53% of total revenue comes from 17% of users from search: a case study of a book e-commerce store in Ukraine

20 March 2023

Share:

This picture best demonstrates that the audience using the website search is the most conversion-oriented.

The convenience and quality of your search directly influence the store's sales.

About the company "Knyharnia Ye", book-ye.com.ua

  • The largest bookstore network in Ukraine
  • Operating in the market since December 21, 2007, with a focus on Ukrainian books since its inception
  • Has 35 bookstores in 18 cities of Ukraine + 3 under construction, as well as an online store, book-ye.com.ua
  • Offers over 21,000 book titles
  • Hosts literary events and presentations

The specifics of the book segment are such that people often type the book title or author's name as they remember it, so the share of search queries with errors is much higher here than in other niches, reaching almost 8%.

For comparison, in a grocery store, this figure is 4%, while in a pharmacy, it is 12%. The more complex the topic, the more errors. And if your search can’t handle them, the customer receives irrelevant or no results at all.

What we implemented at "Knyharnia Ye" to make the conversion rate from search almost 5 times higher within just one month compared to visitors who don't use search on the site is revealed in the case study.

Contents:

  • What the search was like before we came: 5 key problems
  • How we solved the problems before we arrived
  • What we implemented
  • How quickly we implemented it: spoiler! in 4.5 hours
  • Results after implementation🔥
  • Why they chose Multisearch

What the search was like before we came: 5 key problems affecting sales

The main question the company approached us with was: "Does the Multisearch functionality support synonyms in Russian/Ukrainian? For example, if the query is ‘сказки,’ can it show ‘казки’?"

The situation was that the website only had one language version—Ukrainian—so when a user searched for a book in Russian, the search treated it as an error and didn’t return any results.

"Knyharnia Ye" was losing customers daily who searched for "Роулінг" with the Russian "и," or instead of "Будинок легенд," they were looking for "Домик легенд."

Before Multisearch, the CMS module’s search only looked for exact matches, so it didn’t handle:

1. Incorrect keyboard layout, which occurred frequently when users entered the Ukrainian title of a product using an English keyboard layout: ;bcnjrbq ghbyw, ;bdb, ;bhfhlsd, ;flfy ltgti vjl.

2. Queries with errors, when:

  • Author names were confused—Joan Rowling instead of Joan Rowling, Joan Roling.
  • Letters were swapped—"Atlas of Aminals" instead of "Atlas of Animals."
  • Extra characters were added or missing—"Vitka + Halya" instead of "Vitka Halya,"
  • the Russian "и" was used instead of the Ukrainian "і." in “Повість”

3. The transliteration algorithm, when people wrote English names based on how they sounded in Ukrainian — Брідгертонс or Бріджертони instead of Bridgerton.

4. Category relevance. If a person queried "detective," but the book title didn’t include that word, the old search didn’t return a list of books in that genre.

5. Author names and publishers. The search didn’t return any results for these queries.

Examples of queries that returned no results:

How they solved the search problems before us

The company considered two options: tweaking the CMS module or developing their own. After analysis, both options were ruled out, and marketers immediately began searching for a ready-made solution.

Oleksandr Polonsky
Oleksandr Polonsky

Digital Marketing Consultant


I don’t think we would have risked developing our own module. We understand that when developing an IT project, deadlines are always delayed, even when you have a clear specification, and without it, the development could take a year or more.
Now, looking at the numbers, I think if we had developed the module ourselves, maybe in a few years we would have reached the break-even point, but all those years we would have had worse sales results.

What we implemented

 The main problem with the old search was that the CMS module didn’t account for typos or synonyms and only returned queries that exactly matched the book title or author name.

To quickly improve results, the company took the first two steps in setting up smart search:

1. Installed the plugin, and search immediately started:

  • Handling typos, incorrect layout, and transliteration.
  • Performing searches for authors and publishers’ names.
  • Determining the category of a product.
  • Returning more complete results.

Managers of the online store no longer receive complaints from customers about not finding products in the search, although such requests were common before.

2. Added the "Product Popularity" feature. Now, the most frequently viewed and purchased books are shown at the top of the results.

This feature works by learning from client sales over the course of a month after the script is installed. It changes not only the search results but also generates a report on the conversion products and categories.

This is the unique analytics of Multisearch, which helps marketers instantly understand which queries bring the most revenue. The data is updated every 3-5 minutes.

Here is what the Multisearch conversion analytics look like:

In the screenshot, you can see that the query "Agatha Christie" gathered 1435 conversions. Now imagine if it wasn’t functioning on the site 😲.

How quickly it was implemented: Spoiler! In 4.5 hours

 For a quick start, marketers decided to install the plugin, and the process took 4.5 hours.

  • November 8, 9:15 AM: "Knyharnia Ye" provided us with the feed.
  • November 8, 12:18 PM: We sent them the information for search setup.
  • November 8, 1:53 PM: They installed the plugin on the test domain.
  • November 9, 10:26 AM: The client put the search on production.

 Done! At that moment, the online store’s search started working properly.

15 minutes
This is the average time for the plugin integration process via Google Tag Manager. The fastest result was 7 minutes.

The instructions are so simple that you don’t need help from IT specialists.

In the near future, the company plans to improve the website's usability and immediately implement the API, which will provide the online store with unlimited search possibilities:

  • Smart autocomplete
  • Dynamic filters
  • Direct transitions
  • Default autocomplete
  • Manual management of autocomplete

Results After Integration

 We analyzed the changes in key metrics on the "Knyharnia Ye" website one month after implementing Multisearch.

Google Analytics data:

Important❗ The transaction rate in search increased despite an overall decline in the site's transaction rate.

This clearly demonstrates how much the search quality improved. While the site's overall transaction rate dropped by 10%, the transaction rate in search grew to 16%.

Without the overall decline, the search conversion rate would have risen to approximately 30%.

Additional data confirms that users started relying on search more and making more purchases through it:

  • 44.58% increase in unique search queries. Users realized the search works and started using it 2.5 times more actively.
  • ❗Revenue share from search increased from 44% to 52% of total site revenue.
  • 53% of total revenue comes from just 17% of users who use search.
  • ❗Conversion rate from search is nearly 5 times higher than for visitors who don't use search.

Key nuance in the plugin❗
The client's analytics does not include search queries that return no results or where users do not click on results. This means the total number of search queries in analytics is usually lower after installation. 

However, as seen in the screenshot, the share of visits with search usage increased from 16.06% to 17.35%, even as overall traffic grew during the holiday season.



NEW. Feature Update: Automatic Replacement of Russian Words with Ukrainian Equivalents

Previously, online stores had to manually add Russian book titles, author names, and categories as synonyms to ensure Ukrainian search results. Now, the system automatically translates queries and delivers results immediately.

Why They Chose Multisearch.io

Oleksandr Polonsky
Oleksandr Polonsky

Digital Marketing Consultant


Installing Multisearch, even during the free test, instantly solved our issue with typos.
And I’m really impressed with your support. When something breaks, I send a message in chat, and within 30 minutes, everything is working again.
Working with you brings peace of mind — that’s the best way to put it.

Want to see how a new search can boost your store's sales from day one? Integrate Multisearch here. 14-day free trial.

Share:

Автор