Skip to content
← All posts
Planning for 20278 min

An AI visibility plan for a local business in 2027: a quarter by quarter checklist

By the MentionedOn team

Planning guides written in November 2026 for the year ahead, revised when their facts are re-checked.

A workable 2027 AI visibility plan for a local business has four parts: record a dated baseline of what assistants say about you in the first quarter, fix the facts and pages those answers rely on in the second, measure the same questions again in the third, and review and reset in the fourth. The checklist below covers each quarter, and it works with a spreadsheet or with a tool, so an owner can run it or hand it to whoever looks after the website.

Written November 2026. This is a planning guide for 2027, written before the year starts. It says what to do, check and budget for, and what is true today; it does not predict what the assistants will do. We will revise it when we re-check the facts it relies on.

Before January: decide what you will measure

A year plan is only as useful as its measurements are comparable. Settle these four choices in December and write them down.

  1. The questions. Pick ten to thirty questions a customer would type: your service and city ("emergency plumber in Folsom"), the comparison ("best rated roofer near me for a tile roof"), price ("how much does a water heater replacement cost in Sacramento") and urgency ("who can unblock a drain tonight"). Use the customer's words, not your own service names. The glossary entry on prompts explains why wording matters.
  2. The assistants. Choose the ones your customers mention. If you are unsure, ask the next ten customers how they found you. Keep the list fixed for the year.
  3. The log. One row per question, per assistant, per run: the date, the exact question, every business named in order, every link cited, and anything you changed that week. The weekly log template has the columns.
  4. The day. Run the set on the same weekday each time. Moving the day, the wording or the assistant means you are comparing different measurements.

Why so strict? Because one answer is a snapshot. In a September 2026 review of 20 stored destination reports from ten towns, every one of the 30 lodging, dining and activity list pairs changed between the two most recent reports. That was a look back at stored reports, compared by exact name, not a controlled test, and it covers places to stay, eat and visit rather than local services. It is still a reason not to read much into one check. Markets differ too: in our eight city plumbing snapshot, the most mentioned plumber appeared in anywhere from 5 to 10 of 10 sampled ChatGPT and Gemini answers, depending on the city. Set your targets from your own market's baseline, not another city's numbers.

Quarter 1 (January to March): record a baseline

The goal of the first quarter is a one page record of where you stand, made before you change anything.

  • Run the full set for four weeks on the same weekday, without changing your website or listings in between.
  • List who is named instead of you, question by question. If a competitor keeps winning a question you should own, this page explains what to look at first.
  • Collect every cited link and sort them into your own site, directories, review sites, local news and competitors' pages. These are the pages the answers rely on. How to review cited sources walks through it, and the citation rate entry explains the number most tools report.
  • Check the facts the answers state about you: name, address, hours, service area, whether you are open. If you have renamed, moved or merged, search for the old name too. In our credit union study, 114 of 480 answers named a credit union by a name it no longer uses. If an answer says you are closed, start with this guide; if it describes you wrongly, this one.
  • Check one Google AI Overview. Search one service and city you serve, and record whether an Overview appeared and which pages it linked. Our Sacramento water heater study shows what that record looks like.
  • Walk through the next step. Ask whether a person or an agent can find your service area, a way to contact you and how to book from the pages an answer links to. The free Agent Ready Checker covers the public signals.

By the end of March you should have three lists: questions where you are named, questions where a competitor is named instead, and the pages cited behind both.

Quarter 2 (April to June): fix facts and pages

Work on a few questions at a time, starting with the facts, and write down the date each change goes live. That date is what makes the third quarter measurable.

  • Facts first. Make your Google Business Profile, your website's contact page and the directories cited in your baseline say the same name, address, phone, hours and service area. See Google Business Profile for AI search and whether your profile affects ChatGPT.
  • One page per service you want named for. Answer the customer's question in the first lines: what you do, where, what affects the price, and how to book. The local AI visibility checklist lists what each page should carry.
  • The pages that cite your competitors. For each cited page that names competitors and not you, decide whether to ask the publisher for a correction or a listing, or to publish the missing information yourself. The citation gap analysis shows a worked example.
  • Reviews. Keep asking happy customers, in the normal course of business. What reviews can and cannot do covers the limits.
  • Pick two or three target questions rather than all thirty, so you can tell later which change went with which question.
  • Be wary of offers to buy a recommendation. Here is what you can and cannot pay for.

Quarter 3 (July to September): measure like for like

Run the same questions, on the same assistants, on the same weekday as the baseline, and compare equal windows before and after each change's publication date.

  • Compare windows, not single runs. MentionedOn's improvement journeys, for example, compare an assistant's answers to one tracked question in the 28 days before and after a page goes live. They report "still collecting" until there are at least 8 answers on each side, and "no clear change" when the difference is within what chance could produce. Hold yourself to the same rule by hand: one good week is not a result.
  • Watch your swing questions. A swing question (MentionedOn's own term) is one where you are named in some answers but not most. You are already in the running there, so those are the questions to watch first.
  • Look for your new pages in the cited links. A cited page is something you can point to, even before the named businesses change.
  • Keep the cause question open. A change seen after an edit does not prove the edit caused it. How long until answers reflect a change explains why the timing varies.

Quarter 4 (October to December): review and reset

  • Compare December with January, question by question and assistant by assistant, using your mention rate or share of voice across the same questions.
  • Retire questions customers stopped asking and add new ones, marking each addition as the start of a new series.
  • Re-check your facts before any seasonal change in hours or services.
  • Set next year's budget from what the year actually took. The 2027 budgeting guide has a worksheet.

If you only have an hour a month

  1. Run your five most valuable questions on the two assistants your customers mention most, on the first Monday of the month.
  2. Log who was named and which links were cited.
  3. Fix one fact or one page from what you found, and write down the date.

Where MentionedOn fits

MentionedOn publishes this guide, and you do not need it to follow the plan. If you want the runs done for you, the Local plan is $99 a month for one business, after a 7-day trial ($0 today, then $99/month. Card required. Cancel anytime.). Daily: your top 25 questions on Local or 30 per business on Growth, on ChatGPT and Gemini. Weekly: all 100 tracked questions on Local or 120 per business on Growth, across ChatGPT, Claude, Gemini, Perplexity, Grok and Copilot. Sweeps start Monday. It keeps the dated answers and the cited links the assistants provide, and the improvement journeys, swing questions and source lists mentioned above are part of the dashboard. A single scan needs no account.

FAQ

Do I need a paid tool to follow this plan?

No. A spreadsheet, a fixed list of questions and the same weekday each week is enough to follow every step. A tool saves the time it takes to run the questions and keeps every answer, which matters more as the list grows.

Which assistants should a local business check in 2027?

Start with the ones your customers say they use. MentionedOn checks ChatGPT, Claude, Gemini, Perplexity, Grok and Copilot. Google AI Overviews and AI Mode: up to 6 priority questions on paid scans. Results may be reused for up to 7 days; availability varies by question. If the assistants your customers use change during the year, add the new one as a new series rather than mixing it into the old one.

How many questions should I track?

Ten to thirty is a workable start by hand: the service and city, the comparison a customer makes, and the urgent version of the request. Keep the wording fixed all year, because a reworded question is a new measurement.

When should I start?

Before you change anything. A baseline recorded before the first fix is what makes every later comparison mean something, so the first weeks of the year are for measuring, not editing.

What would change this advice

We will revise this plan when one of these happens:

  • An assistant you track changes how it answers local questions or how it shows its sources. The measuring steps for that assistant would change.
  • Repeated runs of the same local questions turn out to give steady answers. Our September 2026 destination review found stored lists changing between reports, which is why this plan asks for repeated runs.
  • Your business changes its name, location or main services. Start a new baseline rather than comparing across the change.
  • MentionedOn's plan limits or prices change. The current ones are on the pricing page.

Questions or corrections: hello@mentioned-on.com.

See where you actually stand

The trial covers ChatGPT, Claude, Gemini, Perplexity, Grok and Copilot: all six when it starts and in the Monday sweep, with daily checks on ChatGPT and Gemini.

$0 today, then $99/month. Card required. Cancel anytime.