In 2010, Steve Phillips took a new advertising-research method to prospective clients. They were pleasant about it, which in business can be a dangerous sign. The method was smarter, more emotional and, by his account, not different enough to make anyone abandon the tests and historical norms they already trusted. The interesting response came somewhere else: clients perked up when he talked about getting familiar research done much faster and for less money. The rejection contained a product brief.
Phillips and a small crew spent nights and weekends building an automation engine for roughly two years. In March 2013, the first customers arrived at ZappiStore, an online storefront where companies could buy automated versions of established research methods. The name was not coy. It was meant to be an app store for market research.
Today the company is Zappi, with more than 300 employees, offices in Boston, London and Cape Town, and a list of publicly named customers that reads like a supermarket aisle: PepsiCo, McDonald’s, Mars, Heineken, Reckitt and Colgate-Palmolive. It sells software for testing product concepts, packaging and ads, plus continuous brand tracking, AI-assisted analysis and professional services. The buyer is usually an insights, innovation or marketing team that needs evidence before committing a large budget to an idea.
The problem was never a shortage of surveys
Traditional custom research can be excellent, but its operating rhythm is awkward. A team briefs an agency, negotiates scope, waits for a questionnaire, waits again for respondents and receives a report weeks later. By then the packaging may be locked, the media slot booked or the loudest executive already convinced. The work is also expensive enough that teams ration it, testing the final candidate instead of using feedback to improve a series of candidates.
Zappi standardizes that repeated work. A customer chooses a proven methodology and audience, uploads the creative or concept, launches a study and reads results in a dashboard. Automation handles survey programming, fieldwork coordination, common analyses and reporting. The company does not claim every question fits a template; it sells strategic guidance and custom services too. But the common path is software, not a chain of email.
What does it cost? Zappi does not publish a simple price card. It sells Premium and Enterprise subscriptions, then credits that pay for research projects, trackers and services. Audience, geography, sample and support change the bill. The useful public comparison is relative: in a commissioned Forrester study, interviewees said the composite organization cut research-campaign costs by 40 percent. A personal-care insights executive in the study said a concept test cost at least 30 percent less. Those figures model a large, repeat user, not a promise to a curious startup with one survey.
The modeled economics
A large composite customer cut campaign costs by 40%, saved 10-15% of insights staff time and reached payback in under six months.
The product changed when the company stopped selling speed
The first ZappiStore proposition was faster and cheaper research from a shelf of methods. That got attention, investment from Kantar in 2014 and a £12 million round led by Prime Ventures in 2016. It also carried a ceiling. A store helps a buyer complete a transaction. An enterprise platform has to become part of how a company works.
By 2018, Phillips said the app-store days were over. Zappi dropped “Store” and retired the “faster, cheaper, good enough” posture. The platform would automate process, scale expertise and help insights departments make better decisions. That sounds like positioning copy until you see the architectural implication: research could no longer end as an isolated score.
Zappi’s sharper difference is the learning loop. New studies use consistent measures and sit beside past studies. Teams can compare a fresh snack concept with earlier snacks, inspect what drove purchase interest, find relevant norms and carry those lessons into the next brief. Every project adds to an organized proprietary data set. The immediate result answers today’s question; the accumulated data improves tomorrow’s question.
McDonald’s describes the shift neatly. Instead of testing to earn a good score, its team uses standardized data to test, learn and iterate toward the strongest version. For a new shake flavor, a researcher can revisit the drivers of interest across previous shake concepts and see how consumers described what they wanted changed. The old work becomes a shortcut, not an archive.
PepsiCo helped turn a tool into a system
Zappi did not invent this operating model in isolation. PepsiCo became a demanding co-creator. Its Labs program says the companies developed solutions that could handle thousands of pieces of brand content, then scaled them globally. Within nine months they had produced deployable tools; PepsiCo says the broader partnership cut its average ideation-to-innovation time by 60 percent and trained more than 400 insights professionals.
That collaboration produced a useful enterprise lesson: do not ask a lighthouse customer only for a testimonial. Ask it to supply edge cases, outcome data and enough organizational complexity to break the product while it is still fixable. Zappi’s Amplify advertising system was co-developed with major advertisers and validated against market-mix outcomes. PepsiCo executive Stephan Gans says creative effectiveness improved by about 30 percent across its advertising. That is a customer-reported result, not an automatic outcome of opening an account.
The current product line follows the same logic across three jobs. The Innovation System screens and improves ideas, concepts and packaging. Amplify tests advertising from an early thought through TV, streaming, digital and static executions. Brand Health Tracker continuously measures awareness, consideration and other indicators. The connective tissue is more important than the menu: product, ad and brand data can inform one another instead of living with separate vendors.
Slower cycle
Project data
Enterprise guidance
Connected data
More setup
Fewer built-in norms
AI is useful here only when it remembers something
Zappi’s 2025 AI Concept Creation Agents generate test-ready product concepts from a customer’s past concept-testing data, or from category norms when proprietary history is thin. The concepts go through Activate It testing, and the response flows back into the next generation. Zappi reported that customers created concepts 30 times faster at one-twentieth the resource cost, with 90 percent rated highly distinctive by consumers. Again, these are company-reported aggregate outcomes. The more durable idea is that generic generation gets grounded in a specific brand’s evidence.
That is also where Zappi sits in a crowded market. Kantar, Ipsos, NIQ/GfK and YouGov offer deep research expertise and global infrastructure. Qualtrics, Attest, Suzy, quantilope and Toluna cover overlapping self-service, agile or experience-research jobs. New AI tools can summarize interviews and draft surveys quickly. Zappi’s claim is not that nobody else can run a concept test. It is that its systems, norms and connected history make repeated product and advertising decisions better as usage grows.
What to steal - and when not to
The Zappi playbook in four moves
- Listen for the urgent problem hidden behind a polite reaction to your original idea.
- Automate the repeated process, but preserve expert help for unusual questions and organizational change.
- Use a demanding customer to co-create the system and validate it against real business outcomes.
- Keep the output structured and reusable so each transaction improves the next one.
There are conditions under which this does not work. Standardization is weak when the question is genuinely novel, the audience is rare, the cultural context is delicate or the team needs exploratory conversation rather than a score. Automation cannot rescue a confused brief or poor sample. A company running two unrelated studies a year will not build much of a flywheel. Nor will a multinational benefit if teams refuse common measures, hoard their data or keep buying exceptions until the “platform” behaves like a collection of custom projects.
Zappi itself had to change leadership for the next phase. After the $170 million growth investment led by Sumeru Equity Partners in 2022, longtime founder-CEO Phillips moved to Executive Chair and Chief Innovation Officer in October 2024. Aaron Kechley, an enterprise SaaS operator, became CEO with a mandate to sharpen execution and commercial growth. It is a familiar handoff: the person who sees the break in an industry is not always the person who should build every process at scale.
The company reported more than 150 product releases in 2025, added brand tracking, tested 90,000 ideas and passed 1.2 billion consumer data points across its ecosystem. Those numbers sound like activity until viewed through the original failure. A slightly better method could not persuade buyers to change. A faster workflow did. A connected memory may persuade them to stay.