Website Redesign Lead Scorer

If you sell web design or web builds, finding businesses is the easy part. The part that eats the day is working out which 40 of the 300 on your list are worth an email, and what to put in the first line so it does not read like every other agency pitch.

I built this Actor because I kept doing that by hand. You open a prospect's site. The footer says 2018. The cookie banner turns out to be decorative, because Google Analytics fired before you touched it. You run axe and find sixty accessibility failures. Four minutes gone, on one company, and there are 299 left.

Feed the list in here instead. You get one row per site: a 0 to 100 redesign score, a HOT, WARM or COLD band, every reason with its numbers, and the three plainest reasons already written as sentences you can paste into the email.

What it does

Each website is loaded in a real Chromium, the way a visitor would load it. Four things get measured.

How old the site looks. The newest lastmod in the sitemap. The newest date in the page's structured data and in its <time> elements. The footer copyright year, the Last-Modified header, the <meta name="generator"> tag, the jQuery version. Dates the site publishes about its own content carry the weight here. A footer year is usually printed by a script and rolls over on its own every January, so it only gets used when there is nothing else to go on.

What loads before a visitor chooses anything. Every network request is recorded from the moment the page starts loading, and for three seconds after. Nothing is clicked. No banner is dismissed. Whatever fired in that window fired before a visitor had a say. The Actor then names the well-known tracking services it recognises. It counts the cookies already set, split first-party and third-party. It looks for a privacy policy link, and for a consent tool from a list of the usual vendors.

Accessibility. axe-core is injected into the rendered page and run against the WCAG 2.1 A and AA rules. You get the number of failing elements, the split by impact, and the five rules that failed on the most elements.

The basics. Mobile viewport tag. https, and whether the plain http address redirects to it. Plain-http files on a secure page. Page weight and request count. Contact form present or absent, plus the social networks linked.

Those four feed a documented rubric that produces the score. No personal data is collected anywhere: phone is a yes or no, social links are hostnames only, and no email address is ever read or stored.

The pitch angles

This is the field I actually built the Actor for. A score tells you who to email. It does not tell you what to say, and a generic opener gets deleted.

So the top three reasons come back rewritten as one plain sentence each, with the real numbers already in them. I keep them flat and unexcited on purpose, because the specific detail is what earns the reply and adjectives only get in its way. Some real ones from live runs:

• The footer still says 2019 and the newest page in the sitemap is from 2021.
• The newest page in the sitemap is from 2020, 6 years ago.
• Google Analytics, the Meta pixel and Hotjar load before a visitor makes any cookie choice.
• The homepage has 34 serious or critical accessibility problems, all of them from one rule: elements must meet minimum color contrast ratio thresholds.
• The pages carry no mobile viewport tag, so a phone shows the desktop layout shrunk to fit.
• The site still loads jQuery 1.11.3, and the current major version is 3.

Every number in one of those sentences is read back out of the row it describes. Nothing is generated, so nothing can be invented.

What you need

Nothing. No API key, no account anywhere, and no proxy for the great majority of business websites.

Run it with no websites at all and you get two sample rows in demo mode, charged nothing, so the shape is visible before you point it at a real list.

How to run it

A prospecting list, homepage only, all four checks:

{
  "websites": "smithandsonsplumbing.co.uk\nthe-old-mill-cafe.com\nhttps://example-garage.co.uk",
  "maxPages": 1
}

Deeper on a shortlist, following up to five internal pages per site:

{
  "websites": "smithandsonsplumbing.co.uk\nthe-old-mill-cafe.com",
  "maxPages": 5,
  "checks": ["freshness", "accessibility", "consent", "tech"]
}

A fast first pass with no browser. You keep the copyright year, generator, viewport and https. You lose the tracker list and the axe-core numbers:

{
  "websites": "a-very-long-list-pasted-here",
  "useBrowser": false
}

The OUTPUT record in the key-value store holds the summary: sites processed, the HOT, WARM and COLD counts, events charged, and the top 10 by score with their first pitch angle.

Output fields

One row per website, grouped the same way the checks are.

The result itself, and what happened during the visit.

Field	What it holds
input	your line, exactly as pasted
final_url, status	the final URL after redirects, and its HTTP status
browser_used, browser_error	false on a browser failure, plus the reason
pages_scanned, page_title	pages behind this row, and the homepage title
redesign_score, score_band	0 to 100, and HOT, WARM or COLD
reasons	every rubric line triggered, with its numbers and points
pitch_angles	the top three of those, one pasteable sentence each
checks_run	the checks switched on, for a fair comparison between rows
error	the reason for no score, or null
checked_at	when

These are the freshness signals.

Field	What it holds
newest_signal_year, newest_signal_source, stale_years	the newest dated evidence, its origin, and its age in years
sitemap_newest_lastmod, copyright_year	the two strongest individual signals
jsonld_newest_year, time_tag_newest_year, content_newest_year	dates from the page's own markup
last_modified	the header, reported but never scored

These cover consent and tracking.

Field	What it holds
consent_view_available	false without a browser, marking the rest of this group unknown rather than clean
cmp_detected	the consent tool found, or null
trackers_before_consent, trackers_before_consent_count	named tracking services from before any consent choice
cookies_before_consent, third_party_cookies_before_consent, first_party_cookies_before_consent	cookies already set at that point
has_privacy_link, has_cookie_policy_link	booleans

These come from axe-core and the cheap static counts.

Field	What it holds
a11y_ran	false on an axe failure, with every number below null rather than zero
a11y_violations_total, a11y_critical, a11y_serious, a11y_moderate, a11y_minor, a11y_serious_plus_critical	failing elements, not failing rules
a11y_rules_failed, a11y_top_rules	the rule count, and the five worst with their element counts
images_total, images_without_alt, missing_html_lang, inputs_without_label, empty_links_or_buttons, heading_level_skips	cheap static counts, present even without axe

These are the technical basics.

Field	What it holds
page_weight_kb, request_count	HTML, CSS and JS only, with images, media and fonts blocked
has_viewport_meta, uses_https, redirects_to_https, mixed_content_requests	the basics
generator, jquery_version	the page's own claim about what built it
has_contact_form, phone_found, social_links	a boolean, a boolean, and hostnames

The score

Six lines, adding up to 100.

Signal	Max	Points
Staleness	35	years since the newest dated evidence: 6 or more → 35, 4 to 5 → 26, 3 → 18, 2 → 10, newer → 0
Accessibility	25	4 per critical failing element, 1.5 per serious one, capped at 25
Consent and tracking	15	4 per tracking service loaded before a choice, up to 12; plus 3 for no consent tool; plus 3 for no privacy policy link; capped at 15
Mobile viewport	8	8 for no viewport meta tag
https	7	7 for no https at all; otherwise 3 for a plain http address that never redirects, or for plain-http files on the page
Old platform	10	6 for jQuery 1.x, 4 for jQuery 2.x, 4 for a generator tag naming old WordPress, Joomla 1 to 3, Drupal 7, FrontPage, Dreamweaver or MSHTML

HOT is 70 and above. WARM is 40 to 69. COLD is under 40.

A line only fires when it has something to say, and an unmeasurable signal never scores at all. A site whose accessibility could not be tested does not collect 0 accessibility points as though it had passed. It gets no accessibility line, and a11y_ran says why. Turning a check off in the input removes its lines too, so only compare scores across runs with the same checks.

Pricing

Pay per event: one site-scored charge per site that produced a score. A site that could not be fetched, that answered 403 or 404, or that you typed wrong costs nothing at all. The demo run costs nothing. Following internal pages with maxPages adds no charge, and neither does switching all four checks on. The price is per site, whatever you turn on.

Limits

• The homepage only, unless you raise maxPages. That is deliberate. For prospecting, the homepage is what the buyer will look at and what you will talk about.
• The consent findings describe what the page loaded on its own, before anyone chose anything, and they stop there. Whether any of it breaks a law is a question for the prospect and their lawyer. It also makes a far better opening line than an accusation.
• The tracking services named come from a list of well-known ones. A site can run something the list does not cover, so read the count as a floor.
• Some large publishers serve their consent tool from their own domain, so ad blockers cannot see it. That hides the vendor name from this Actor too. Where the shape is there but the vendor is not identifiable, cmp_detected says a first-party consent tool rather than guessing at one.
• The accessibility numbers are axe-core on the pages scanned, against WCAG 2.1 A and AA. Automated testing catches a portion of the real barriers. Treat the number as a reason to open a conversation, and never sell it as a WCAG audit.
• Blocking images shifts the accessibility count very slightly, because an image that never loads never reaches the DOM as one. I measured this on a live site: 7 critical failures with images on, 6 with them off, a single image-alt node of difference. Everything else matched to the node.
• A footer copyright year is usually generated, so it loses to dates the site publishes about its content. Last-Modified is a server file timestamp rather than evidence that anyone touched the design, so it appears in the row and stays out of the score.
• JS-heavy sites and sites behind a bot check can beat the browser. Those fall back to a plain HTTP read, browser_used comes back false, and the tracker and axe fields come back null rather than zero. Any site answering 400 or above gets no score at all, because treating a bot-block page as a homepage is how a lead list fills up with confident nonsense.
• Small business sites on shared hosting sometimes refuse a datacentre IP outright. One site in my own testing returned 200 from my laptop and 401 from an Apify server on the same afternoon. That is the case proxyConfiguration exists for, and the row says 401 rather than inventing a score.
• Page weight covers HTML, CSS and JS. Images, media and fonts are blocked to keep runs cheap, so the number sits below what a visitor really downloads.
• Only public http and https hostnames are accepted. IP addresses, local hostnames and everything else are refused before any request goes out.

Pair it with a workflow

If you run n8n, my free templates on https://n8n.io/creators/fractional-hq take a list like this Actor's output and turn it into a cold email draft per prospect, plus the follow-ups. Filter to score_band of HOT, drop pitch_angles[0] into the opening sentence, and the specific detail that makes the email land is already written for you.

The scoring model and its blind spots are written up at https://fractionalhq.uk/use-cases/redesign-lead-scoring

Built by Fractional HQ

I run Fractional HQ, a one-man studio for design and automation. I do not write code by hand: I architect, review and hold the standard while AI agents do the typing. That method runs my own estate: seven brands, a Postgres operations system, more than 100 scheduled jobs, published n8n templates, Zapier Silver certified.

Want this wired into your CRM and outreach, or built and handed over working? Get in touch: https://fractionalhq.uk/brief