# Message test — https://asksensei.dev/ After reading your page, only 4 of 15 personas could name what kind of product this is, unprompted. - **Page tested:** https://asksensei.dev/ - **Audience tested against:** An independent B2B SaaS founder with a working product and a public website, preparing for a launch or customer outreach. They write their own copy and can make changes themselves or through an AI builder. They have limited time and budget, want actionable feedback, and already use ChatGPT or Claude. They would need a clear reason to pay for a separate review tool. - **Personas:** 15 simulated - **Report:** https://grader.wynter.com/r/sensei-what-strangers-think-you-sell-HMU4Nuo > These answers are generated by AI, scored on Wynter's B2B Message > Layers framework using behaviorally-diverse simulated personas. The > methodology is real and the critique is directional. What a simulated > persona cannot have is a live budget, a renewal coming up, or a boss > asking about this quarter. --- ## 01 · The scores Every persona answered all four questions. These are four independent proportions of the same panel, not stages of a funnel. | Layer | Question | Cleared the bar | Strength | Of those who passed | | --- | --- | --- | --- | --- | | 1. Clarity | Do they understand what you do? | 4/15 | 78% | all with reservations | | 2. Relevance | Can they tell what it solves, and who it's for? | 15/15 | 99% | 14 without hesitation, 1 with reservations | | 3. Value | Do they actually want it? | 8/15 | 52% | all with reservations | | 4. Differentiation | Is there a reason to pick you over the alternatives? | 9/15 | 56% | all with reservations | **Brand alignment** (a side metric, not one of the four layers) — 15/15, 78% strength (all with reservations). Does the page read like the company you actually are? **Fix first: Clarity.** Earliest failing layer, walking the sequence in order — not simply the lowest score. ### What they thought you sell 7 of the personas who named a category got it wrong: - 2× “AI-powered landing page / positioning audit tool” - 1× “AI product/positioning critique tool” - 1× “AI-based product/positioning audit tool” - 1× “AI-powered homepage/positioning audit tool” - 1× “AI-powered landing page / product audit tool” - 1× “AI-powered website/positioning audit tool” --- ## 02 · What to change, layer by layer Ordered worst-first. Specific edits, not a restatement of the score. ### Clarity **Replace the kanji axis glyphs and "Lock in. Strong, with one soft axis." with plain labels.** Terms like "soft axis", "composite · unedited" and the character marks make a first-time reader decode the results card before they can trust it. Use plain labels such as "weakest area: Presence" next to each axis. *effort medium · impact high · tested against Plain language* **Fix the hero score display so the composite reads as a normal number out of 100.** The hero shows "-177/100" and axis values like "Market -252", which read as a broken placeholder rather than a score. Show a score inside a stated range and label what high and low mean. *effort low · impact high · tested against Plain language* **Add one sentence under the hero headline explaining what Sensei does to a URL.** Nothing at the top says the reader pastes a URL and gets a scored critique plus one rewritten line; that only becomes clear at the worked example far down the page. State the input, the output and the time in one line before the sample read. *effort low · impact high · tested against Show the product early* ### Value **Add a named customer with role and company beside the getmeerkat.dev sample read.** Every example on the page is anonymous, so the sample read looks like a mockup built for the site. Attach a real name, role and company to at least one reading. *effort medium · impact high · tested against Proof next to the claim* **Add an outcome line under "Same product. Now a paragraph a stranger can hold onto."** The before/after rewrite proves the copy changed but not that anything commercial moved. Add what happened after the fix shipped, such as the score change and the conversion or signup difference over a stated period. *effort medium · impact high · tested against Tie the feature to the outcome* **Add a second worked example from a different business type below the first.** One example makes the result look like a lucky case rather than something repeatable. Show a second before/after, ideally from an independent selling a service, since the page claims to serve both audiences. *effort high · impact medium · tested against Proof next to the claim* ### Differentiation **Move the before/after rewrite above "Why it cuts" so it appears before the argument sections.** The rewritten headline is the thing that proves this is not another free grader, but it sits below the hero and the sample card. Put the concrete before/after directly under the hero. *effort medium · impact high · tested against Show the product early* **Source the "11,400 runs this week" figure or drop it from the after example.** A usage number with no date, source or definition reads as invented and taints the confident tone around it. State where the count comes from and what counts as a run. *effort low · impact medium · tested against Proof next to the claim* ### Brand alignment (side metric) **Rewrite "Why it cuts" and "Past the number" as headings that state their point.** Scanning the headings alone gives no story, because they name a mood rather than a claim. Lead each with what the section concludes, such as the re-read proving whether the fix worked. *effort low · impact medium · tested against Headings stand alone* --- ## 03 · What is working ### The homepage names its audience and problem without requiring inference Eight respondents said the hero copy states who the tool is for and what problem it solves on first screen, with role-specific vocabulary. Several called out that no guesswork was needed. > the hero line "For founders & independents" plus "You're too close to see what's broken. Sensei isn't" tells me the problem (blind spots in your own positioning) and the audience in the first two lines, no hunting required > > — Solo Founder, Software, 1-10 > the hero line "You're too close to see what's broken" plus "For founders & independents" right at the top tells me who this is for and what pain it's hitting > > — Founder, Technology Startups, 1-10 > the header says "For founders & independents" right up top, and the "Who Sensei reads" section spells it out explicitly with two named buckets (Founders building a product, Independents selling a service). The problem is stated almost immediately too: "You're too close to see what's broken" > > — Solo Founder, Software, 1-10 > It's clear fast — the hero line literally says "For founders & independents" and "You're too close to see what's broken," and further down it splits into two named reader types with their own language > > — Solo Founder, Software, 1-10 > the hero line "For founders & independents. You're too close to see what's broken. Sensei isn't" tells me the problem (blind spots in your own homepage/positioning) and names the two audiences right there, no digging required > > — Independent Founder, SaaS, 1-10 ### The before/after headline rewrite is the proof that lands Five respondents pointed to the concrete worked example as evidence the product produces real output rather than generic promises, and as the thing separating it from free alternatives. > The "re-reads that exact thing and proves the lift" mechanic and the actual before/after example (the "Stop prompt engineering" headline rewritten to "Turn 15-minute prompt battles into 2-minute wins") is the one concrete thing that would pull me toward this > > — Solo Founder, Software, 1-10 > The "952 readings delivered" line and the actual before/after rewrite example on getmeerkat.dev ("Stop prompt engineering" → "Turn 15-minute prompt battles into 2-minute wins") is the one thing that would tip me toward this over a competitor > > — Founder, Technology Startups, 1-10 --- ## 04 · What the personas said ### The score display reads as broken or unfinished Three respondents said the scoring format looks like a placeholder or is obscure, which directly damaged their confidence in the product. > the friction was the score display itself: "-177/100" reads as broken or a placeholder bug, and "Composite · unedited" doesn't explain what unedited means until much later > > — Solo Founder, Software, 1-10 > my hesitation was the score itself, '-177/100' and 'Composite · unedited,' which reads like a broken widget or an inside joke rather than a real metric > > — Solo Founder, Software, 1-10 ### There is no named customer anywhere on the page Six respondents said the absence of named customers, logos, or a second case makes the single example feel like a mockup and blocks credibility. > there's no named customer, no logo, nothing beyond one anonymized "getmeerkat.dev" sample — I'd want to see two or three real founders say "this caught something my team missed and it moved our conversion," > > — Founder, Technology Startups, 1-10 > What would rule it out is the "952 readings delivered" line sitting right next to zero named customers or logos of companies that actually shipped the move and grew > > — Solo Founder, Software, 1-10 > the "952 readings delivered" and the one example (getmeerkat.dev) are the only proof points, there's no named customer saying "this caught something my team missed," and I don't know if the rubric holds up across different products or is just plausible-sounding generic advice. I'd want to see 2-3 named users with before/after scores > > — Independent Founder, SaaS, 1-10 > there's no second real customer example, no named company saying "yes, this genuinely moved my conversion rate," so I can't tell if that progression is a real user or an illustrative mockup > > — Solo Founder, Software, 1-10 ### Copy examples are not accepted as evidence that scores move revenue Three respondents said they need conversion metrics or before/after score data tied to outcomes; improved copy scores alone do not prove commercial gain. > I'd want to see the "re-read scores the same axes against the last read" claim proven on someone else's site with before/after conversion numbers, not just copy diffs > > — Founder, Technology Startups, 1-10 > I'd just run the free read myself and see if the "one move" actually moves the number before I'd bother pulling in anyone else to discuss it. > > — Founder, Technology Startups, 1-10 ### The 952 readings figure is unsourced and reads as unverifiable Three respondents flagged the usage statistic as having no source or context, which undermines the credibility the confident tone is trying to build. > the "952 readings delivered" and "11,400 runs this week" numbers have zero source or context — no cohort, no before/after aggregate, nothing > > — Independent Founder, SaaS, 1-10 > unsourced usage numbers ("11,400 runs this week") that read like placeholder stats rather than something a two-person shop would actually have and choose to show. > > — Independent Founder, SaaS, 1-10 > No VC slop," "Sensei refuses to hand you five balanced options"), no corporate hedging, which is the register I actually respond to — though I'd note that same confidence is exactly why I want to see a second real customer example before I trust it's not just well-written copy > > — Solo Founder, Software, 1-10 ### The mechanism is only understood after reading the worked example Three respondents said the actual mechanism is unclear from the top of the page, and one read the core product as a standard website grader in mystic framing. > I'd call it an AI-powered landing page audit / conversion copy critique tool, not anything more exotic than that — the "Sensei" branding and mystic framing is trying to make a fairly standard category (website grader) feel like something new. > > — Founder, Technology Startups, 1-10 > It's an AI-generated critique tool for your homepage/landing page — you paste in your product URL or copy, and it scores you on four axes (Market Fit, Strategy, Presence, Edge), gives you a "one move" to fix the weakest spot, then re-scores you after you ship the change. > > — Solo Founder, Software, 1-10 > Where it got slightly murkier was the actual mechanism — I had to read into "Why it cuts" and the worked Meerkat example to understand it's specifically a landing-page/positioning critique with a scored rubric and a re-read loop > > — Founder, Technology Startups, 1-10 ### The audience label is generic where a product category is expected One respondent wanted the specific product category named rather than the broad founder label. One framed the value as a pre-consultant positioning check. > I'd want to see my actual category named, not just "founders & independents" — something like SaaS, dev tools, agency sites — because right now I have to self-select into "founder" without knowing if their rubric was built for consumer apps, B2B tools, or something else entirely > > — Solo Founder, Software, 1-10 > If it actually worked, it would save me a round of "does this even make sense to a stranger" before I put money behind a consultant or a fractional CMO > > — Founder, Technology Startups, 1-10 --- ## 05 · The hardest read An adversarial pass over the findings. Every claim below was checked against the panel's own answers; unsupported ones were dropped. - **The page has no credibility infrastructure, so every confident claim it makes lands as assertion.** *(high)* Six respondents flagged the total absence of named customers or logos, three called the 952 readings figure unsourced, and three refused copy examples as proof of revenue impact. Nothing on the page survives verification. - **The one asset that works is also the only asset carrying the page, making the whole argument single-threaded.** *(high)* Five respondents named the before/after rewrite as the proof that lands, but six said a single example without a second case reads as a mockup. The strongest element is structurally fragile. - **The page defers comprehension: the hero sells but does not explain, forcing readers deep before they understand what they are buying.** *(high)* Three respondents only grasped the mechanism after the worked example, and one read the product as a standard website grader in mystic framing. Clear audience targeting up top does not compensate for an unexplained mechanism. - **The page proves capability but never proves consequence, leaving buyers no reason to pay.** *(high)* Three respondents explicitly demanded conversion metrics or before/after score data tied to outcomes, stating improved copy scores alone do not demonstrate commercial gain. The worked example shows output, not return. - **The product's core output is its weakest visual element, undermining the demo it depends on.** *(medium)* Three respondents said the score display reads as a placeholder or is obscure, damaging confidence directly. Since the worked example is the page's only persuasive proof, a broken-looking score corrupts it. - **Relevance is doing all the work and it is not enough to close anyone.** *(medium)* Eight respondents credited the hero with naming audience and problem without inference, yet the page then fails on evidence, sourcing, and outcome proof across three separate negative themes. Attention is earned and then spent on nothing. --- ## 06 · Who answered | # | Role | Industry | Company size | | --- | --- | --- | --- | | 1 | Solo Founder | Software | 1-10 | | 2 | Founder | Technology Startups | 1-10 | | 3 | Independent Founder | SaaS | 1-10 | | 4 | Solo Founder | Software | 1-10 | | 5 | Founder | Technology Startups | 1-10 | | 6 | Independent Founder | SaaS | 1-10 | | 7 | Solo Founder | Software | 1-10 | | 8 | Founder | Technology Startups | 1-10 | | 9 | Independent Founder | SaaS | 1-10 | | 10 | Solo Founder | Software | 1-10 | | 11 | Founder | Technology Startups | 1-10 | | 12 | Independent Founder | SaaS | 1-10 | | 13 | Solo Founder | Software | 1-10 | | 14 | Founder | Technology Startups | 1-10 | | 15 | Independent Founder | SaaS | 1-10 | --- ## 07 · Before you act on this The methodology is real, and the critique is directional. What a simulated persona cannot have is a live budget, a renewal coming up, or a boss asking about this quarter. **Validate anything you're betting on with real ICPs who are actually in-market.** Being wrong is more expensive than you think. Finding out is cheaper than you'd guess. Wynter runs message testing with verified B2B professionals — trusted by HubSpot, RingCentral, Shopify, Cognism, Paddle, Veeam, Rippling and Miro. This report is kept for 60 days from 2026-09-26, then deleted along with the personas and their answers.