Lockerfella: #1 on Google, ChatGPT and Gemini in 21 Days, With Zero Backlinks

On 8 April 2026 we launched lockerfella.co.uk for a one-man locksmith business in Brewood, Staffordshire. The domain was brand new. It had no backlinks, no reviews, no history of any kind. Nineteen days later, ChatGPT and Gemini both named Lockerfella first for "Brewood locksmith". On day 21, Google's organic results agreed. This page is the full account: what we built, the decisions behind it, the alternatives we rejected, and what the May 2026 core update did to it afterwards.

The Client: A One-Man Locksmith in Brewood

Sean Hamilton has spent 30 years on the tools. He runs Lockerfella as a one-man, one-van operation covering Wolverhampton, Cannock, Stafford, Walsall and the South Staffordshire villages around Brewood. His problem was the standard one for an independent tradesman: the search results for his area were owned by national directories and incumbent firms with two decades of reviews. A new website with no authority enters that market at the very bottom.

The brief was blunt: make the phone ring from Google, without the budget for a long backlink campaign or paid ads. That constraint shaped everything. If we couldn't buy authority, the site had to win on the two things Google's recent updates reward most: genuinely useful first-hand content and flawless technical delivery.

What We Built: Area Pages, Schema, AI Discovery Files

One disclosure first, because we hold our case studies to the same honesty as our editorial standards: Lockerfella is hand-coded PHP, not WordPress. We chose that to hit aggressive Core Web Vitals targets on a tight budget. Every other pattern on this page, the content model, the schema, the AI Discovery Files, applies directly to the WordPress builds we deliver every week.

The build, in the order of what mattered:

• 18 hand-written area pages. Each one carries two or three real, recent call-outs: the customer's first name, the verified street, what was wrong, what was fitted, and the actual price charged (between £30 and £249 across the set). No two pages share a paragraph. This is the opposite of the templated doorway pages most trade sites ship.
• Open pricing. No call-out fee, the phone quote is the price paid, payment after the job. Published on a dedicated pricing page while most competitors hide behind "call for quote".
• A real About page. Sean's name, his 30 years, his DBS check, his £1M public liability insurance, his 12-month workmanship guarantee, and photographs of the actual man and the actual van.
• Schema on every page. LocalBusiness, Service, FAQPage and BreadcrumbList entities throughout. Google's Rich Results Test returned 6 valid items and 0 errors at launch.
• All 9 AI Discovery Files at the domain root, following the AI Visibility specification that our founder authored. These give ChatGPT, Gemini and Claude a machine-readable account of who the business is and what it does.
• Performance as a budget, not an afterthought. The dated PageSpeed Insights run of 29 April 2026 shows mobile scores of 100/100/100/100 with First Contentful Paint at 1.0s, Largest Contentful Paint at 1.7s, Total Blocking Time at 0ms and Cumulative Layout Shift at 0. The compressed HTML weighs 34.8 KB.
• Hosting that stays out of the way. The site runs on 365i's standard Linux web hosting with the bundled CDN: server TTFB measured at 28ms, and an A+ security grade with all 8 security headers present in 365i's free HTTP Header Inspector. Notably, this is the standard £14.99-a-month tier, not a premium plan.
• A backlink magnet with a sense of humour. Added on 9 June 2026: Ask an Expert Locksmith, a daft little game. Ask the padlock anything starting with "Could Lockerfella" and it replies with one of those gloriously British rhetorical questions that always means yes ("Would a dad guard the good scissors with his life?"), with Sean's real, costed answers further down the page. You can't help sitting there trying questions, and every verdict comes with a shareable image card.

The Build Decisions, and What We Rejected

A case study that hides its rejected options isn't one, so here is what we turned down and why:

• Rejected: a page-builder WordPress build. For this budget and these speed targets, a typical page-builder stack would have shipped hundreds of kilobytes of CSS and render-blocking JavaScript, capping mobile PageSpeed in the 80s. Where a client needs WordPress (a CMS, a shop, a team editing content), we build WordPress and engineer the same disciplines into it. Lockerfella didn't need a CMS.
• Rejected: templated area pages. Fifty pages with the town name swapped in would have given Google a doorway-page pattern to demote and the AI engines nothing distinct to quote. We shipped 18 real ones instead of 50 fake ones.
• Rejected: stock photography. No gloved hands on a generic lock, no cyber-blue padlocks. Every image is Sean, his van, or his actual work.
• Rejected: stretching the trade-body badge. Sean is under two years trading and not yet eligible for Master Locksmiths Association membership, so the MLA badge is switched off in the site's config until he is. Most competitors would have stretched it. Trust signals you can't verify are liabilities now, not assets.

The Results: #1 on Google, ChatGPT and Gemini in 21 Days

The timeline, every step of it dated and screenshotted in the original build write-up:

1. 8 April 2026: launch. Zero backlinks, zero reviews.
2. 9 April: Sean posts a five-star Google review of the build, unprompted.
3. 27 April (day 19): fresh, logged-out tests on ChatGPT and Gemini both return Lockerfella as the first recommendation for "Brewood locksmith".
4. 29 April (day 21): Google's organic results follow. Lockerfella holds #1 on all three systems. The site still has zero backlinks.

Sean's own reaction, in his verbatim Google review of the launch: "Really impressed with Press Forge. Their website alone shows how professional and clean their work is. You can tell they've got loads of experience and actually understand what businesses need, not just how to make something look nice." Three weeks later he added a second review on our sister company's profile: "I know how important Google PageSpeed is, and this site hits 100 across the board, which I never thought I'd see."

What the May 2026 Core Update Did

Google's May 2026 core update ran from 21 May to 2 June. Core updates are where content-led rankings either hold or evaporate, so we treated it as the first real stress test and documented the results the day it completed.

Lockerfella came out stronger. The day the rollout finished, its uPVC diagnosis page held #1 organic for "door won't lock wolverhampton" and its lost-keys advice page held #1 for "lost house keys wolverhampton", both ahead of national directories with years of history. These are exactly the problem-led, first-hand pages Google says the update rewards.

"We recommend waiting at least a full week after a core update completes before analyzing your site in Search Console."

Google Search Central, core updates documentation

And the quieter win: on "locked out wolverhampton", a head term Lockerfella does not rank for organically, Google's AI Overview names Lockerfella as its first recommended locksmith, phone number included. The AI layer has read the content and trusts it ahead of the classic ranking signals. In our experience that's the leading indicator, with the organic position following as local signals build.

What It Hasn't Won Yet: The Wolverhampton Head Term

Here is the part most agency case studies leave out. On the bare commercial query "locked out wolverhampton", Lockerfella is not in the organic top set. That space belongs to incumbents and directories, one of which carries 5,921 Google reviews. Bare emergency head terms are decided less by any single page's quality and more by local prominence: proximity, the Business Profile, and sheer review volume feeding the Map pack. An eight-week-old business has not banked those signals, and no amount of clever content shortcuts it.

We say this for a simple reason: if an agency promises a new local business the head terms in week one, they are either guessing or lying. The honest sequence is the one Lockerfella followed: long-tail first, AI layer second, head terms as reviews accrue over months. The two #1 positions and the AI Overview citation are the proof the strategy works; the missing head term is the proof we won't pretend it's magic.

The next lever is already live. The site ranked its first #1s with zero backlinks, but head terms eventually want links too, so we built it something worth linking to: the Could Lockerfella page exists precisely to earn shares and backlinks that a plain service page never attracts. Reviews build prominence, linkable assets build links, and the head term follows both.

Applying This to a WordPress Build

Everything that ranked Lockerfella is a pattern, not a one-off. On the WordPress sites we design and build, the same playbook looks like this: non-commodity pages written from the client's real job records, conversion-focused design with the client's actual photography, schema on every page, the 9 AI Discovery Files at the root, and a performance budget enforced before launch rather than patched after. We applied the identical discipline to this very site, documented in our PageSpeed 100 case study.

One honest calibration: a hand-coded site makes the perfect 100s easier. A well-built WordPress site using these patterns typically lands in the high 90s, which is competitively equivalent; what matters is being decisively faster and more useful than the sites you need to outrank. If you run a trade business, the locksmith industry page shows how we apply this thinking to your sector.

Frequently Asked Questions