Eyerim - online optician for anyone

Data analysis

Next, we analyzed all kinds of data related to the users, the way they move on the site, also taking into account how they move on desktops and mobile devices.

We used four tools to conduct an analysis.

Based on the qualitative data obtained from Hotjar, we could learn where the users are encountering problems due to badly prepared interface or about programming errors that prevented access to the shopping cart. Hotjar also allowed us to review the common behaviour of users.

The quantitative data provided by Google Analytics has enabled us to determine the bounce rate and to obtain information on devices used to browse products on eyerim.com. It also provided very informative demographic data.

Hatimeria's QA engineers have been also monitoring basic indicators of well performing ecommerce system, using NewRelic and Google Analytics for last few months. Collating data gathered by them with UX team conclusions, allowed us to prepare more complete gameplan and prioritize goals.

Defining problems

After pre-interviews with users, we have defined a few main problems

1. 1Filters on mobile devices were not intuitive.
2. 2The product information on the product card was named differently in the shopping cart.
3. 3Users wanted to be able to check how the product looks like at different stages of the shopping path.

Then we conducted an expert analysis which showed us more issues to solve:

1. 1Long page loading time and server response.
2. 2No descriptions inside input and select fields.
3. 3CTA hidden below the fold.
4. 4Users having a problem with going back to shopping after removing the product from the cart.
5. 5People escaping the cart after clicking on the product photo in the basket.

DevOps improvements

Simultaneously with the work of the UX & UI team, our DevOps team analyzed the data obtained in the initial phase of Blackfire and NewRelic. We have identified bottlenecks and planned to level them in cooperation with server administrators.

The most important points that contributed to increasing the efficiency of the eyerim.com store were:

• implementation of ElasticSearch to work with and speed up category pages

• optimization of SQL queries efficiency on a category, product, and home pages

• improvement of cache usage to relieve the database work

• improvement of frontend (images, CSS, compression) according to Google specialists

• optimization of indexers and other background jobs that caused errors disrupting store’s operation

• removing unnecessary records from the database to make it lighter

• general server clean up to remove unnecessary files slowing down the IO operations

• general JS scripts cleanup to remove scripts that were unnecessary or caused errors and increased page load time

• the general review of web transactions, identifying suspected/unnecessary traffic and reducing it

In addition, we have eliminated numerous errors that have not slowed the page but have had an impact on its availability, eg: Server crashes and Server Unavailable.

The work we carried out brought positive changes in:

• page load

• response time

• server uptime

Conversion rate vs loading time

We recorded a conversion rate increase by almost 30%, comparing the second quarter of 2017 (1.45%) with the second quarter of 2018 (1.87%). In turn, we analyze the first quarter of 2018 (1.69%) with the second quarter, and we see a significant increase of ~ 12%.

The increase in conversion is strongly connected with the decrease in page load time and server response time. The chart below shows how the page loading time decreased between the first and the second quarter of 2018.

Testing

At each stage of implementing changes, our QA team ensured that the changes we implement were perfect in every respect.