Overview
Business idea
Customers could buy food adapted to the parameters of their cats while adjusting the amount and frequency of payments.
Purchase flow
The customer purchased by going through a few-step form. After completing the record, the system proposed the food amount, type and flavours adapted to the given animal. Then, the ordering party could edit the set, size and frequency of food. The first trial package of food was at a promotional price.
Problems
- A small number of customers
- The high cost(s), poor ad performance
- Widespread resignation after receiving a trial box.
- Low brand trust
- Critical technical errors on the front-end and back-end.
Goals
The main goal set before me was to improve the degree of acquiring new customers. The second goal was to extend the customer’s payment period.
Approach
To learn more about the problems, discover their sources and causes, and identify blockers and errors, it was necessary to do quite a lot of research.
From the beginning, the platform had essential integration with Google and Facebook analytics. There were also logs from the payment operator. On the other hand, technical issues could be tracked with the help of Jira task history and project history in the GIT repository. A valuable source of information is also customer correspondence with customer service, opinions about the company and user activity.
Obtaining some clues, clues to finding the causes of errors required more data, so I recommended:
- More accurate segmentation of users in analytics by location, number of animals, basket value, location and degree of funnel completion.
- Adding events sent after completing each step of the funnel.
- Implementation of user session recording.
- Implementation of heat maps, scroll maps.
- Adding a form (several checkboxes) in the cancelation flow to obtain information about unsubscribing.
- Customer service contact with a short survey for current, former and would-be customers. The survey was to show the strengths and weaknesses of the platform.
- Conducting website tests on various screens by manual testers who do not know the website. The testers had to sum up different customer cases.
- Gather information about technical problems of the website, their duration and source.
- Monitoring keywords about the company, service and products to get to know people’s opinions on these topics.
Due to the lack of all data and the simultaneous need to implement corrections, I made the following assumptions:
- Based on the knowledge and studies previously performed in the company.
- Using my knowledge and experience.
- Looking for analogous cases in analyses and case studies on the Internet.
- Analyzizng the actions of competitors.
- Concluding from research results when more than 50% of the results (with at least 100 trials) indicate certain decisions, or I await the results on a group of 1000 cases.
Initial thesis
- Visual and technical errors are behind the basic conversion problems.
- The diet selection form and the screen are too difficult.
- Advertisements are targeted at people not in locations served by the company.
- The problems with cancelling a subscription, the high price after the end of the trial, or the problem with payments. But before that, you need to research customers’ opinions about the service and products.
Research
Phase 1 – Instant fixes
To take corrective action, I immediately started checking the funnel on various screens and pointing out elements and errors, which minor graphical corrections will improve the company’s assessment in the eyes and improve the usability base. In phase 1, we have already managed to get a few per cent higher user engagement. First of all, check out with many fields and functionalities such as order summary, order calendar, delivery address validator and various payment methods was subject to a deep check—a recommendation to check the correct operation of the analytics.
Phase 2 – Analysis of the knowledge accumulated by the company
Analysis and familiarization with the information and business studies obtained so far.. Familiarization with the characteristics of the company, product.
Phase 3 – Research based on existing data
Based on the initially collected data, it was possible to verify:
Google/Facebook Analytics, Jira:
- Check the conversion graph over time. Finding disturbances in the chart, mainly dips. And then an attempt to correlate them with campaign changes, but above all, with development work and technical problems.
- Checking the conversion results broken down by device type.
- Finding the location with the lowest conversion and checking if the platform supports their orders.
- Checking the flow of users between pages to determine a potential conversion blocker.
Checking logs of the payment operator and order history:
- Verify how many of the orders failed due to payment or logistics problems and the reason for the issues.
- Checking how many subscriptions have been cancelled due to payment or logistics issues.
- An attempt to extract repetitive problem cases.
Phase 4 – Communication with customer service, user feedback
- An attempt to pinpoint the primary users’ problems with the company, product, and website to obtain insights about the reasons for not converting or reasons for cancelling the conversion, as well as customer needs.
- Customer service was responsible for contacting customers to conduct surveys to get more information from would-be, current and former customers.
Phase 5 – Analysis of the business environment and competition
- Examining how the competition performs similar tasks. What advantages and problems does they have?
- Comparison of how the company compares to the competition.
Phase 6 – Solving technical issues related to production deployments
At the study’s beginning, it turned out that some problems were helpful with errors after frequent production deployments. It was necessary to check how such a situation came about and what the consequences were
Phase 7 – Implementation of manual testers to the project
- To speed up obtaining larger amounts of data, a wide group of testers, over 1000, was employed, which was divided into different groups due to the type of device, number and characteristics of animals, food preferences.
- Based on user feedback and user recordings, it was possible to analyze problems with testers on an ongoing basis and in-depth.
Phase 8 – Conclusions and solutions based on the fastest obtained conclusions
Phase 9 – A/B tests
Users were randomly sent to the old or new version of the funnel to assess the effectiveness of the fixes
Phase 10 – Enhanced analytics analysis after initial improvements
Phase 11 – Analysis of user sessions, heatmaps, scroll maps
- Search for blockers in the entire flow or on individual steps.
- Observation of user behaviour on the food preferences selection screen.
Phase 12 – Monitoring “signals” from users
- Conversations with customers.
- Monitoring communication between customers and customers service.
- Monitoring mentions and activity about the company
Conclusions
Low conversion rate
Basic problems
- Front-end of the mobile version makes it challenging to purchase.
- Errors in the functionality of hc on the front-end side./li>
- Back-end errors periodically prevent the purchase.
- Poor design quality.
- Aesthetic imperfections.
Checking for problems with the store’s credibility, or rejecting or even preventing the purchase
Ads campaings
The main source of traffic to the website were advertising campaigns targeted at customers from the supported regions. However, there have been cases of campaigns reaching people whose location was not supported.
The funnel – Formularz zamówień
Already done
- The form is divided into several steps.
- Progress and several steps are shown.
- Most of the steps are of similar complexity, which gives the user how much time he needs to complete them.
- Simple fields to complete.
- The form remembers some data and copies it on further steps.
To improve
- Complex case handling with multiple animals.
- Simplifying the first step. Instead of downloading personal data, only zipcode validation or animal input.
- Considering the possibility of zipcode validation in the first step if the users come from ads from a given location.
- Improving selection of fields in sections.
- Equalizing the time spent on each slide./li>
- Simplifying food choices—resignation from extensive editing possibilities for simple pieces with a recommendation.
- Improving of the cat breed selection box.
- Hints on the cat’s weight are based on the rest of the data.
- Preventing the selection that the cat is still growing and, at the same time, setting the age above one year.
- No validation of fields and error messages.
- Improving the mobile version and better adjust buttons and fields for mobile devices.
- Funnel loading time, in particular, the calculation of the nutrition plan.
- Adding additional descriptions and hints to help users and indicate the purpose of completing the form.
Checkout
The final screens of the shopping path are made of modified Woocommerce templates and several external solutions. The visual hierarchy and aesthetics issues need to be improved, as well as basic usability, especially on mobile devices. Through checkout is not optimal for the shopping path. It contains many fields to be completed with address validation which works very slowly.
Validation, in particular, becomes problematic when an error is made in the address. In addition, the address is also entered in the previous step, and even photocopying it still visually shows a high degree of complexity. The site offers a calendar to choose the date of the first delivery, which works quickly enough. Also, the thank you page itself requires quite a lot of time from the client.
Choice of food preferences
Users had the option of editing the diet proposed by the system. They could change the calorific value and choose the taste and type of food from a dozen options. The data and recordings of the session showed that users had some problems with this purchasing step. Approx. 20% of users dropped out at this stage. They just left the page or experimented with different options and didn’t move on.
Product quality
Among people highly aware of cat nutrition, the company’s offer did not appeal at all. Customers rated the product poorly or as an uncertain choice.
Lack of credibility
The website’s appearance, the lack of reasonable opinions, well-developed social media, and influencers around the brand caused, or at least an accurate presentation of the product; all this meant that users did not want to be bound by the subscription model with this company.
Loading and response time
The website showed significant drops in speed at the time of loading, calculating the nutrition plan, and converting the price. The exemplary price conversion took up to 2 seconds, and there was no visual communication that the system was working and not crashed. Users who refreshed the page either lost some of their settings or were reverted to step one and left the page.
The number of times prices are shown and their inconsistency
Prices were posted in many places. The sheer amount of this communication could be outdated. Also, the prices were given for the entire order, the trial in the promotion, and as a daily one; moreover, the delivery cost appeared separately. At further stages of the check, due to technical errors, the prices were not consistent with each other.
Technical issues
The website has repeatedly encountered technical problems that prevented the purchase. Errors hampered the funnel or the cash register itself. At checkout, it happened that the calculation of prices did not work correctly, payments, address validation, or the calendar
Subscription cancelation
Clients opinions about the products
Some customers expressed dissatisfaction with product quality, product quantity (too much), problems with deadlines.
Payment issues
Very often, any problem with the renewal payment resulted in the suspension of the subscription. There was no system for repeating payment attempts and contacting customers with customer service.
Technical issues
Subscriptions were unresponsive due to some intermittent issues with logistics integrations why customers quit.
The Subscription’s cost
A common reason for unsubscribing was that the subscription basket was full. The first order was often much lower due to the reduced calculation time and discount. In addition, the inconsistency of prices on the front end gave the false impression of lower prices.
The trial package period was two weeks, but sometimes some food plans were written out for a much extended period. Shorter delivery times were also problematic for some customers. Most subscriptions are based on 4-week periods,, which users are used to.
Problems editing subscriptions and payments
The study also outlined the next work stage, which is modifying my account. Users quit due to numerous problems with my performance and, thus, with the option to edit subscriptions or edit payment methods.
Navigation issues, saving form state
Often going back in the form from the browser’s history, or refresh caused breaking the funnel.
Solutions
Sample wireframe for the new funnel
Problem
Solution
No credibility
- No credibility.
- UI quality improvement.
- Placing customer reviews, approvals, certifications, veterinarians’ opinions.
- Showing the number of orders, serviced cats
Increased involvement in completing the form
- Improvement of copywriting in the funnel to indicate the “reward” after completing the form.
- Clearer indication of the short time needed to complete the form.
- Adding hints for fields that caused problems based on session recordings.
Front-end functionality improvements
- Improved navigation.
- Fixed form saving.
- Improvement of the race search dropdown – adding a search engine, placing the “I don’t know the race” option at the top.
- Improving the operation of interactive forformraz elements.
Slow loading/respone time
- Optimization of front-end <> back-end communication, and optimization of processes on the back-end.
- Improving the UX related to loading. Communication with the client that everything is going according to plan.
Simplification of the form
- Opting out of zipcode validation in the first step for ad users,
Improved grouping of fields. - Optionally show some fields depending on previous selections.
Simplifying the choice of food preferences
- Giving up the option of adjusting flavors and calories.
- Presentation of several ready-made sets with an indication of automatic selection of recommendations.
Address validation
- Improvement and acceleration of address validation.
Price presentation
- Improved price calculations
- Limited number of places to present the price.
- Unification of displaying prices for a week or for the entire order
Including shipping costs in the cost of the diet.
Price shock
- Extending the effect of the starting coupon on 3 orders while reducing the price reduction on each order.
- Standardizing the length of the settlement period to 4 weeks.
- An attempt to add another factor to the delivery time calculation, such as a psychological price barrier.
Technical errors
- Implementation of procedures and automatic and manual solutions aimed at checking the entire website with each deployment of a new version of the code.
- Implementation of solutions that help to catch orders with problems with payments and logistics.
Payment problems
- Implementation of solutions to obtain payment
- Improvement of recurring payments
- Improvement of editing payment methods by customers
- Implementation of notifications about payment problems
- Implementation of procedures for customer service to contact customers
Summary
This project struggled with several problems from various topics related to building a web application that sells a physical product. From the side of the website and the order form, it struggled with fundamental problems that should not be present to such a large extent on the live version. In addition, frequent failures and errors also disturbed the analysis.
Removing the fundamental problems visibly improved the statistics but stabilized them.
As a result of A/B tests, simplifying the choice of the type of food improved the conversion rate by about 10% and showed that the idea of the whole business needs to change slightly. However, users were not entirely interested in the complete karma edition.
Despite continuous technical improvements, subscriptions were cancelled due to the product and prices.
Photo by Alicja Gancarz on Unsplash
