This page is about the operating model: what it costs, who does it, what happens each month and when to stop. The tactics are a separate question and they are covered in AI search for small business. Most small business programmes do not fail on tactics. They fail because nobody owned them and nothing was diarised.
Why the operating model is the harder half
The work itself is not complicated. Fix your entity, finish your listings, write a good page a month, ask for better reviews, check your prompt grid. Any competent person can execute it.
What kills it is that none of it has a deadline. Nothing breaks if you skip a month. No client chases you, no invoice goes unpaid, no system fails. It is the most postponable work in a small business, and it happens to be work that only pays off if you do not postpone it.
So the useful part of this page is not what to do. It is how to make it survive contact with a busy year.
The budget question, answered properly
The wrong question is “what should I spend a month”. The right one is “what can I still be spending in month 12”.
The corroboration layer accumulates. Directory consistency, review depth, third-party mentions and published answers all compound slowly and stop compounding the moment you stop. A programme running at a modest level for 12 months beats one running hard for four and stopping, and it is not close.
Three honest tiers.
Time only. Roughly 12 to 16 hours a month of your own time, no cash beyond hosting. Everything in this cluster is doable at this level. It is slower and it is a real version of the work.
Time plus selective help. Your time on the foundation and the relationships, paid help for the writing or the technical work. Common and sensible. The trap is paying somebody for the parts you could do and keeping the parts you will not do.
Delivered. Somebody else runs it. MarGen publishes its figures rather than quoting case by case, which is worth knowing when you collect comparisons: only two of the seven UK agencies we track publish anything at all. Before you gather quotes, read what a fair monthly price looks like and red flags when hiring.
Whichever tier, decide the 12-month number first and divide. Not the other way round.
Who owns it
One person. Named. With time in the diary.
That sounds obvious and it is the single most common point of failure. The usual pattern is that AI search belongs to “marketing”, which in a small business means the owner, the office manager and whoever last built the website, none of whom think it is theirs.
What the owner needs:
- Two to four hours a week, diarised. Booked like a client appointment. Moved rather than cancelled
- Authority to publish without a three-person approval chain. Approval friction is what turns one page a month into four a year
- Access to the accounts. Google Business Profile, the website, the directories, Companies House filings. Hunting for logins consumes a surprising amount of the available time
- A place to record things. One spreadsheet, one document. The prompt grid, the listing audit and the enquiry source log
They do not need to be a specialist. The knowledge in this cluster is a few hours of reading. The scarce resource is protected time and clear ownership.
The first month, which is different
Front-loaded on purpose. Roughly 20 to 30 hours, and the only month that needs more than a couple of days.
| Task | Hours |
|---|---|
| Entity audit: every public source found and recorded | 6 to 8 |
| Corrections: registers, directories, social profile fields | 4 to 6 |
| Google Business Profile completed properly | 3 |
| Schema and robots.txt checked, crawler access confirmed | 2 |
| Baseline: 20 prompts across 5 assistants, answers recorded | 2 |
| Choosing the 15 questions to answer over the year | 3 |
| Setting up the enquiry source field and the log | 1 |
Do it in two blocks rather than spread across four weeks. It is coherent work and it goes faster in one sitting. The audit method is how to audit your off-site AI ecosystem.
The monthly rhythm
Same order every month. Roughly 12 to 16 hours.
Hour 1: test. Your 20 prompts across the assistants your buyers use. Record the answers verbatim, dated, in the same document. Do not summarise them.
Hour 2: read what changed. Compare with last month. New names appearing, your own presence, anything stated about you that is wrong. Wrong statements are the most valuable finding, because they point at a source you can go and fix.
Hours 3 to 10: write one page. One properly. Answer in the first 60 words, the detail, the honest limitation, a named author, a visible date. One good page a month is 12 a year, which is more than most small businesses manage and enough to matter.
Hours 11 to 13: off-site. Two or three things: a directory corrected, a review request batch sent, a trade body listing updated, a comment offered on sector news, an event listing checked. Small, boring, cumulative.
Hours 14 to 15: admin. Update the enquiry source log. Check nothing broke. Note next month’s page.
Half an hour: read the enquiry sources. The only number with revenue attached. Ask on the form and read the answers.
The three signals to stop
Write the stopping rule before you start, because in month five everybody wants to move the goalposts in whichever direction the numbers went.
All three of these must be true before you conclude it is not working:
- The prompt grid has not moved by month 6. Same prompts, same wording, honestly compared
- Server logs confirm the AI crawlers are fetching your pages. If they are not, you have a technical problem, not a strategy problem, and stopping teaches you nothing
- The entity audit is verified clean. If your address still disagrees across four sources, the programme has not actually been tried
If all three hold at month 6, stop and put the money somewhere else. That is a legitimate outcome and having decided it in advance is what makes it a decision rather than a surrender.
If any of the three fails, you have found the actual problem, which is more useful than the original question. What the realistic curve looks like is in when will you see AI citations.
What goes wrong, ranked by frequency
Stopping at month four. By far the most common. The foundation work produced a small early movement, months two and three felt slow, and the slow part is where the compounding lives.
Nobody owned it. Covered above, and it is the root cause of most of the rest.
Publishing in bursts. Six pages in February and nothing until October. Recency is weighted, and a site that went quiet reads as inactive.
Buying tools before doing the work. A subscription is easier than an audit and it will not change a decision you make in year one.
Skipping measurement, then arguing about results. Without a recorded baseline the month-six conversation is two people with opinions.
Chasing the broad term. The category keyword is not available to you and it will absorb everything. Specificity is the whole small business advantage and it is explained in AI search for small business.
A note on what this is not
This page is deliberately not a pitch and not a pricing page. If you want the service, the packages are published, the method is published and the results are documented rather than described, including a small dental directory that went from zero to 66,000 impressions.
If you want to run it yourself, everything needed is in this cluster and free. That is a genuine option and it is stated here because most businesses at this size should take it for the first six months and decide afterwards with evidence.
AI Search Marketing for Small Business: Common Questions
How much should a small business spend on AI search marketing?
Enough to sustain it for a year, which matters more than the monthly figure. A programme that runs for 12 months at a modest level beats one that spends heavily for four months and stops, because the corroboration layer only accumulates with time. Set the budget by asking what you can maintain to month 12, not by what you can afford in month one.
Who should own AI search work in a small business?
One named person with two to four hours a week protected in the diary. It does not need to be a specialist and it must not be shared between three people who each assume somebody else is doing it. Ownership failure, not knowledge failure, is what stops most small business programmes.
How many hours a month does this take?
About 12 to 16 hours a month once the setup is done, and 20 to 30 hours in the first month because the foundation work is front-loaded. Roughly half of the ongoing time goes on writing, a quarter on off-site consistency and outreach, and the rest on measurement and admin.
Should a small business hire an agency for AI search?
Do the foundation layer yourself, because it is checkable and has a finish line, and paying a retainer for a week of work is poor value. Consider help at the corroboration and publishing stage, which needs sustained effort rather than knowledge, and which is where most small business programmes quietly stop.
What does a monthly AI search routine look like?
One hour testing your prompt set and recording the answers, one hour reviewing what changed, six to eight hours writing one good page, two hours on off-site consistency and outreach, and half an hour updating your source-of-enquiry log. Same order every month, same day if you can manage it.
When should a small business stop an AI search programme?
Set the rule before you start. A reasonable one is: if the prompt grid has not moved by month six, and crawler logs confirm the pages are being fetched, and the entity audit is verified clean, then stop and reallocate. All three conditions must hold, because stopping while the crawlers were blocked teaches you nothing.
Can this be done alongside a full-time job running the business?
Yes, at 12 to 16 hours a month, if it is diarised rather than fitted in. The businesses that fail at this are not the ones without time, they are the ones without a fixed slot. Two half-days a month, booked, protected and treated like a client appointment, is enough.
What is the cheapest version that still works?
Foundation work done properly in month one, one good page a month, quarterly listing checks, and a monthly prompt test. That is roughly eight hours a month and no cash cost beyond your time. It is slower than a funded programme and it is a genuine version of the work rather than a token one.
Where to Go Next
The tactics this operating model runs: AI search for small business · AI visibility strategy · how to audit your off-site AI ecosystem
Decide whether to buy help: in-house versus agency · can I do GEO myself · how much of my team’s time does GEO take
Set expectations: when will you see AI citations · what if I see nothing after six months · what happens if I pause GEO after three months
Have it delivered: MarGen is a UK GEO agency with published pricing, a published method and documented results.