Frequently Asked Questions

You need two things:

• An account at run.99takes.com. It starts with 350 credits, valid for 30 days, and needs no payment details.
• Claude: add 99takes as a connector in claude.ai or run one command in Claude Code. The steps are at /en/setup.

Not required:

• No code snippet on your page
• No tracking tool

Then name your landing page or paste two versions of your copy in the conversation, and describe your audience briefly (e.g., "IT managers in mid-sized companies" or "parents comparing tutoring services").

Yes. We started in B2B. The method works wherever people weigh an offer against concrete criteria before they decide, and that includes consumers and voters.

✅ Perfect for:

• B2B software and SaaS
• Consulting services and agencies
• Technical B2B products (e.g., IT security, HR tools, analytics)
• Services requiring explanation (e.g., financial consulting, insurance)
• Considered consumer purchases (e.g., education, real estate, subscriptions)
• Political campaigns and public information

❌ Less suitable for:

• Impulse purchases decided in a few seconds
• Pages nobody reads because the price alone decides

Rule of thumb: If your audience compares alternatives and has concrete questions ("Do I need SSO?", "Is this course funded?", "What does it cost me per month?"), 99takes works well.

You can prompt a persona round yourself. 99takes adds what a prompt alone does not give you.

→ More about the technology

99takes names the price before every run.

When your credits run out, no new run starts until you top up or the next month begins. Nothing is charged afterwards.

Plans and credit packs: Pricing.

Comparing two versions takes a few minutes. The result appears right in your conversation with Claude.

On request 99takes turns a study into a report with its own link you can pass on.

If we do the work for you, we agree the schedule with you.

The jury judges, it does not write.

You or Claude write the versions in the same conversation, and 99takes checks them. You get:

• A grade per persona for each version, with reasons
• The objections: where a persona gets stuck and what it is missing

For a whole landing page there is a section-by-section build that ends with a check against the old version.

Yes. Runs are billed in credits, not per page, so you can check as many pages as your credits cover. Comparing two versions with 12 personas, each also reading under the others’ jobs, takes about 73 credits; 1:1, each persona only under its own job, about 19.

Plans, credit packs and what a credit is: Pricing.

Which audience the copy is for and whether it lands there: whether the people at your core are convinced or only those at the edge, whether everyone likes it a little and nobody picks it first, whether your brand promise holds, and which core claims competitors make the same way. The overview is at www.99takes.com/en/what-99takes-checks.

No. The jury is curated and runs on Claude. What makes 99takes are audiences with evidence and recorded versions, runs that are comparable with each other, and a hit rate per jury from reported market results.

You report which version won in the market. That builds a hit rate per jury, always with the number of results and a range. It is published from ten decidable results on; before that it reads "not measured yet".

No – it complements them.

99takes is not an A/B testing replacement, but a precursor:

Bottom line: A/B tests tell you what wins. 99takes tells you why it's not working.