Frequently asked questions about 99takes
Answers about getting started, pricing, the simulation and our service. Don't see your question? Email run@99takes.com.
99takes is a tool inside Claude that tests landing pages, ads and other marketing copy on personas from your audience before anything goes live. It shows which version the personas prefer, what makes them hesitate, and why. 99takes is a product of Tappla UG (haftungsbeschränkt), Germany.
An account at run.99takes.com. It comes with 350 credits that are valid for 30 days, and you don't need to enter payment details. Add 99takes as a connector in claude.ai or run a single command in Claude Code (steps at www.99takes.com/setup). Then give Claude your landing page or paste two versions of your copy. There's no code snippet to add to your page and no tracking tool to set up.
99takes works in Claude: as a connector in claude.ai and the Claude desktop app, and in Claude Code. There is no version for ChatGPT, Gemini or Copilot and no public REST API; in Claude Code you can connect with an API key from your account instead of signing in. What goes in: a page by its URL, copy you paste, competitor URLs, and your own interviews or support tickets as evidence. What comes out: the result in your chat and, on request, a report with its own link to share. There are no plug-ins for CMS, analytics or ad platforms.
Yes. Once you sign up and confirm your email address, you get 350 credits, valid for 30 days, with no payment details required. Comparing two versions with 12 personas, each also reading with the other personas’ jobs in mind, costs about 73 credits. A 1:1 comparison, with each persona reading with only its own job in mind, costs about 19. After that, you buy credits when you want to keep testing, with no subscription.
Personas from your audience read your landing page or two versions of your copy. Each persona scores every version and explains why: where they got stuck, what was missing, what convinced them. An audience typically has 10 to 14 personas who differ clearly, even when they share the same role, and a run includes all of them, up to 24. Each persona also reads with the other personas’ jobs in mind.
Who the copy is really for and whether it lands with them. It shows whether your core audience is convinced or only the people at the edges, whether everyone likes it a little but nobody picks it first, whether your brand promise holds up, and which of your key claims your competitors make in the same way. The overview is at www.99takes.com/what-99takes-checks.
Yes. Give it the URL and 99takes checks the copy, the structure and reading flow, how the design comes across on desktop and mobile, and how well the page is put together. If you name competitors, or 99takes finds some, it compares your key claims with theirs. You end up with an action plan, most important items first.
Yes. 99takes writes challenger versions to compete with your current copy, drawing on the research into your audience or on the objections raised in a run. Any number that doesn't appear in the source material is flagged for you to verify. You see every version before it's tested. Three headline versions cost about 6 credits.
Yes, in the design check. The personas look at your page on desktop and mobile and tell you how it comes across. That verdict is separate from the copy score and isn't calibrated, so there's no hit rate for it. A check costs up to 12 credits, plus 2 to 3 credits for redesign suggestions.
No. The jury is curated and runs on Claude. What 99takes adds: audiences backed by evidence and saved as fixed versions, runs you can compare with each other, and a hit rate per jury based on reported market results.
You can prompt a round of personas yourself. What's hard to build on your own: audiences backed by public sources, runs that stay comparable with each other, a hit rate per jury based on reported market results, and a competitor check with verbatim quotes. The 99takes jury itself runs on Claude.
You report which version won in the market. Those reports add up to a hit rate per jury, always shown with the number of results and a range. It's published once there are ten conclusive results; until then it reads “not measured yet.”
No. A simulation doesn't measure the market. It shows you, before launch, where your copy loses your audience, so your A/B test starts from a well-founded hypothesis. If you have enough traffic for clean A/B tests, your own numbers are the better source.
A comparison of two versions runs right in your conversation with Claude, and that's where the result shows up. How long it takes depends on the scope and the number of personas. On request, 99takes turns it into a report with its own link that you can share. If we do the work for you, we agree on a schedule together.
Teams and agencies that want to test a landing page, ads or emails before they go live. In B2B that includes software, IT services, consulting and engineering, for example. It also works for considered consumer purchases and political campaigns. It's less suited to impulse buys that people decide on in a few seconds.
You can use 99takes yourself, starting with 350 starter credits, then with credits from €17.85, no subscription. If we handle the research, audience and report for you, a study costs a fixed €990 and a retainer from €1,500 a month. Pack prices include VAT, service prices are net. Details at www.99takes.com/pricing.
Yes. Your agency can use 99takes directly, with as many audiences as it has clients. With the service, we deliver the research, the audience and the report, and your agency puts the recommendations into practice.
Read on: how the buyer simulation works, how to connect 99takes to Claude, how the hit rate per jury is built and what the plans and packs cost.
Still have questions?
Try it with 350 starter credits, valid for 30 days, no payment details needed. Or let us do it for you.
Create an account