Five new leads: their site today, our build, and the postcard that introduces it
Updated after Carl's review pass: expert-plain headlines, standalone service copy, one CTA with FAQ and guidance below the form, unified value icons, two-colour discipline, real named attributions, and premium 2x device mockups on the cards. All five built to the lindeutemiljö flow, source-copy-first, we-form voice, variant pools, FAQ + schema, review gallery, knowledge below the form. Every postcard claim was re-verified against their live site before the PDF was rendered, and the print gate decoded every QR.
How the copy is layered: the four tiers, in plain words
Every sentence on a build comes from exactly one of four layers. The layer decides who wrote it, how often it repeats across sites, and what is allowed in it.
| Tier | What it is | Written | Repeats? | Rules | On these five |
|---|---|---|---|---|---|
| 1 · Business-specific | Their name, phone, towns, reviews, photos, their own site copy, verified facts | Harvested per lead, never invented | Never | Every fact carries a source and a date. Missing fact = omitted sentence, no fallback prose. | Acer's "thirty years to British Standards", BGG's testimonials naming Luke and Brian, Brunel's dropped-kerb council help, Digg's Pershore and Claines jobs. |
| 2 · Niche-shared | Guidance cards, FAQ sets, service vocabulary, advice articles | Once per trade | Across every site in the niche | True of the trade anywhere. Useful enough to change a decision, or it is cut. | The fencing FAQ on Shrivenham, the paving guidance on Brunel and BGG. |
| 3 · Region-shared | Local facts: conservation areas, geology, flood risk, local species | Once per geographic cluster | Within the cluster only | Checkable local facts. The cheapest specificity available: it makes a page that could only be from here. | Cotswolds cluster only so far. Hereford, Bath, Swindon and Worcester need their own clusters written, so these five carry Tier 2 in that slot. |
| 4 · Global-shared | Process, areas copy, enquiry copy, mid-CTA, why-choose values | Once, as pools of 3 to 5 variants | Everywhere, but deterministically deduplicated | Neighbours in the same sub-cluster never draw the same variant, so no mail-merge tell and no duplicate-content penalty. | All five drew different process and enquiry variants. |
Shrivenham Fencing — priority 64, the strongest lead of the wave
Verified today: nothing answers on port 443 on either host, so the site is plain http and browsers mark it "Not secure". No viewport meta, 102 words. Card claim: "your site has no padlock, so browsers mark it 'Not secure' before anyone reads a word."
Acer Tree Services — priority 64, best affordability evidence in the pipeline
Verified today: footer reads "© 2017… Proudly created with Mad Dog Web Design" in the rendered DOM. Card claim: "your site has not changed since it was built, and the footer admits it."
Digg's Gardens — priority 49
Verified today: below the fold the homepage embeds the Facebook feed, complete with like counts and comment threads. Card claim: "below the fold, your homepage is your Facebook feed." Note: no phone number on record, so the build carries no call buttons. The card itself needs only the postal address, which is verified, so it can post; a phone would strengthen the preview site, not the card.
Baths Green Gardens — priority 40
Verified today: the first testimonial sits 62% of the way down the document, under a page with visible typos ("Availble", "hugh choice"). Card claim: "the reviews that sell you sit low on the page, long after most visitors stop reading."
Brunel Driveways — priority 35
Verified 2026-08-07 (agent screenshot): service-card headings break mid-word at desktop width: "Paving and driveway/s", "Garden landscapi/ng". Card claim: "the headings on your homepage break in the middle of words."
The scores, and the likelihood they buy
Same two scorecards as the first test, scored by me (I built these, so read it as a structured self-assessment). "Buy likelihood" weighs the owner score against how big the jump from their current site is, whether the firm has money, and whether we can even reach them.
| Lead | Owner /25 | Homeowner /25 | Jump vs their site | Can pay? | Reachable? | Buy likelihood |
|---|---|---|---|---|---|---|
| Acer Tree Services | 22 | 22 | Huge. 2017 template to a modern build carrying their own credentials. | Strongest in the pipeline: accreditations cost real money. | Phone + address verified. | HIGH |
| Shrivenham Fencing | 21 | 21 | Huge. Five lines and no padlock to a full site. | 14-yr Ltd, VAT registered, accounts current. | Phone + address verified. | HIGH |
| Brunel Driveways | 22 | 21 | Moderate. Their site works; ours is visibly better made. | 7-yr Ltd, 31 reviews, wide service area. | Phone + address verified. | MEDIUM |
| Baths Green Gardens | 21 | 21 | Large. Same content, professionally presented. | 9 years trading; craft 8 site suggests nobody spending on web. | Phone + address verified. | MEDIUM |
| Digg's Gardens | 19 | 20 | Large. Feed to a real site. | 8-yr Ltd, work clearly flowing. | Postal address verified; phone still unknown (Apollo has nothing for them). | MEDIUM · card can post |