The team knew what the page was supposed to say
The founders, marketer, and product agent shared months of context. They knew the category, customer problem, product mechanism, and roadmap.
That knowledge made the page easy for them to understand.
A new visitor would see only the headline, supporting copy, interface examples, proof, and CTA. The company needed to test the gap between intended meaning and received meaning.
Another internal review would not create that distance. An agent with access to the strategy could identify copy patterns, but it could not provide genuine first exposure.
The brief protects the cold start
The participant must not have seen the product, company, or page. They receive a public URL and a perspective, not the intended positioning.
Example perspective:
“You are responsible for product, customer success, support operations, or customer insight at a software company. Review this page as you would during an initial evaluation.”
Before the participant sees background material, Taskin asks:
- 1.What does the product do?
- 2.Who is it for?
- 3.Which problem does it solve?
- 4.What do you expect after clicking the main CTA?
- 5.Which claim do you trust most?
- 6.Which claim do you trust least?
- 7.What information is missing?
- 8.Which words or sections are unclear?
- 9.How is the product different from alternatives you know?
- 10.What is your strongest reason not to continue?
The participant records immediate interpretation before considered advice.
Execution
The marketing agent submits the public page, target perspective, clean-start rule, questions, and evidence format.
The participant confirms prior exposure, opens the page in a clean session, and records their first description before visiting other pages. They reference exact sections when explaining confusion or trust.
Verbatim response, observation, and recommendation remain separate. The participant is not instructed to be positive, validate the founder, or solve the whole positioning problem.
The result returns to the marketing workflow. The agent compares the participant's language with the intended proposition. A marketing owner decides what to change.
A second test uses a different participant. The first reviewer can no longer provide first exposure.
Structured result
prior_exposure: false
participant_context: <truthful relevant background>
first_product_description: <verbatim response>
perceived_audience: <participant response>
perceived_problem: <participant response>
primary_cta_expectation: <participant response>
most_trusted_claim: <specific claim and reason>
least_trusted_claim: <specific claim and reason>
missing_information: <list>
unclear_language: <list with page references>
perceived_alternatives: <named or none>
strongest_objection: <verbatim response>
recommendations: <separate optional section>Acceptance test
The task passes when:
- prior exposure is confirmed;
- every required question is answered;
- first impression precedes additional context;
- specific page elements support the feedback;
- verbatim reactions remain distinct from interpretation;
- the participant's background is represented truthfully;
- uncertainty is visible;
- no answer is rewritten to sound more favorable.
How the team should use one person's feedback
One participant can reveal a comprehension failure. One participant cannot estimate market-wide prevalence.
The company can fix an obvious gap immediately, test the issue with more people, or compare reactions across real segments. It should not call a single response “customer research” without qualification.
The most useful comparison is intended versus received:
| Intended | Received | Decision |
|---|---|---|
| Product definition | Participant's first description | Match, partial match, or mismatch |
| Target buyer | Perceived audience | Clear or mis-segmented |
| CTA promise | Expected next step | Aligned or surprising |
| Credibility | Trusted and untrusted claims | Keep, prove, narrow, or remove |
| Differentiation | Named alternatives | Distinct, familiar, or unclear |
What to measure
- definition match;
- target-buyer match;
- CTA expectation accuracy;
- trust gaps;
- unclear terms;
- missing decision information;
- accepted copy changes;
- improvement in the next cold-start round;
- later page conversion, measured separately.
No conversion lift is claimed here; production data is measured separately by the company.
When this pattern fits
Use it before paid campaigns, product launches, positioning changes, pricing-page changes, homepage redesigns, or investor/customer outreach.
It complements, but does not replace, customer interviews, analytics, experiments, or broader message testing.
What the company kept on record
The company retained the exact page version, the task reference, the participant starting condition, verbatim responses, the changes made, and the second-round result.
The record shows the before-and-after copy and which human observation led to the change. No testimonial is attributed to the customer.