Ten months of QA for a taste-testing platform where the data is the product

Websiteflavorwiki.com ↗

Client

FlavorWiki

Industry

Consumer insights · Food & beverage

Engagement

2022 – 2023 · 10 months

QA team

2 full-stack QA engineers

Platforms

Websites · Mobile Web · Consumer Portal · B2B Platform · Visualization System · API · Database

About the project

FlavorWiki is a digital market research and consumer insights platform. Food, beverage and consumer product brands run taste tests with real consumers, who try products and answer structured sensory surveys on their own phone or laptop. FlavorWiki then uses artificial intelligence to turn that data into flavor profiles, preferences and market-by-market insight.

That makes the platform two products in one: a consumer portal simple enough for any taster, in any market, to complete a survey accurately, and a B2B platform where brand and research teams build studies and read results through charts, dashboards and exports. If a survey records the wrong answer or a chart misreads the data, a brand makes decisions on bad insight.

QAonTOP provided two full-stack QA engineers for a ten-month engagement from 2022 to 2023, working in Jira and Slack alongside FlavorWiki’s product and data teams.

What made this hard

In market research, a quiet data error is worse than a visible crash. FlavorWiki needed every answer captured correctly and every chart to tell the truth.

1. CHALLENGE 01

   Surveys must record exactly what tasters mean

   Scales, rankings, branching logic and timed tasting steps. A mis-mapped answer or broken skip rule silently corrupts a whole study.

2. CHALLENGE 02

   Charts are only as good as the numbers beneath

   Flavor profiles, preference maps and segment comparisons had to match the raw responses across filters, markets and exports.

3. CHALLENGE 03

   Consumers on every kind of device

   Tasters join from older phones, tablets and laptops in many countries and languages. The portal had to behave the same on all of them.

4. CHALLENGE 04

   Two audiences, one data model

   A change made for brand users in the B2B platform could break the consumer portal, and the reverse. Both had to be tested as one system.

How QAonTOP tested FlavorWiki

1. 01

   Map each study end to end

   Traced a study from creation in the B2B platform to consumer invitation, survey completion and final report, and turned every step into test cases in Jira, with daily triage over Slack.

   Tools  Jira · Slack

2. 02

   Validate survey logic and data capture

   Tested every question type, branching path and validation rule, then queried the GraphQL API through Postman to confirm each answer was stored exactly as given.

   Tools  GraphQL · Postman

3. 03

   Visual and data-visualization testing

   Compared dashboards, charts and exports with expected values calculated from known test responses, and ran visual checks on layout, labels, legends and colour across screen sizes.

   Tools  GraphQL · BrowserStack

4. 04

   Every browser, every device

   BrowserStack real devices and mobile emulators covered the consumer portal on the browsers and phones tasters actually use, including slow networks and translated interfaces.

   Tools  BrowserStack · Emulators

5. 05

   Usability and load

   Usability walkthroughs of the taster journey to reduce drop-off, plus JMeter load tests for large panels submitting surveys at once and for heavy dashboard queries.

   Tools  JMeter

The engagement in numbers

Before and after the engagement

What we covered