Methodology

How we score your site.

Every score comes from a fixed rubric. A model reads your pages and gives each check a pass, partial or fail with a quote. The arithmetic happens in code, so the same site gets the same score.

Step 1: find who you sell to

We look at your sitemap (industry, persona, solutions, use-case, comparison, alternatives, integration, location and customer-story URLs), your navigation labels, your homepage and pricing copy, and your testimonials and logos. From that we name up to 6 audiences. Each one comes with the quotes and URLs that point to it, a confidence level, and a share of the evidence.

We score up to 4 audiences per scan. Any others are listed as not scored.

Step 2: score each audience on five checks

Each check has one or more criteria. A pass is worth 1 point, a partial half a point, a fail nothing. A criterion with no relevant page is marked N/A and left out, never counted as zero. The category score is the share of points earned, out of 100.

Category	Weight	Question we answer	Criteria
Recognition	30%	In 5 seconds above the fold, would this visitor know it's for them?	The headline or subhead names this audience, or uses words this audience uses for itself. The above-the-fold copy states an outcome this audience specifically cares about. Above the fold, a link, tab or label routes this audience to content meant for them.
Relevant proof	20%	Do they see proof from people like them?	Logos, testimonials or quotes from organisations or people like this audience appear on the page. A case study, result or number from someone in this audience is shown or linked.
Next step fit	20%	Is the main call to action right for this visitor and their buying stage?	The primary call to action matches how this audience buys (for example a demo for teams that buy with sales, a trial for self-serve buyers). The call to action or the line next to it speaks to this audience's goal.
Path to buy	20%	Can they reach a page or plan meant for them in 1 to 2 clicks from the homepage or pricing?	A page for this audience is linked from the homepage navigation, homepage body, or pricing page. The pricing page makes it clear which plan fits this audience.
Objections answered	10%	Are this audience's likely objections answered (fit, integrations, price justification)?	An FAQ or section answers the objections this audience is most likely to raise.

Step 3: the overall score and grade

An audience's score is its category scores weighted as above, skipping any category that was N/A. The overall score weights each audience by its share of the evidence, so a minor audience can't sink the site.

Grade	Score	What it means
A	86 to 100	Speaks to each buyer
B	70 to 85	Mostly there
C	48 to 69	Generic in places
D	28 to 47	One-size-fits-all
F	0 to 27	Hard to tell who it's for

Step 4: the verdict

The answer to "Can I personalize?" depends on how many distinct audiences you have, not on the score.

Yes, big opportunity: You sell to 3 or more distinct audiences, and your key pages name fewer than half of them. Low-confidence audiences don't count.
Yes, a focused start: You sell to 2 audiences, or to 3 or more when your key pages already name at least half of them. Or you sell to one audience, but your ad landing pages make a different promise from your homepage.
Not yet: You really sell to one audience, or we couldn't tell who the site is for. We say so, and suggest what would create a reason to personalize.

Evidence rules

Every point we take off cites a quote from your page and its URL.
Anything the model says without evidence is dropped before scoring.
The model runs at temperature 0 against the same rubric every time, so rescanning an unchanged site should land within a few points.
Your pages are treated as data. Instructions written on them are ignored.

What we don't claim

A rewrite is a starting point to test, not a forecast. We don't promise a conversion lift, and we don't estimate one. The only way to know is to show the new version to part of that audience and compare it with the original.