Top of Search logo
Framework · Working plan

The 90-day local visibility plan.

The sequence we run in client engagements, laid out for an owner to work through alone. Three phases, one focus each, roughly two hours a week.

Ninety days is the shortest window in which AI visibility work produces measurable movement. Engines need time to crawl, retrieve, and trust changes, and citation behavior typically starts shifting 60 to 90 days after the work lands. This plan sequences the work so the slow-compounding moves start first and the before-and-after comparison actually holds up.

The order is the point. Every audit we run finds the same pattern: businesses publish content on a foundation that cannot support it, then wonder why nothing moves. Identity first, structure second, content third, corroboration alongside, measurement throughout.

You cannot rush trust, but you can stop delaying it. Everything in this plan is either free or already yours to fix.
Top of Search
Before day one

Set the baseline you’ll measure against.

One sitting, about an hour. Build your prompt set: the questions a ready-to-buy customer asks at each stage, discovery through branded. Run each one in a private window across ChatGPT, Perplexity, and Google, where the AI answer is now the default experience. Screenshot every answer, note who was named and which sources were cited, and file it with the date. This is the before picture, and in 90 days it is the only evidence that matters.

Days 1–30

Foundation: one business, verifiable everywhere

The unglamorous month. Nothing here is visible to customers, and everything later depends on it. An engine has to be able to verify you to cite you.

01

Unify your business identity

Run the Entity Consistency Check, then fix what it finds: one canonical name, one phone, one address, one description, made identical across your website, Google Business Profile, and every directory and profile that mentions you. In our audits this is the most common critical finding, and it is the root cause behind half the other gaps.
Week 1
02

Rebuild your Google Business Profile field by field

Exact primary category, every service itemized, attributes set, hours current, photos from real jobs this year. Then restart a steady posting cadence: recent work, seasonal reminders, offers. Engines use the profile to verify you, and an abandoned profile reads as an unverifiable business.
Week 2
03

Add schema markup

LocalBusiness and Service schema carrying the same canonical name, phone, and address as everything else, plus FAQ schema where you genuinely answer questions. Use the free Schema Builder, have your developer install it, and validate. In our audits this is routinely the highest-leverage single fix.
Week 3
04

Fix crawlability and page structure

Confirm AI crawlers are not blocked in robots.txt, then restructure your core pages: one question per section, the answer in the first sentence under a real heading. Not a redesign, a tightening pass on the pages you already have.
Week 4
Days 31–60

Content: publish the answers, starting with price

Now the foundation can support weight. This month you publish the pages engines lift passages from, in the order buyers ask.

05

Publish your pricing page

Real prices or ranges, what moves them, and what is included. The price question is the one customers are most likely to ask AI and least likely to ask you, and published pricing pages are among the most-cited business-owned URLs in our tracking. If you publish one page this quarter, publish this one.
Week 5
06

Deepen your two money service pages

Pick the two services that pay for everything else. Each page gets specifics an engine can quote: process, timeline, service area by name, what it costs, what can go wrong. Thin area pages that all say the same thing get consolidated or made genuinely local.
Week 6
07

Write the cornerstone guide

The one thorough guide your market lacks, shaped by your prompt set: the whole buying decision on one page, with named authorship and a visible date. Topic guides are the most-cited content type in market after market we track. A rigorous ranked list or comparison, one that includes options that are not you, is the strong alternative if your category already has a good guide.
Weeks 7–8
08

Add the FAQ layer

The questions buyers ask right before they buy, answered in one to three sentences each, marked up with FAQ schema. Pull them straight from your prompt set stages, especially trust and urgency.
Week 8
Days 61–90

Corroboration and the re-run

Engines trust what independent sources confirm. This month builds the agreement around your claims, then measures what moved.

09

Complete the directories your category leans on

Find the two or three directories AI actually cites for your category, Angi and HomeAdvisor in the trades, Zocdoc and Healthgrades in health, Zillow and Realtor.com in real estate, and complete your profile like a page an engine reads, because it is. Same canonical identity as everywhere else.
Week 9
10

Rebuild the review pattern

Ask every completed job, and steer the ask toward specifics: the service, the town, the timeline. “Replaced our water heater same-day in Leland” corroborates three claims at once. Respond to every review, because responses are content on a page engines read.
Weeks 9–12, ongoing
11

Put one proof asset on a platform you don’t own

One answer-shaped YouTube video (how you price, what a job looks like), or genuine participation where your trade is discussed. YouTube is now the most-cited domain in Google AI Overviews, and community threads carry weight on every engine we track.
Weeks 10–11
12

Re-run the baseline and read the movement

Same prompts, same private window, same engines. Compare against day one: appearance rate across the set, not any single answer. Read who got cited where you did not, because that should be next quarter’s list. Expect movement, not domination, and expect Perplexity to register the work first.
Week 13
What to expect

The arithmetic.

With existing content and authority, citation behavior usually starts shifting inside this window. Building from very little, the 90 days lay the foundation and the visibility shifts start to land in the following quarter. Answers vary run to run on every engine, so judge frequency across the fixed prompt set, and hold the set steady so the comparison holds up.

If day 90 shows no movement anywhere, the plan did not fail silently: the re-run tells you exactly which layer is stuck. That is also where the formal audit can help, because the fix is usually structural and specific to you.

Cited

The moves that work, every Thursday. Free.

Subscribe

Cited, the weekly newsletter on AI visibility for local business owners.

Common questions

Before you start.

How much time does this actually take?

Plan on roughly two hours a week, with weeks 7 and 8 heavier because of the cornerstone guide. The two exceptions worth delegating are schema installation, which needs whoever maintains your website, and the guide itself if writing is not your strength.

Can I skip the foundation month and start with content?

You can, and it is the most common mistake we see in audits. Content published on an inconsistent identity and an unstructured site earns a fraction of its potential citations, because the engine cannot verify who wrote it. The first month is what makes the second month work.

Do I need to buy any tools?

No. Every tool the plan references, the Prompt Set Builder, the Entity Consistency Check, the Schema Builder, and the Citation Check, is free on this site, and the measurement is manual prompt runs in a private window. Paid tracking tools automate the measurement, and they are worth it for an agency or an ongoing engagement, but the plan does not require one.

What if I already rank well in regular Google search?

Then you will likely see movement faster, especially in Google AI Overviews, which correlate with existing rankings. Your gap is usually that pages written to rank are not written to be quoted. The structure and content months matter most for you.

What order do the engines respond in?

In our tracking, Perplexity usually registers new pages first because it retrieves live on nearly every question. Google AI Overviews follow where you have index standing. ChatGPT tends to move last on unbranded questions. Judge the whole set at day 90, not any single engine.

What happens after day 90?

The plan becomes a cycle: re-run the baseline quarterly, keep the review and posting cadences running, and add one substantial content piece per month against the next prompt-set gap. If you would rather have the layer owned for you, that starts with the AI Visibility Audit, since every partnership begins from one, and your day-90 re-run will make it a well-informed conversation.

Anna Curry
Anna Curry

Anna Curry is the founder of Top of Search, a marketing consultancy in Wilmington, NC that helps owner-operated local businesses get cited by AI search engines. She has worked in marketing since 2008 and writes Cited, a weekly newsletter on AI visibility for local business owners.

Want the 90 days scored before you start?

The audit runs your business against real customer prompts across five engines and hands you this plan already prioritized for your gaps, with the baseline built.