search engine in your website


Addicted to Google but want to quit? Looking for an answer and can’t find one? There are alternatives to the most dominant engine on the net. The engines that will rediscover the internet for you

Kyle Monson, PC Magazine Published: 31.10.07, 18:08

Google wants to reach the moon , and maybe one day it will also own the Earth (temporary name). In the meantime she has managed to educate us all to search and find everything in her search engine. According to some market research, Google’s market share of the global search market ranges from 40 to 60 percent (depending on who is asking). The firm grip on the market is certainly justified. The search engine is agile, and is equipped with a highly successful user interface. If you add Yahoo and MSN to Google, you will reach 90 percent of the market.

But have you ever felt like trying another search engine and seeing where it sends you? Google’s search results are based on relevance and popularity, which means you will not always be able to discover new districts (which may be very relevant to you, but are gaining low popularity among the surfers). Here are 14 ways to search and find yourself anew.

Technorati

If you do not include blogs in the daily search menu, you are missing out on a lot of excellent content, which tends to be too fresh for the crawlers and indexes of the typical search engine (who attach too much importance to popularity at the expense of “freshness” lettuce). Among Google’s alternatives in this regard – the best tool we have come across is Technorati .

The blog search service offers a wealth of ways to search for hot content. There is a list of the most popular searches, a list of blogs with the largest amount of inbound links, and a list of music titles, movies and games to which most bloggers create links. The search is easy, and you can sort the results by the date of their publication (to retrieve the latest information) or by the relative importance of the blog (according to an index of the number of inbound links. The more there is, the higher the importance or “authority”).Technorati

ChaCha

Can’t find what you are looking for? The ChaCha Search service allows you to have a live conversation with real professionals, who will guide you regarding the secret wording of search queries, depending on the interest. The search is quite disruptive, which makes its use amusing, and is completely free.

Rollyo

Let’s start with the fact that the name Rollyo is an abbreviation for Roll Your Own Searh Engine. A search engine allows you to conduct general searches, by category, on blogs and online. It offers the user to create a search engine for specific sites for the user to choose. For example, I created an engine that searches only on my favorite music blogs.

You can also see other people’s search engines. There are relatively few cases where you will want to limit the search to individual sites, but if this ability interests you, then Rollyo makes life easier.

Kosmix

This is a thematic search engine . The search is performed by categories such as health, cars, travel, finance, politics and video games. Do not bother to search through it queries on current topics, because it tends to be out of date. This fact makes it possible to find information on issues such as global warming, but makes it difficult to reach, for example, Ahmadinejad’s recent speech. On health issues, for example, the search engine provided links to a wealth of information on disease prevention, treatment methods and risk factors.Kosmix

Ask.com

This rich site uses a ranking mechanism based on popularity, but not in the usual sense of the word. Instead of getting a list of results sorted by the popularity of each link, here the search results rely on popularity among sites that are considered search experts. We did a number of searches, which yielded quite different search results from the results we got on Google – on the same query. We love the preview of the search results provided by the engine, as well as the collection of Smart Answers it offers.

Clusty

An engine with a pretty shocking name, but a fresh approach to how search results are displayed. While Google’s engine displays the results in a simple list, Clusty starts by concentrating search results (aggregation) from several search engines (Google is not among them), and then sorts the results into clusters. At this point the user can refine the search, to narrow down his results. For example, searching for the phrase Dell XPS will return a list of XPS models, reviews, and links to sales sites (grouped into categories).

StumbleUpon

This StumbleUpon search engine allows toolbar downloaders to rank web pages they reach. The more he learns about user preferences, the more accurately he is able to direct them to points of interest.

You can add to the list of friends people who share common hobbies. The preferences of the additional members will help prevent filtering the search results even further (assuming they characterize you as well). The site tends to be slow from time to time, but it may link you to content to which you would not otherwise be exposed.Stumble

Draze MetaSearch

Draze Engine  Center (via aggregation) Search results from Google, Yahoo! And MSN. The home page is reminiscent of the typical Google interface. You can view results in the search of one or more of the search engines, and even subtract from the search results those that appeared in one of the engines.

A useful mechanism allows you to peek into a page that is between the search results (clicking on the word Peek-a-Boo opens a peek window with the scroll option). This way you can save a lot of time and open only really relevant pages.

FindSounds

The FindSounds engine   also deserves to be included in the search engine team as well. Want to find sounds online? This is your address. Here you will find sounds of animals, tools or natural phenomena. The search results are audio files in various formats such as AIFF and WAV.

AfterVote

The AfterVote search site centers search  results from the three major engines – Google, Yahoo! And MSN, and adds a social component, by allowing surfers to vote “for” or “against” the site. The search for it is one of the most sophisticated we have come across. Search results include Alexa and PageRank statistics, Digg and del.icio.us buttons and even Wayback Machines archive pages.

Can’t Find On Google

It’s less a search engine and more a bulletin board . It allows desperate surfers to share with other surfers queries that have raised pottery in the well-known engine, in the hope that someone can help them find what they are looking for. The problem is that there are far more people who post queries to the site that did not deliver the goods, than people who are interested in helping. If you are really desperate, you may want to donate your own orphaned query.Cant Find it On Google

UFOCrawler

You know that the truth is out there, and that Google does not tell you everything. At UFOCrawler you will find what you are really looking for: conspiracies, intrigue, nuclear ambiguity, UFOs and lots of other interesting things. Many search results repeatedly point to the same sites. You probably need to know where to look.

Ask Vox

Looking for meaning? Do you only have a question? We asked Ask Vox what love is (in English, so that he would understand), and he replied in plain English (through computer speakers) that love has many faces, and that there are loves that are worth dying for. Ask Vox is an answer search engine, for those who are looking for an answer and do not find one. This engine is in its infancy, and it asks anyone who gets an answer to stay around a bit and contribute answers to others’ questions. We were happy to find out that Vox at least knows the creator of the world and Bill Gates. That’s something too.

Baidu

You have no reason to enter this search engine (unless you read Chinese), but it is worth knowing about its existence (for example if you are considering entering the Chinese market). It will not replace Google. Is the Chinese Google, in the God of Knowledge. It can be conveniently searched using the Chinese-language pin-yin phonetic spelling (written in Latin letters). If you really want to take a walk in it, make sure you can move your browser to Chinese Simplified encoding. In two words: really interesting – Zhen you yisi.

nternal search engine on the site – without a doubt this is one of the most important tools for improving conversion rates on sites in general and e-commerce sites in particular, however absurdly it is one of the most untreated sites in the vast majority of sites – almost without thinking, measuring or working on performance and improvement, no Too bad? Shame shame!

So take a breather and let’s dive into the most comprehensive guide ever written on an internal search engine on websites, the following guide includes reference to all the layers you need to turn the internal search engine on your site into a well-oiled conversion machine: UX, UI, micro copy, measurement, implementation, work with Data and more.

First of all, let’s start with data that illustrates the potential of the internal search engine on your site:

  • A user who has performed a successful search on the site is 2-5 times more likely (depending on which study they are referring to) to make a purchase in relation to a regular user
  • According to data from The State of site search in 2019 there was a 56% increase in the use of internal search engines on sites compared to 2018
  • 46% of “serious” customers who come to the site with the intention of finding a particular product, begin their consumer journey by searching

In short this important search we have already said?

It is important to remember that an internal search engine is an important component in almost any site, especially on ecommerce sites. But the larger the site and the wider range of content and / or products, the greater the importance of the search engine or in other words – the importance of a search engine on a site that sells only one product in 3 variations is minor while on a fashion site with 500 items or marketplace With 100,000 items it is an essential commodity that sits on a very central artery in the conversion process.

So come on, let’s dive…

Rule No. 1 – Prominent and central location

In the relationship between us and our users, the rule is simple – our job is to make life as simple and comfortable for them and their job is to be as loyal as possible and convert as much as possible. And when we talk about convenience and simplicity in the context of a search engine, we are talking first and foremost about a prominent and clear position. So this is our first rule – note that your search engine is in a prominent and central position in the vast majority of site pages (except for pages like personal area, check out, such operational pages and others where it is sometimes incorrect to integrate the search engine). Prominent location means – any user who comes to the site can in a split second and without reading anything understand where to search.
For your convenience, here are some examples of how to make it stand out (and how less):

Note the super prominent location on Amazon’s website (both mobile and desktop) – there is no doubt that you can not miss (Amazon is a great example of another variety of effective search applications that we will address later in the article)

Amazon search engine

Also on the IKEA UK website you can see the search bar located in a very conspicuous and unavoidable way (again they too, like any international brand that understands where the world is going, did not miss the mobile)

The search engine at IKEA
The search engine at IKEA

And if we talked about amazon.com it is impossible without its sworn nemesis Walmart who also did not miss (did you notice the similarity? Suppose it is coincidental…)

The search engine at walmart
The search engine at walmart

And of course there are also less good applications, for example – on the (excellent in general) site of Terminal X you can see that they missed when it comes to the search bar and instead of highlighting the bar and making users enjoy the richness on the site, choose to keep the search bar both mobile and desktop

The search engine in Terminal X.
The search engine in Terminal X.

Century21 also missed the opportunity

The search engine in the C21
The search engine in the C21

The same goes for Adobe who really bothered to hide the search engine (they have great software and lots of content and information to offer, isn’t it a shame ?!)

Search engine in Adobe
Search engine in Adobe

So what do we take from here? It is very important to place the search engine in a very prominent place on the site, such that your users will not have to search even for a moment. Be careful:

  • Place the search engine at the top of the screen (in the area of ​​the site’s logo and maple)
  • Place it outside a sidebar (do not accidentally poke a hamburger in a mobile)
  • If there is no place to spread the search bar across the entire width of the screen, insert the search pane open on 1 / 3-1 / 2 of the width, in Hebrew sites it is recommended to put it on the left side (because our eye begins the process of scanning the site from left to right in reverse F) And on English sites it is recommended to put it on the right side

Attached is a diagram representing how different areas of the screen receive “initial attention” from users, as you can see the upper area receives the highest initial attention (it makes sense that users will usually start the page from the top), with the top-left corner receiving the highest attention Most sites in Hebrew and top-right sites in English

Recommended location Search bar
Recommended location Search bar

Rule No. 2 – Open and accessible search bar

Have you ever wondered why on the Amazon site the main thing you see when you arrive is a wide and huge search bar? (With a wide field and a yellow button?) Why does Google have an open and not closed bar? The answer is simple – one click less! Search Invitation = Higher conversions, so who are we to argue? An open search bar is without a doubt:

  1. More inviting 
  2. More efficient (saves at least one click) 
  3. Allows you to convey messages on the search bar (see below for a separate reference to the topic)

Here are some excellent examples of particularly inviting search lines:

So while it is not a search engine within a site but a site that is a search engine, but still – it is impossible to talk about search without taking examples and learning from those who more or less invented the search experience – Google. Pay attention to the prominence and manner of ordering (and other elements we will talk about later in the guide) of the search bar when reaching Google on both mobile and desktop

Google search engine
Google search engine

Even at houzz (for those who do not know – a particularly successful start-up of an Israeli couple that deals with the worlds of home design, inspiration, marketplace, connection to suppliers and more) did not miss and make sure to place the search bar in a very prominent, open and inviting way

The search engine in Houzz
The search engine in Houzz

And how is it possible without the Chinese e-commerce giant (and the site that sells the highest turnover in the world!) Ali Express

Search Engine Owners Express
Search Engine Owners Express

And here are some sites that missed the opportunity:

Zara

Bazaar Search Engine
Bazaar Search Engine

Global Sunglasses Giant (Oakley)

The search engine in okley
The search engine in okley

Rule # 3 – Use the search bar as a tool to create demand and interest in the search process

Remember that users who search are more likely to convert? Excellent! So it is quite clear that you have a clear interest (assuming your search is good and assuming you have applied as many insights from this article as possible) that your users will perform a search.

A great way to get them to perform the same long-awaited search (in addition to rules 1-2 I mentioned – a prominent location and an open search field) is an inviting and relevant micro-copy in the search bar. Instead of listing – search or search, SEARCH or GO It is possible (and recommended!) To list more inviting things that highlight the benefits of searching your site such as: highlighting the variety / wide selection, a light explanation of the ways you can search or just motivation for action. Here are some sample texts:

Start your purchase right here

Over 1,000 items are just waiting for you

Search among more than 10,000 products 

Search by item number, product name or brand 

Come find your next vacation

Just choose from over 1,000 designer clothes

As mentioned – the main goal is to invite the user to perform a search while highlighting and highlighting features or capabilities in the search engine and / or your site in general, things that are worth considering incorporating in the micro-copy of the search bar:

Large selection – If you have a large selection of articles / recipes / products, etc. and this is your relative advantage, it is worth highlighting it in the search bar as well.

Different options to perform a search – if you can search in different and convenient ways, for example – search by name, by part number, by manufacturer, etc.

Motivation for emotional action adapted to the concept of the site – for example if your site is a vacation site, then you can write something like: Look for your next vacation here…

A good example can be found on Spotify

The search engine on Spotify
The search engine on Spotify

And the international fashion giant ASOS

Asus search engine
Asus search engine

Rule # 4 – Always use a magnifying glass

As mentioned, our users should not think too much and certainly not get tired so it is important for them to ensure as consistent a user experience as possible. It does not make sense that on a particular page you put the search bar in a map and another page (because it works) in a potter or another location, it produces unnecessary cognitive noise that will surely lead to less good results in the user experience and your ability to lead the user to the desired result (conversion). As mentioned above, the ideal location is up in a prominent place (in applications where there is a lower “sticky” bar, there are schools that say that it is correct to access the engine from the bottom left due to the convenience of access with the finger, but we did not see significant data and provide support. Ours is to place up and in a prominent place)

Consistency in place

Rule # 8 – Make the most of the search dropdown

Most of their searches are performed by your users on Google, Amazon, Spotify, Netflix and the like and this is for them the benchmark for comparison, yes – yes, they compare you to Google, Amazon and other global tech monsters and therefore they expect you to understand what they want without having to Say it explicitly. Challenging, no ?!

A great way to do this is to use AUTOCOMPLETE, before we dive into how to do it right, it’s right to hone two concepts: 

AUTOCOMPLETE – is a technique in which a search engine automatically completes the search of the surfer based on the string he has already typed, for example if the user typed – aro, then the engine will offer him to complete the string to ‘closet’, ‘closet’, ‘chimney ‘ Etc. etc’. In essence an auto-completion mechanism is one that is based on an alphabetical order, meaning if I searched for the string ‘arrow’ it will offer completions with ‘a’ after the string (if any) and then ‘b’ and so on. 

AUTOSUGGESTION – This is a smarter mechanism designed to deduce which completions and / or suggestions may interest you as possible search results based on various variables such as – the string you typed so far in the search bar, common searches, past visits and searches, site trends and more.

As much as possible we strive to use AUTOSUGGESTION as much as possible and “simple” AUTOCOMPLETE as little as possible, as the relevance of the completion suggestions we give our users while they search is crucial, with the best way to do it being a combination of the two. 

So do you make proper and smart use of auto-completion mechanism and suggestions? Let’s go over some points on the subject:

  1. Be sure to calibrate the AUTOCOMPLETE mechanism in a smart and non-technical way, for example – if a person types ‘H’ we want him to first get a complement to the word ‘shirt’ and only then a complement to the word ‘corset’
  2. Combine as much data and data as possible in order to “guess” what the user wants or alternatively interest him in relevant things, by flooding with – hot searches and recent searches (assuming it is a repeat user) prioritization of autocompletions by frequency and not by A-B in a simplistic way. Most search engines will prefer non-domestic and without too much thought (which is of course a mistake). Let’s take an example from Google (feel free to try it for yourself on an endless variety of search strings) – if you start typing the ‘all’ string, the logic is that you will first get options that start with the string you typed ‘all’ followed by the letter ‘a’, meaning ‘prison’ and so on. Then ‘dog’, and ‘calg’ and so on, but in fact this is not the case. The first result you will get is – ‘general’ (‘to’ after the string ‘all’) and then ‘calcalist’ (‘as’ after the string ‘all’) and then ‘rule insurance’ and only in the ninth result will you get the option ‘dog’ *, Which was supposed to be (if you look in terms A – B) higher than all those who came before it .. Why? Because Google understands that the alphabetical order of the results is irrelevant and tailoring to the need (user, trends, common searches, etc.) is much more important. So we will probably not be able to produce a “Google” search experience on the site, but striving and aiming for it is certainly possible (and desirable) so it is very important that your autocompletions be as smart as possible and based not only on regular alphabetical order but also on search frequency. / T. For example – if it is a user who usually buys kitchen products on the site and he started typing on the site (let’s say for example that the site deals with household products) the string ‘Maz’ then it would probably be more correct to bounce the results in the first completions – ‘fork’ or ‘festive forks’ And not necessarily the phrase ‘double mattress’. * It is important to note that Google search is constantly changing so it is possible that if you search for the string ‘all’ while reading the guide you will get a different position for the ‘dog’ suggestion (and not ninth place as happened in the example we mentioned),Autocomplete
  3. Number of search results
  4. If your search has different types of relevant objects for example: product vs. content article or product name vs. brand or category name, it is recommended to separate them very clearly graphically and divide the space between them in a way that illustrates the frequency of the search. For example – if a particular user searches the NI string on a shoe site, this string can of course be relevant to any pair of NIKE shoes but also to a NIKE brand page or an article on the NIKE site so it is very important for the user to understand before choosing the auto-complete option – Does he choose a particular model (then clicking on the completion should take him to the model page) or does he choose a brand page (then choosing completion will take him to a general page / brand page) or he chooses a page that will take him to a content article and so on. Beyond the fact that the user is supposed to understand where he is going, it is very important that graphically the relationship will also be relevant, if NIKE has 100 products on the same site, 6 brand pages (for all kinds of campaigns, etc.) and 15 content articles, it is unlikely that the proposals for completion will be presented in the same relative and prominent, and it is clear that in such a situation (assuming it is an ecommerce site of course) the product pages themselves should have much more prominent How much more prominent? It very much depends on how people behave on the same site – if it is a site for selling shoes with extensive activity in the content world and the data shows that customers who interact with the site’s content then tend to buy more, then maybe the articles should gain significant weight and if not, then maybe just a mention. at all. The options are varied, but what is important to remember is that this dose is important to perform on a data basis and to improve from time to time. How much more prominent? It very much depends on how people behave on the same site – if it is a site for selling shoes with extensive activity in the content world and the data shows that customers who interact with the site’s content then tend to buy more, then maybe the articles should gain significant weight and if not, then maybe just a mention. at all. The options are varied, but what is important to remember is that this dose is important to perform on a data basis and to improve from time to time. How much more prominent? It very much depends on how people behave on the same site – if it is a site for selling shoes with extensive activity in the content world and the data shows that customers who interact with the site’s content then tend to buy more, then maybe the articles should gain significant weight and if not, then maybe just a mention. at all. The options are varied, but what is important to remember is that this dose is important to perform on a data basis and to improve from time to time.
    Result type
  5. Apart from offers for automatic completion (more or less smart), remember that the search completion mechanism can also be an additional advertising channel for promotions, value offers, etc. (Of course, try to link as much as possible the promotion / value offer to the worlds of content the user is looking for. Strips of promotions as part of offers for auto-completion or alternatively incorporate banners (delicate) as part of the search engine dropdown
    Combining promotions
  6. Avoid the inconvenience of proposing autocomplete that will lead to the “No results” page, if the system offers autocomplete you must make sure that there are optional search results behind this complement

*** Please note, when you use an auto-complete mechanism you may well create a problem for yourself when it comes to measurement, since ostensibly if a user came to your site and typed ‘Hu’ and you gave him a specific completion then he selected it and moved to the product page, ostensibly in terms of systems The measurement was not performed here at all, although for you it is highly desirable to measure and analyze this data, later we will review those who avoid this distortion

**** In any case, whether you choose an engine that is based only on completeness according to A-B or an engine that also incorporates smarter completions (personal, data-based, etc.), it is important that the logic of A-B be logical and relevant and not Technical. It does not make sense that if a user searched for the string ‘candle’ he would get first completion ‘candle status’ and second completion ‘candle’ or ‘candles’, there should be clear logic that prefers full words first and then prefixes and so on (again, unless the ‘candles’ completion ‘Progressed in line due to intervention of “mind” to recommendations). The basic logic of a-b should be run according to the following legality:

Search Results
Search T
  • Gentle apology / identification with the situation
  • Suggestion for spelling errors
  • Offer to contact customer service and accessibility of an easy way to contact (in fast ways not by email or fax but chat / WhatsApp or phone)
  • Suggestion for other products that may be of interest, any site and any field with its inaccuracies but by and large it should be – hot products that might interest you or hot categories
  • Possibility to perform another search as simply and conveniently as possible

So how is it possible without examples? Let’s look at some sites that do this great, and the first example among them is the GAP US site on both mobile and desktop: 

Maximum
Maximum

Norstrom also does not miss:

Improvement suggestions for finding results
Improvement suggestions for finding results

And also in Pot Locker:

Improvement suggestions for finding results
Improvement suggestions for finding results
Maximum

Notice how in this example Amazon automatically corrects the ‘kitchentable’ query to ‘kitchen table’ in the most natural way possible:

And in this example Urban Outfitters do exactly the same thing:

Maximum
Maximum

Here are some examples of sites that are missing out on this important point, such as Shufersal:

No results

AM: PM:

No results
No results

Rule # 10 – If your search engine made cuts and filters automatically, reflect and make them accessible

If in the search all kinds of sub-cuts were made in order to produce accurate results, reflect this to the surfer and make them accessible in a simple way so that he can understand the process and more importantly move within it more smoothly. For example – suppose a certain user searched for the query ‘red running shoe for men’ and then he comes to a page with only 4 results, but above the results there are a kind of “tags” of ‘man’, ‘running’, ‘red’ (these tags represent the sub-cuts That the engine performed to get results), then with a simple click the user can remove the ‘red’ and get results for running shoes in a variety of colors (instead of re-searching). Did you save him time? Did you create a good and requested experience? You will receive your salary by increasing the conversion rate. And how is it possible without examples:

An excellent example from the Foot Locker USA website:

Mirrored sub-cuts

And also from HOUZZ:

Mirrored sub-cuts
Mirrored sub-cuts

Rule No. 11 – Allow easy orientation on the results page

A search results page with too many results can (and probably will) lead to bad results so it is very important to allow as many filters and filters as possible, also – if there are common filters or recurring patterns, it is important to use them and shorten the way. For example – let’s say you have decided to allow 3 Sorting Options:

From the new to the old or vice versa

From A to Z and vice versa

By best sellers

And suppose you set the basic option to be from A to T, does that mean you necessarily chose the right option? Really not sure! It is very worthwhile to work with the data in order to see if a pattern is identified and if so – to adopt it. For example – if you see that a large part of your surfers change the way they display and choose to display the products from new to old, then it may be worth considering making this option the default (of course you should measure and see if the change improves your KPIs regarding search, but it is without Potential provider). The same goes for other elements such as – the amount of products on the page, in cases of many results most sites will display browsing (so-called pagination) between pages, and it can be the default that you set is 20 results per page but most users replace it and choose 50 results per page. Regarding the quantity of items in a row, etc., etc. Pinterest is an example of a site that provides an excellent experience in navigating between the various search results:

The search engine on Pinterest
The search engine on Pinterest

Rule No. 12 – Measurement, measurement and measurement again

Measurement is without a doubt the most important thing you can do to turn your on-site internal search engine into a well-oiled conversion machine! The measurement helps in constantly improving the user experience, in learning the engine to better answer search queries, understand what interests your users and more and more.

So how do you do it right? What is important to measure? Let’s dive…

The most common measurement tool in the world when it comes to measuring websites is – Google Analytics (or GA for short) and GA has quite a few Build in capabilities when it comes to measuring and tracking searches, but is that enough? Definately not! This is at best a good start, not beyond that. But if this is a start, then let’s start with that:

How do you connect your GA account to a search so that every search query on the site will be recorded in your GA account?

Enter your Google Analytics account settings to view level

Select View Setting and then Site Search Settings

Google Analytics min Admin ➤ View ➤ View Setting ➤ Site Search Settings

Scroll down to the Site Search Settings section that looks like this and turn on Site Search Tracking

site search settings

Now you need to define for GA what is the parameter in the URL structure of your site that represents a search query. How do you find this parameter? Just go to the site and perform a search, inside the URL of the results page (after you type a query and perform a search) a parameter will appear indicating that it is a search results page like some letter (usually q or s) or a word (usually search or query ). For example, if we search for the word ‘fruit’ on a site, the search results page URL might look like this:

https://www.mydonain.co.il/? q = fruit

* The parameter we are looking for is the same parameter q from the URL – in general all the parameters we want to pass to GA come after the question mark (for example UTM SOURCE, UTM MEDIUM etc.), but in this case because you searched the site the URL should be clean From other parameters so you can take the parameter that appears after the question mark and set it as a parameter that represents a search results page. Note that you check the option – Strip Query Parameters Out Of URL, so that GA will clear all the parameters from the URL and present you in the search report (which we will discuss in detail immediately) the queries themselves in a “clean” way

query parameter
Amazon search engine

On sites where it is possible to search within certain categories, it is advisable to activate the Site Search Categories option in GA, which allows you to analyze the data while cutting between categories and the like. For example in amazon you can select (by clicking on the dropdown that says ALL) a specific category on the site and search only within it

In such a case the fact that the search was performed within a particular category and not within all the options on the site should somehow be reflected on the results page (by indicating a parameter in the URL), for example:

https://www.mydonain.co.il/? q = watermelon & sc = fruits and vegetables

In the case before us the parameter ‘q’ represents the search query and the parameter ‘sc’ represents the category in which the search was performed, in this case if we want to monitor the categories we will type the parameter – sc in the Category Parameter

Again, if we want to get the “clean” results it is important to check Strip Category Parameters Out Of URL

Category parameter

If your search is not structured in the way I mentioned and how the URL on the results page contains the search query, the category (if you have a search within specific categories) and a dedicated parameter that allows it to be diagnosed, for example in this case:

https://www.mydonain.co.il/s/ Fruits

or

https://www.mydonain.co.il?s_ Fruits

Or any other structure that is not: question mark >>> any parameter >>> equal sign >>> search query

You have two options to get the information (this so vital) into your GA:

  1. Make sure your development team makes the required adjustment to the results page URL
  2. Create a bypass using a dedicated JS in Google Tag Manager

How do you do that? Before we dive into the technique of actual implementation in GTM I want to explain the essence of the move, what this JS does “in a big way” is to extract the search query from the URL and return it to Google Analytics plus a question mark and equal, so even though the URL Appearing in the browser (and we as surfers see) is without question and equal, the URL transmitted to GA as part of the report is yes in the correct structure so GA will know how to use SiteSearch capabilities. Enter GTM and create a Variable type – Custom JavaScript, in which we enter the following JS:

function () {

var pagePath = window.location.pathname;
var searchParam = ‘/ search /’;
if (pagePath.indexOf (searchParam)> -1) {
return searchParam + ‘? s =’ + pagePath.split (searchParam) [1] .split (‘/’) [0]
}

Let’s unpack the JS for a moment:

var pagePath = window.location.pathname

Here we generate a variable called pagePath whose value we assign is window.location.pathname

var searchParam = ‘/ search /’

Here we create a variable called searchParam and assign it the value of what we have in the URL on the site (instead of the desired state), in this case instead

/? s =

There it is

/ search /

if (pagePath.indexOf (searchParam)> -1)

Here we condition the condition of the function that says in simple language that if the variable – pagePath (which is actually our URL) contains the value of the variable searchParam then the following thing should happen

return searchParam + ‘? s =’ + pagePath.split (searchParam) [1] .split (‘/’) [0]

Which is simply – splitting the URL and replacing that part

/ search /

In this part

/? s =

Want to make sure you understand?

What will the function look like if our search site looks like this:

https://www.mydonain.co.il/s/ Fruits

And we want q to be our search parameter?

Take a second to think…

True – the function will look like this:

function () {

var pagePath = window.location.pathname;
var searchParam = ‘ / s / ‘;
if (pagePath.indexOf (searchParam)> -1) {
return searchParam + ‘ ? q = ‘ + pagePath.split (searchParam) [1] .split (‘/’) [0]
}

So what will happen at GTM?

We will create a new JS variable

We will then make a change to the GA base script so that the page to be passed to GA as part of the Page View will be the page with the manipulation we performed (the replacement) and not the page as it appears in the URL, and then because GA henceforth sees a correct page with proper structure Of results page – question mark, parameter, equal mark henceforth Site Search will work properly. Did we understand? Excellent – so let’s dive into the technical…

Log in to the Tag Manager account and go to the Variables tab

Create a Pride Variable Script in Google Tag Manager

Adding a variable defined by the user (User-Defined Variables)

User-Defined Variables

Creating a unique JS variable (Custom JavaScript) for the example we called the variable – Search-mihi.market

Create Custom JavaScript

Add our JS code (for your convenience you can copy it from here)

function () {

var pagePath = window.location.pathname;
var searchParam = ‘/ search /’; // replace this with your page path before the search term
if (pagePath.indexOf (searchParam)> -1) {
return searchParam + ‘? s =’ + pagePath.split (searchParam) [1] .split (‘/’) [0]}
}

It will look like this:

 Google Tag Manager

We will now modify and adjust the general GA tag (Page View tag) by adding Fields to Set, with the page name field and the dynamic value of the variable we created (remember our example is – Search-mihi.market)

Set a tag for the variable in Google Tag Manager

You can now enter the Site Search settings and set it as we specified at the beginning of Rule # 12.

Rule No. 13 – Performing analysis and ongoing analysis

Okay, so after we made sure the search data got to GA properly, it’s time to see what it looks like and what can be deduced from it.

First and foremost – where are we? Inside the GA account, under Behavior category >>> Site Search, where GA’s ready reports for us to search the site await us. Please note that these reports will be filled in automatically from the moment you properly define the search (the information will be collected from now on and not backwards)

Site Search

What fun, we have come to the part that relates to Site Search Let’s review all the good it offers us: 

The first tab – Site Search Overview:

As the name implies, here we can see general data about the searches on the site (you can of course use all the “usual” tools of GA such as cutting dates, using segments, etc.): For your convenience we have listed the various parts in Overview, let’s go part-by-part

Internal Search Terms - Google Analytics

Part 1 – Sessions with Search: The reference is to the number of activities (sessions) on the site during which a search was performed. Note that the figure only refers to searches that were completed as part of a search – meaning those in which the user clicked on a search and reached a results page, even if it was empty.

If a user touched the search engine and did not perform a search or started typing and stopped / regretted his search will not count here

If a user typed a string in the search bar and then selected one of the autocompletions (one that passes to a specific page and not a results page) then he will also not count here as a search session

Part 2 – Total Unique Searches: This refers to the number of different searches performed on the site less duplicates. For example if a user performed two different searches on the same visit (once ‘shoes’ and a second time ‘basketball’) then they will count as two searches, but alternatively if he performs two identical searches (‘shoes’ and then again ‘shoes’) on the same visit they are Will count as one. If we divide part 2 by part 1, we get the average of the various searches that users search for a visit (in our example 3,021 / 2,016 = 1.498), that is – the users of this site perform about 1.5 different searches per visit on average

Part 3 – Number of results pages viewed in relation to Searches (Results Pageviews / Searches): The average number of times the search results page was displayed, from these three data together (1, 2, 3) you can extract the number of duplicate searches that were in the same session (e.g. Searched for ‘shoes’ on that visit), how? So:

2,016 * 1.67 = Number of result pages = 3,367

The amount of unique searches is as stated = 3,021

This means that the number of duplicate searches performed in the same session is – 3,367-3,021 = 346

This result (346) is divided by a total of pages of results (3,367) and we get – 10.28% For convenience we call this figure part 3.1 (later you will understand why) – part 3.1 will be defined as the percentage of double searches (Double Searches)

This figure is particularly interesting for troubleshooting and / or debugging a search engine that produces duplicate searches or “gets stuck” on certain searches or loads the results page very slowly and causes users to type again or click on the search and the like, as this is not a likely scenario. A visit will perform the same search (this is likely in cases where a user leaves a session and returns after a few minutes and is told about the same session, but this should be in a very marginal percentage). Let’s say that if the amount of searches returned in your search engine (which as mentioned we saw how to extract it) reaches the order of 10-15% of the total unique searches on your site, then you may very well have a problem with the search engine and worth checking

Part 4 – Search Exit%: Here we will see the percentage of visits that ended right on the results page, that is – a user entered the site, performed a search, came to a results page (note that the “No Results” page is also set as a results page) and this The last page where he visited the same session. A high rate of search after ports usually indicates an inefficient search engine that provides users with irrelevant results (in quantity, level of accuracy, etc.), your ambition should be to reach a port of search after 10% and even less – it is not a one – two day process, But within a period of 8-10 weeks of thorough work on the search engine (learning new queries on a regular basis, working on “no results” pages, strengthening the auto-completion mechanisms, “training” for common errors and more) it is definitely possible

Part 5 – Search Refinements%: The percentage of times a user was forced to perform another search immediately after performing a search, for example – if a user searched your site for the query – ‘chicken soup’ and immediately afterwards searched for the query ‘chopped liver’, He will then be counted as a user who performed a repeat search. Please note, if the user performed the additional search on the same session but after going through additional screens on the site, this search will not be considered as a repeat search. for example:

Search query ‘Slippers’ >>> Reach the results page >>> Search the query ‘Ballet shoes’ >>> Reach the results page

In this case this user will be counted as a repeat searcher

Search query ‘Slippers’ >>> Reach the results page >>> Enter one of the products displayed on the results page >>> Search the query’ Ballet shoes >>> Reach the results page

In this case the user will not be counted as a repeat search user

A high percentage of repeat searches will indicate a poor search engine, as the user does not get what he is looking for and is forced to perform another search.

By and large our ambition is for as few repeat searches and as few post-search exits as possible

Part 6 – Time after Search: Here we show the length of time the user stayed on the site after performing a search (the period of time is calculated from the moment the search query is performed until the last page of the visit *). Usually this figure will be in high correlation to the percentage of departures after a search, a high percentage of departures will usually lead to a low stay time and vice versa. In general, a long stay after a search is a positive sign (although too long a stay can indicate problems such as confusion of users and their inclusion in inefficient flows), but it is very subjective to your site.

* It is important to remember the (somewhat strange) way that GA calculates temporary stays in general. Because GA feeds commands that are shot from the site to your GA account (called hits) and through them it builds sessions (it actually sees that the hits that come from your visit are with the same user ID and create from them a “sequence” called Session), panels, events. Etc., he has no real way of knowing how long a user has been on the site (there are detours that can be done to know, but it is not for this guide) but by subtracting the date of the hit from the first hit, if for example a user came to the site Yours at 22:53:12, so the first hit that GA received is at 22:53:12 and then the user visited the site and entered a number of pages, with the last one being at 22:56:44 and did not perform any action there. Another that produces another hit, so GA will calculate the dwell time (duration of the session in this case) as the time elapsed between 22:56:44 and 22:53:12 (h – 3:32 min). The same is true in the case of stay time after the search, It will be calculated as the difference between the time the search page was displayed (the time stamp of the same hit) and the time of the last hit that the user created (it could be that after the last hit he stayed another time on the same page, GA will not know if you do not tell him ). In short – the issue of length of stay in GA is not the most accurate and requires detours for really accurate results, but as mentioned – this is an issue for another guide, it was just important for us to be aware of the relative inaccuracy in the data

Part 7 – Average Depth of Search (Avg.Search Depth): Here we will see the average number of pages the user viewed after performing the search (including the results pages themselves, so that Avg.Search Depth is always greater than 1)

This is the place to show you the index we built here at mihi.market, we call it – Search Score and it is built from the following formula:

Part 4 * Part 5 * Part 3.1 / (Part 6 * Part 7) = Search Score

Or more simply:

Search Score

If we take the numerical example attached above the resulting Search Score will be:

Counter – 3.35 * 1.84 = 6.164

Denominator – 10.28% * 30.55% * 23.28% = 0.00731

The total Search Score in this example is 843

In general, the higher the Search Score, the better the search quality on your site, this is a very subjective score for your site and is mainly designed to make sure that any change you make does not adversely affect another site, so this index should be in front of your eyes from the moment you start working on The search engine and one goal – to increase the score.

Let’s take another numerical example, a site whose data is its data:

Time after Search – 3.5 דקות
Avg.Search Depth – 2.5 מסכים בממוצע
Search Refinements – 25%
Search Exit – 25%
Double Searches – 10.5%

ה-Search Score של האתר הוא:

3.5 * 2.5 = 8.75 במונה

25% * 25% * 10.5% = 0.00656

הציון במקרה הזה יהיה 1,333. האם זה אומר שבהכרח שהאתר הזה טוב יותר מהאתר הקודם? סביר להניח אבל לא בטוח, שכן יכול להיות שהאתר הראשון הוא אתר תוכן שמעוניין להביא משתמשים לזמן שהייה מאוד גבוה וכמות עמודים גבוהה בביקור בעוד האתר השני הוא אתר מידע מאוד פונקציונלי שהמטרה שלו היא שמשתמש ימצא את המידע וילך לדרכו, לכן קשה לשפוט מבלי להשוות “תפוזים לתפוזים”, מאידך אם מדובר באותו אתר בשני מועדים שונים, אפשר להגיד בבטחה שהמצב החדש הוא טוב יותר מהמצב הקודם, משמע – התוצאות של מנוע החיפוש השתפרו

בנוסף ניתן לראות במסגרת ה-Overview של ה-Site Search גם את הביטויים הנפוצים, עמודי הפתיחה (העמודים בהם מתבצעים הכי הרבה חיפושים) הנפוצים וקטגוריות נפוצות (במידה ומדובר באתר עם חיפוש פנימי בתוך קטגוריות, במידה ולא – אז הנתונים כאן יהיה לא מוגדרים). אוקי אז סיימנו עם לשונית ה-Overview, עכשיו נעבור ללשונית הבאה – Usage

הטאב השני – Usage:

כאן נוכל ללמוד בהרחבה על ביצועי הסשנים בהם בוצעו חיפושים באתר (כמות, שיעורי נטישה, משתמשים חדשים, דפים נצפים לסשן, המרות וכו’). גוגל אנליטיקס מציג לנו כברירת מחדל את ההשוואה בין סשנים ללא חיפוש לבין אלו עם חיפוש וביצועיהם.
נחלק את הדוח הזה ל-3 חלקים – Acquisition, Behavior ו-Conversions:

Acquisition – כאן GA מראה לנו מהי ההתפלגות בין סשנים שנעשה בהם חיפוש לבין אלו שלא והאם משתמשים חדשים עושים יותר שימוש בחיפושים לעומת אלו שכבר “מכירים” את האתר. התובנות העיקריות שניתן להוציא מנתון שכזה קשורות בחשיפה למנוע החיפוש באתר ורמת השימושיות בו. לדוגמה – אם רואים אחוז נמוך יחסית של ביקורים עם חיפוש, נכון יהיה לתהות האם שווה בכלל להשקיע במנוע החיפוש? ואם רוצים להשקיע, אז אולי שווה למקם אותו במקום אחר? בצורה אחרת שתבלוט יותר? וכדומה 

Behavior – מאפשר לנו לייצר השוואה קלה בין ביצועי סשנים שנעשה בהם חיפוש לבין אלו שלא נעשה בהם. אם המדדים של השורה השנייה (Visits With Site Search) טובים מאלו של השורה הראשונה, הרי שמנוע החיפוש באתר כנראה עושה עבודה טובה על המדדים המרכזיים של האתר (לא בהכרח ש-Bounce Rate נמוך או כמות ביקורים בסשיין או זמן ממוצע חייבים להיות מדדי ההצלחה בכל אתר, אבל נניח שבגדול מדובר במדדים חיוביים) – מצמצם את שיעור הנטישה, מביא את המבקרים להתעמק ביותר דפים ולמשך זמן ארוך יותר, והכי חשוב הוא ממיר יותר טוב. אם המצב הוא הפוך, תצטרכו לעשות עבודה מקיפה כדי להבין מה עליכם לשפר על מנת להשיג תוצאות טובות לעומת סשנים בהם לא מתבצע חיפוש באתר.

Conversions – המדד הכי חשוב (בהנחה שמנוע החיפוש של האתר פועל בצורה סבירה לכל הפחות) שאומר שעם חיפושים באתר משיגים יותר יעדים, הרי שיש לתת למנוע החיפוש מקום משמעותי יותר באתר ולהניע את המשתמשים לעשות בו יותר שימוש

Analytics data

הטאב השלישי – Search Terms:

דוח זה יורד לעומק של שאילתות החיפוש עצמן ומספק לנו נתונים קונקרטיים עבור כל אחד משאילתות החיפוש שחופשו באתר.

מדו”ח זה נוכל להבין מהי איכות המענה שהצלחנו לתת למשתמש עבור כל אחד מהחיפושים – האם הוא נאלץ לצאת מן האתר או לבצע חיפושים חדשים? עד כמה העמיק באתר לאחר ביצוע החיפוש – מה שעשוי להעיד על חיפוש שהניב עבורו תוצאות טובות וכך הלאה. כהרגלנו ולנוחיותכם, פירקנו את הדו”ח לחלקים, בואו נעמיק בכל אחד בנפרד

Internal Search Terms - Google Analytics
  1. מונח חיפוש (Search Term) – כאן נראה את אותה שאילתת חיפוש (מונח חיפוש) שאליו מתייחים הנתונים, כל שורה מייצגת מונח חיפוש אחר. חשוב לזכור שהרשימה הזו לא כוללת מונחים שהמשתמש הקליד ובעקבות השלמה אוטומטית עבר ישירות לעמוד מוצר / תוצאה, כמו כן לא יוצגו כאן המחרוזות שהמשתמש הקליד לפני שבחר בהשלמה אוטומטית. לדוגמה – אם משתמש הקליד את המחרוזת ‘חו’ ואז בוצעה השלמה לחולצות והוא בחר אותה, במקרה כזה הוא יעבור לעמוד תוצאות חיפוש של חולצות ולכן המונח חולצות יופיע כאן ויספר, אך המחרוזת הראשונית של המשתמש ‘חו’ לא תופיע (זה נתון חשוב אם רוצים לשפר את מנגנון ההשלמה האוטומטית)  דוגמה נוספת – אם משתמש הקליד את המילה ‘חו ובחר בהשלמה אוטומטית של ‘חולצה טי פרחונית’ ועבר לעמוד של אותה החולצה, במקרה שכזה מבחינת GA בהגדרות הבסיס שלו, לא התבצע כלל חיפוש ולכן הנתון לא יופיע באף אחד מדו”חות ה-Site Search כולל לא בדו”ח ה-Search Terms
  2. סך החיפושים הייחודיים  (Total Unique Searches) – כמות החיפושים היחודיים (שבוצעו במסגרת סשיינים שונים) שבוצעו על אותו מונח חיפוש במסגרת הזמן שפילטרנו. שימו לב – אם משתמש חיפש מונח מסוים פעמיים באולו סשיין הוא יספר כאן כחיפוש אחד, אך אם אותו משתמש יחפש את אותו המונח בשני סשיינים שונים, זה יספר כאן כשני חיפושים שונים
  3. תוצאות הצגת דף / חיפוש (Results Pageviews / Search) – עמודי תוצאות שנצפו ביחס לחיפושים, נתון זה מתייחס לכמות עמודי התוצאה השונים שמשתמש ראה לאחר ביצוע חיפוש, כאשר ברור לחלוטין שאם משתמש נאלץ להיכנס למספר תוצאות שונות על מנת לקבל את מבוקשו, אזי – החיפוש לא עובד בצורה אידיאלית. המצב האידיאלי הוא שמשתמש יבצע חיפוש, יקבל בדיוק מה שהוא צריך ויתקדם משם להמשך ה-Funnel (השארת ליד / רכישה / פתיחת מנוי וכו’), במידה וזה לא המצב והמשתמש נאלץ לחפש את מבוקשו בין תוצאות החיפוש השונות, אזי החיפוש לא מספיק “חכם” או “ממוקד”, עם זאת – הנ”ל מאוד סובייקטיבי ותלוי במונח החיפוש עצמו, בסוג האתר וכו’ (כנראה שאם משתמש נכנס לאמזון ומחפש ‘מחשב נייד’ לא סביר שהוא יסתפק בצפייה בעמוד מוצר אחד, אלא יעבור מגוון רחב של עמודי מוצרים מהתוצאות עד למציאת המחשב הנכון)
  4. אחוז יציאות לאחר חיפוש (Search Exits%) – מספר היציאות מהאתר מיד לאחר ביצוע חיפוש ביחס לכמות החיפושים הייחודיים שנעשו בו עבור מונח חיפוש ספציפי. מדובר באותו מדד כמו שסקרנו ב-Overview רק שהפעם הוא מתייחס למונח החיפוש הספציפי הזה ולא כללי לכל פעילות החיפוש באתר
  5. מיקודים של חיפושים (Search Refinements%) – כנ”ל, אותו מדד כמו ב-Overview רק הפעם בפריזמה של מונחי חיפוש ספציפיים
  6. זמן לאחר החיפוש (Time after Search) – כנ”ל
  7. אחוז העומק הממוצע של חיפוש (Avg. Search Depth) – כנ”ל
  8. קטגוריות חיפוש באתר (Site Search Category) – כפי שפירטנו לעיל, באתרים גדולים עם מגוון רחב של מוצרים ניתן (ולעיתים רצוי) לייצר חיפוש פנימי בין קטגוריות, במקרה כזה כאן ניתן יהיה לראות את כל שבעת המדדים הללו כאשר ההתייחסות היא לקטגוריה, לדוגמה נניח שיש לכם באתר 5 קטגוריות שאפשר לחפש בתוכן, אז תוכלו לראות את אחוז המיקודים, אחוז היציאות, סך החיפושים היחודי וכו’ – עבור כל קטגוריה בנפרד

הוספת מאפיין משני (Secondary Dimension) – כמו כמעט כל הדו”חות ב-Google Analytics גם הסקשיין של Site Search לא מספק מענה מלא על כל השאלות As Is ועל מנת שנוכל ללמוד ממנו את כל מה שמעניין אותנו, חשוב לעבוד יעיל ונכון. אחד הכלים הטובים (והפשוטים מאוד ליישום) הוא שימוש ב-Secondary Dimension שמאפשר לנו ללמוד ולהעמיק יותר בנתונים עצמם. להלן כמה דוגמאות מצוינות לשימוש ב-Secondary Dimension בהקשר של דו”ח Search Terms ובהקשר של שאר דו”חות ה-Site Search בגוגל אנליטיקס:

הוספת Secondary Dimension של סוג מכשיר (Device Category) – עשוי להתאים לכל דוחות החיפוש באתר ולשפוך אור באילו סוגי מכשירים בוצעו יותר חיפושים וכיצד התנהגו המשתמשים לאחר הופעת החיפוש בכל סוג של מכשיר (זמן שהייה, יציאה, המרה ועוד). דוגמה פרקטית – נניח שאתם רואים שעבור מונח מסוים שמחפשים יחסית הרבה יש תוצאות מעולות בדסקטופ אבל במובייל הוא מקבל מדדים לא טובים (אחוז יציאה גבוה, כמות מסכים נצפית לאחר חיפוש נמוכה וכו’), מה יכול להיות שקרה? יכול להיות שהמונח הזה מייצר הרבה השלמות אוטומטית והדרופדאון מוצג בצורה שגויה / מפורקת במובייל מה שגורם לאנשים להתבאס מהחוויה? או שהמונח הזה מייצר הרבה תוצאות ובמובייל הרבה תוצאות חיפוש מייצרות בעיה גרפית או טעינה איטית במיוחד וכך הלאה. לא תדעו אם לא תכירו את הנתונים ותבדקו

הוספת Secondary Dimension של מקור טראפיק לאתר (Source / Medium / Campaign) – כאן נוכל להסתכל על מקורות ספציפיים והאופן שבו הם התנהגו ביחס לחיפוש, לדוגמה – נניח שנראה שגולשים שהגיעו מפייסבוק מקמפיין מסוים חיפשו יחסית הרבה מונח מסוים שמתחבר לקמפיין, אולי אפשר ללמוד מזה על כך שהעמוד נחיתה שאליו הם הגיעו לא היה ברור מספיק? אולי היה נכון להנגיש להם מעבר נוח יותר לאותו עמוד שלא דרך מנוע החיפוש? האם משתמשים ממקורות / קמפיינים מסוימים מחפשים יותר או פחות מאחרים? האם חיפוש משפיע עליהם יותר או פחות טוב? לדוגמה – יכול להיות שלמשתמשים שמגיעים מפייסבוק ממודעות כלליות של המותג החיפוש לא רלוונטי כי הם רמת נכונות יחסית נמוכה לרכישה, בעוד לקוחות שמגיעים מ-Google אורגני מאוד “סגורים על עצמם” ולכן החיפוש ממיר אותם טוב – מגיעים לאתר, מחפשים, מוצאים בדיוק מה שרצו ורוכשים

הוספת Secondary Dimension של עמוד הנחיתה (Landing Page) – כך נוכל לראות באילו עמודים המשתמש שלנו התחיל את המסע שלו באתר, לפני שהוא ביצע את החיפוש (שימו לב, הכוונה לעמוד הראשוני אליו הוא הגיע ולא לעמוד בו החיפוש בוצע בפועל, אותו אפשר למצוא על ידי Secondary Dimension של Page). מה אפשר ללמוד מזה? בואו ניקח דוגמה: נניח שאנחנו רואים שיש הרבה משתמשים שמחפשים דגם מסוים של שרפרף באתר, זאת על אף שחלק גדול מהם התחילו את המסע שלהם באתר מעמוד של כורסה מסוימת שהולכת מעולה עם השרפרף. לא היה נכון יותר שבעמוד של הכורסה הייתה הפניה ברורה ובולטת לאותו שרפרף? או לחלופין איזושהי אפשרות לרכוש את שניהם יחד “כבאנדל”? יש מצב שכן ובכל מקרה – זה בטח שווה בדיקה…

הוספת Secondary Dimension של עמוד היעד לאחר החיפוש (Search Destination Page) – הניתוח הזה הוא קצת “למתקדמים” אבל שווה לגמרי את הניסיון להבין. כאן נוכל לראות מה התוצאה שהמשתמש בחר לאחר שקיבל עמוד תוצאות חיפוש, לדוגמה נניח שנסתכל על מונח החיפוש ‘חולצה’ כאן נוכל לראות את כל העמודים שאליהם הגיע המשתמש לאחר שהוא חיפש ‘חולצה’ ומה זה ייצר. מה עושים עם זה? בואו ניקח דוגמה: נניח שאנחנו רואים שיש עמוד ספציפי של חולצה שרוב המשתמשים שחיפשו ‘חולצה’ עברו אליו אבל המדדים שלו הם לא טובים – אנשים נטשו את העמוד, העמוד לא ממיר וכך הלאה, אולי יהיה נכון לדאוג שהעמוד הזה לא יופיע ראשון במנוע החיפוש לאותה שאילתת חיפוש? כנראה שכן

הטאב הרביעי – Search Pages:

מאפשר לנו להבין באילו עמודים מתבצע החיפוש, העמודות דומות במהות לעמודות שראינו בשאר הדו”חות ב-Site Search אך חשוב להתעמק דווקא בחיתוכים השונים (סימנו אותם ב-1 עד 5 לנוחיותכם):

Internal Search Terms - Google Analytics
  1. Start Page – The address of the page where the search was started. Pages by name (entrance) These are the pages through which the user enters the site
  2. Total Unique Searches – The number of unique searches performed on the same page within the time frame set for this report
  3. Destination Page – The address of the destination page that the user reached after performing the search (usually the search results page that the user reached) will appear here –
  4. Search Destination Page The address of the destination page reached by the user after clicking on the search results

The less we recommend using the Search report as all these insights can also be deduced from the Search Terms report (by using Secondary Dimension)

Rule # 14 – The search process is Funnel, treat it like this

It is important to measure the entire Funnel of the search, we recommend treating the search as a Funnel with three steps: pre-search, results and post-results. 

Until the moment the search query is performed (pre-search) 

From the moment the search was performed until the transition from the results page (results) 

The browsing process after the transition from the results page (post results) 

Conversion funnel with site search

So let’s start from the pre-search stage, the various sub-stages we have here are:

Search Engine Clicking – What percentage of your users even clicked on the search engine? If you have not applied rules 1-3 and your engine is not prominent enough, then it may also be worthwhile to measure the amount of users to which it was at all exposed.

Start typing in the search engine – what percentage of your users have started the process of typing in the search engine (you will be surprised to find that there is no small drop between the number of people who click on the engine and those who actually start typing)

Finish typing and performing VS AutoComplete Search – What percentage of your users performed a search by performing the actual search or by selecting one of the various autocomplete suggestions (as mentioned, not all of them will be reflected in Google Analytics’ search reports – see below for an explanation of how You can also easily measure the autocompletions) 

By and large this part of the Funnel in the part is supposed to look something like this 

Total entering the site 

Total clicks on the engine 

Total started typing 

Out of them click on search XX and those who clicked on one of the alternatives is Y 

Our KPI is for this stage of the Funnel is the maximum end of the process – there should be as little drop as possible between those who touched the search engine and those who performed a search or chose some result from auto-completion 

Pre-Search Panel

The search phase itself – our KPI is one and only in this case – EXIT RATE as low as possible, if a user chooses to end his visit to our site after reaching the results page, then we failed the task because it means that:

The page was not comfortable enough either 

The results did not match the search either 

The page did not respond quickly enough either 

Any combination of these answers

This part of our Funnel will look like this – the amount that goes to the results page, the amount that leaves the search page compared to the amount that continues to one of the results. 

* Please note – more than one site (mainly ecommerce sites) uses the search pages to create pages dedicated to campaigns, for example – search for 10 products in the search bar and promote the results page as the “campaign page”, for example suppose a particular site wants to promote 8 products on the site That are not under a particular category, an effective way to do so would be to generate a search query with the eight part numbers of these products and then take the link of the search results and promote it as part of the campaign. This method can indeed be a good solution to the need we have mentioned (having a page that concentrates the products of the campaign and will be easy to refer to), but on the other hand can produce a significant glitch when it comes to measuring search results. Why? Because in practice we took the functionality of the search and distorted it for other needs and the data was distorted accordingly

  1. Suddenly a search results page is a landing page for quite a few sessions
  2. The exit rate of the search can increase significantly (because unlike a normal situation where a user searches for a phrase and then has to decide whether the results page is right for him or not, in this case he comes to a landing page which is a search page) 
  3. The amount of searches for those terms found on your search page can be distorted and dramatically increased even though it is not true (which will make the other search phrase appear marginal / irrelevant) 

How is it handled? The most convenient way is to simply add some agreed parameter to these pages and then filter it when you get to look at the reports (either in the report itself or through a segment)

Post Search Stage – At this stage we are interested in understanding what happened from the moment the user moved from the results page to the moment he performed the action we want him to perform (leaving a hand, making a purchase, opening an account or any other action). It is very important to examine whether the very transition of the user in the search process (as part of his journey on the site) caused him to:

  1. Added more to cart? 
  2. Did you progress further to the check-out stage? 
  3. Performed a higher percentage conversion?
  4. Returned for more visits at higher or lower rates?
Post Panel Results

Our last debt before we finish – we talked a lot during the article about the fact that when a user chooses an option out of auto-completion Google Analytics does not look at the process as a search process at all and therefore does not reflect this data in the various Site Search reports. Parameter independently by using GTM and Custom JS) but the simplest is to ask your development team to add to the URL of the product page / article or any other result that a user selects from auto-completion (and does not move it to the search results page) the search parameter your.

For example: Suppose a user typed the ‘sand’ string in your search engine and you did not suggest the ‘black brandon button-down shirt’ option, then instead of going to page:

https://www.mydonain.co.il/ Black Brandon button-down shirt

He will move to this page:

https://www.mydonain.co.il/ Black Brandon button shirt? q = sand

So enjoy all the worlds – both convenient and simple auto-completion and proper reporting to Google Analytics about the search string they searched for and the search performed

And finally an extension

An in-house search engine on the site is one of the best tools for improving conversion rates, experience shows that professional and thorough work on a search engine (first as part of a one-time project to define, fine-tune and arrange and then consistently) can lead to tens of percent improvement in your overall conversion rates. , Yes-yes-tens of percent! So the next time you think about whether to “increase media” or “campaign” you will think twice, maybe it is more appropriate to invest this time and attention in improving conversion rates on your site, through CRO in general and specifically improving and investing in the internal search engine is a good start.

Common questions

Do surfers use search engines on websites?

• A user who has performed a successful search on the site is 2-5 times more likely (depending on which study they are referring to) to make a purchase in relation to a regular user. • According to data from The State of site search in 2019, there was a 56% increase in internal search engine use. • 46% of “serious” customers who come to a site with the intention of finding a particular product, begin their consumer journey by an important search Remember that an internal search engine is an important component in almost any site, especially on ecommerce sites. But the larger the site and the wider range of content and / or products, the greater the importance of the search engine or in other words – the importance of a search engine on a site that sells only one product in 3 variations is minor while on a fashion site with 500 items or marketplace With 100,000 items it is an essential commodity that sits on a very central artery in the conversion process.

How to Build Logic for Auto Complete

AUTOCOMPLETE – is a technique in which a search engine automatically completes the search of the surfer based on the string he has already typed, for example if the user typed – aro, then the engine will offer him to complete the string to ‘closet’, ‘closet’, ‘chimney ‘ Etc. etc’. In essence an auto-completion mechanism is one that is based on an alphabetical order, meaning if I searched for the string ‘arrow’ it will offer completions with ‘a’ after the string (if any) and then ‘b’ and so on. Be sure to calibrate the AUTOCOMPLETE mechanism in a smart and non-technical way, for example – if a person types ‘H’ we want him to first get a complement to the word ‘shirt’ and only then complete to the word ‘corset’. Wants or alternatively interests him in relevant things, this by flooding of – hot searches and recent searches (assuming it is a repeat user) prioritization of autocompletions according to frequency and not according to A-B in a simplistic way. Most search engines will prefer non-domestic and without too much thought (which is of course a mistake). Let’s take an example from Google (feel free to try it for yourself on an endless variety of search strings) – if you start typing the ‘all’ string, the logic is that you will first get options that start with the string you typed ‘all’ followed by the letter ‘a’, meaning ‘prison’ and so on. Then ‘dog’, and ‘calg’ and so on, but in fact this is not the case. The first result you will get is – ‘general’ (‘to’ after the string ‘all’) and then ‘calcalist’ (‘as’ after the string ‘all’) and then ‘rule insurance’ and only in the ninth result will you get the option ‘dog’ *, Which was supposed to be (if you look at it from A to B) higher than all those who came before it … Why? Because Google understands that the alphabetical order of the results is irrelevant and tailoring to the need (user, trends, common searches, etc.) is much more important. So “Google” search experience we probably will not be able to produce on the site, But striving and aiming there is definitely possible (and desirable) so it is very important that your autocompletions be as smart as possible and based not only on a regular alphabetical order, but also on the frequency of searches and if possible relevance to that user. For example – if it is a user who usually buys kitchen products on the site and he started typing on the site (let’s say for example that the site deals with household products) the string ‘Maz’ then it would probably be more correct to bounce the results in the first completions – ‘fork’ or ‘festive forks’ And not necessarily the phrase ‘double mattress’. * It is important to note that the Google search is constantly changing so it could be that if you search for the string ‘all’ while reading the guide you will get another position for the ‘dog’ proposal (and not ninth place as in the example we mentioned), the goal was of course to illustrate a point and it – Google does not refer Only to order A-B in order to offer perfection, It is recommended to separate them in a very clear graphic way and divide the area between them in a way that illustrates the frequency of the search. For example – if a particular user searches the NI string on a shoe site, this string can of course be relevant to any pair of NIKE shoes but also to a NIKE brand page or an article on the NIKE site so it is very important for the user to understand before choosing the auto-complete option – Does he choose a particular model (then clicking on the completion should take him to the model page) or does he choose a brand page (then choosing completion will take him to a general page / brand page) or he chooses a page that will take him to a content article and so on. Beyond the fact that the user should understand where he is going, it is very important that graphically the relationship will be relevant, if NIKE has 100 products, 6 brand pages (for all kinds of campaigns, etc.) and 15 content articles on the same site, it is unlikely that suggestions for completion will be displayed. In the same relativity and prominence, And it is clear that in such a situation (assuming it is an ecommerce site of course) the product pages themselves should get much more prominence and space than the article or brand pages, how much more prominent? It very much depends on how people behave on the same site – if it is a site for selling shoes with extensive activity in the content world and the data shows that customers who interact with the site’s content then tend to buy more, then maybe the articles should gain significant weight and if not, then maybe just a mention. at all. The options are varied, but what is important to remember is that this dose is important to perform on a data basis and to improve from time to time.

Where to place the search engine on the site?

Place the search engine at the top of the screen (in the area of ​​the logo and maple of the site) and place it outside a sidebar (do not accidentally stick it in a hamburger on a mobile) If there is no place to spread the search bar across the entire screen 1/2 of the width, in Hebrew sites it is recommended to put it on the left side (because our eye begins the scanning process of the site from left to right in an inverted F shape) and in English sites it is recommended to put it on the right side

What do you do when the search engine on the site does not return results?

If the query the user was looking for has an alternative that you can understand, then suggest it or even define it without asking (just give an indication that this is what you did). For example – if you are an Israeli electrical products site and the user searched for the query ‘nygi’ then over time you should have taught the system that ‘nygi’ is actually a ‘charger’ only in incorrect typing of “Hebrew – English”, in which case it is recommended to display the search results For a charger with an indication that you made this change for it. For example – if you accidentally type in Google the query – ‘chyuj shrv’, Google’s engine will immediately recognize that you are actually looking for – ‘home insurance’ and will show you the relevant results plus an indication that it performed the (so convenient) gesture of auto-repair for you. The mistake If you could not understand “what the poet meant”, you need to provide the user with a clear indication of what query he was looking for and that you could not find a matching result, but at the same time offer him – alternative suggestions, communication with customer service, A convenient way to perform another search, etc. (the main thing is not to leave the site due to the unsympathetic experience). Some things worth considering to include on the “No Results” page: Gentle apology / identification with the situation Suggestion for checking spelling errors Suggestion to contact customer service and accessibility of an easy way to contact (in fast ways not by email or fax but chat / WhatsApp or phone) Suggestion for other products that may be of interest, each site and each area with its accuracy but in general it should be – hot products that may interest you or hot categories Possibility to perform another search in as simple and convenient a way as possible

How to analyze the performance of the internal search engine on the site?

Inside the GA account, under Behavior category >>> Site Search, where GA’s ready reports for us to search the site await us. Please note that these reports will be filled in automatically from the moment you properly define the search (the information will be collected from now on and not backwards) Site Search Overview – as its name implies, here we can see general data about site searches (you can of course use all GA ‘ Such as cutting dates, using segments, etc. Usage – here we can learn in detail about the performance of the sessions in which the site was searched (quantity, abandonment rates, new users, page views per session, conversions, etc.). Google Analytics shows us by default the comparison between sessions Search Terms – Search Terms – This report goes into the depth of the search queries themselves and provides us with concrete data for each of the search queries searched on the site. From this report we can understand the quality of the answer we were able to give the user for each of the searches – was he forced to leave the site or perform new searches? How deep he was on the site after performing the search – which may indicate a search that yielded good results for him and so on. Search Pages – allows us to understand which pages the search is performed on, the columns are essentially similar to