AI PLAYBOOK / FENCING

Key takeaways

THE MECHANISM

When a homeowner types “who should I hire to replace a wood fence in Atlanta,” the assistant doesn’t consult a secret ranking. It searches, its own index, the live web, or both, pulls the sources it trusts about fence companies in that area, and composes an answer from what those sources agree on. The engines differ in the mix (Google AI leans on its search and Maps data; Perplexity shows its citations; ChatGPT blends browsing with what it already knows), but the shape is the same: retrieval, then verification, then a recommendation it can defend.

That last part is the lever. An engine naming a fence company is taking a small reputational risk. It prefers companies whose story checks out everywhere it looks: the profile matches the site, the site matches the reviews, the reviews describe actual fence work, and other sources corroborate all of it. Contradictions are how you get skipped, because the engine cannot safely recommend you, since you’re unsafe to recommend.

THE SIGNALS

One identity, everywhere. Same name, address, phone, and service list on your site, Business Profile, and every listing. Every conflict is a reason to skip you.

Reviews with specifics. “Great job” proves little; “replaced 180 feet of privacy fence in a week, cleaned up perfectly” is quotable evidence. Volume and recency matter; detail is what engines can actually cite.

Pages that answer fence questions plainly. Wood vs vinyl, cost factors, HOA rules, timelines, written so a machine can lift the answer and name the source. This is why the SEO layer and the AI layer are the same work, not competing budgets.

Crawler access. If your site blocks AI crawlers, you’ve opted out of the answer. The robots.txt decision should be a decision, not an accident.

Corroboration. Chamber listings, supplier mentions, local press, directories, third-party confirmation that a real, established fence company stands behind the claims.

THE MEASUREMENT

The honest metric is Share of Answer: take 12 real fence-buying questions for your market, ask them on Google AI Overviews, ChatGPT, and Perplexity, repeat each three times in clean sessions, and count the questions where your company is named in at least two of three runs. That’s your X/12, per platform, dated, with screenshots.

Run it monthly and the trend tells the truth: signals building, or not. Single runs prove nothing, AI answers vary, which is exactly why anyone showing you one triumphant screenshot is selling, not measuring.

Beware two things in this space: percentage “AI visibility scores” with no method behind them, and guarantees of inclusion. Nobody controls the outputs. The work is making your company easy to verify, cite, and recommend, here’s how we run that as a service, measured X/12 with the method public.

THIS QUARTER

In order of leverage: fix every name/address/phone conflict you can find; make the review ask systematic (every job, same-day, compliant, and nudge customers toward specifics by asking about the project, never about the sentiment); publish honest answers to your five most-asked fence questions; check your robots.txt; then claim the two or three local listings that actually matter in your market.

Then baseline your X/12 before the work, so three months from now you’re reading evidence instead of vibes. The Free Visibility Check is exactly that baseline, 12 questions, your score, screenshots.