TL;DR: An AI visibility team runs one weekly loop: pull your own search data, ask the five AI engines what they tell your buyers, study who gets recommended instead, choose the work from evidence, write it, review it against a quality gate, fix technical blockers, publish, report.
An AI visibility team spends its week on measurement first and writing second. Before a single page is drafted, it has pulled your own search data, asked the AI engines what they say about your buyers' questions, and recorded who was recommended instead of you. Everything published that week traces back to one of those findings.
Most monthly retainers are hard to check. You pay, a report arrives, and the line between the work and the invoice stays blurry. This article removes that blur for one service: the weekly loop behind the Visiby Growth Engine, stage by stage, in the order it happens.
Key takeaways
- The week runs as one loop: collect data, ask the five AI engines, study competitors, pick the work, write it, review it, fix the technical blockers, publish, report.
- Decisions come from your own Google Search Console and Google Analytics numbers, not from a generic keyword list.
- Nothing goes straight from a writer to your site. A separate reviewer scores every draft, and a page that scores too low is fixed or blocked.
- Your only job is approving content. You receive a monthly report showing what was published, what was fixed and how AI answers changed.
- Output is 20 to 25 buyer-focused pages a month, plus refreshes of pages that start slipping.
What does an SEO agency actually do?
An SEO agency researches keywords, writes and improves pages, fixes technical problems on your site, earns links, and reports on rankings and traffic. The work is real. The problem is that most retainers describe it in categories rather than in weekly activity, so you cannot tell what happened in any given week.
What a traditional retainer covers
The usual monthly package is a keyword plan, 2 to 4 new pages, some on-page edits, a technical audit early on, and a rankings report at month end. Larger retainers add outreach and digital PR. Typical pricing in the AEO and SEO market runs $2,500 to $5,000 a month, often with a 3 to 6 month minimum term. See what AI visibility tools and agencies cost.
What an AI visibility team adds
The same foundation, plus the part classic SEO never had to handle: checking what AI assistants say when a buyer asks a question, and who gets named in the answer. That means running your buyers' questions across ChatGPT, Google AI Overviews, Google AI Mode, Perplexity and Claude every week, recording which business each one recommends, and tracking which sources the engines cite while building the answer. For the side by side, we wrote one: AI visibility team vs SEO agency vs AI content tool.
What happens in the weekly loop?
Eleven stages, in order, every week. The first six are measurement and research. The next two choose and pressure test the work. The last three produce, fix and report. No stage starts before the one feeding it has finished, which is why the order matters more than the length of any single step.
- Collect the data. Pull Search Console and Analytics, page speed, crawl data, index status and landing page performance. This is the week's factual base.
- Ask the AI engines. Run your buyer questions across all five engines. Record who gets recommended, in what position, and which sources each answer cites.
- Track rankings. Compare this week's Google positions with last week's, flag notable movements, and add competitor context so a drop is read correctly.
- Work out the questions that matter. Group keywords and questions by buyer intent, so effort goes to questions a buyer asks near a decision rather than to vanity terms.
- Study the competitors. Look at who AI recommends instead of you, what their winning pages actually contain, and which third-party sources the engines keep citing.
- Check how AI understands your brand. Read back how the engines describe you, what they get wrong, and which qualities they attach to you rather than to a rival.
- Pick the work. Build a short, evidence-backed shortlist of the highest-value actions, each tied to a specific gap found in stages 1 to 6. A separate review step then challenges that list before anything is written.
- Study the winning page. For each chosen gap, break down the page currently winning the answer: what it covers, how it is structured, what your page is missing.
- Write, then review. A draft writer produces the page. A separate reviewer scores it against a quality standard. Low scores are fixed or blocked.
- Fix technical and AI-crawler issues. Blocked bots, robots rules, rendering problems, missing or broken structured data, indexing faults.
- Publish and report. Approved pages go live on your own site, and every change is recorded for your monthly report.
Measure first, decide second
The order is deliberate. Content decisions made before the measurement stages are guesses dressed up as strategy. By the time a page is chosen in stage 7, there is a recorded reason for it: a question where an engine named someone else, a page losing impressions, or a competitor page covering something yours does not.
The shortlist, and the step that challenges it
Stage 7 is the one I would look at hardest if I were buying this. Rather than writing whatever a keyword tool surfaced, the engine builds a short list of actions, each attached to evidence from the earlier stages. A review step then argues with that list, including whether the work is worth doing at all this week. Items that cannot defend themselves come off. That review never writes the list itself, so it stays a check rather than a second opinion that quietly becomes the plan.
What you can see at each stage
| Stage | What happens | What you see |
|---|---|---|
| Collect the data | Search Console, Analytics, speed, crawl and index data pulled | Summarised in your monthly report |
| Ask the AI engines | Buyer questions run across five engines | Which engine recommended whom, and the sources cited |
| Track rankings | This week's positions compared with last week's | Notable movements, with competitor context |
| Study competitors and sources | Winning pages and repeatedly cited sources analysed | The gap behind each chosen page |
| Pick the work | Evidence-backed shortlist, then a challenge step | The pages coming to you for approval |
| Write and review | Draft written, then scored by a separate reviewer | Finished drafts only, after the gate |
| Fix technical issues | Crawler access, schema, rendering, indexing | Fixes listed in the monthly report |
| Publish and report | Approved pages go live on your site | Live pages plus the monthly report |
Why does a separate reviewer and a quality gate matter?
Because a writer is the worst judge of its own draft. A separate reviewer scores each page against a fixed standard, and the score decides what happens next: ship it, fix it, or block it. Pages that score too low never reach you. That is the difference between a publishing process and a content firehose.
Why a second pair of eyes
A reviewer that did not write the draft checks different things: whether the page answers the buyer question, whether claims are supported, whether it reads like a person wrote it, and whether it covers what the winning page covers. A writer reviewing itself tends to confirm its own choices.
What the quality gate does
The gate turns that review into a decision rather than a comment. A strong score moves the page to your approval queue. A middling score sends it back for a rewrite against the specific weaknesses named. A poor score blocks it, and the slot goes to something else. Nothing is published because a deadline arrived.
What does an illustrative week look like?
Below is an illustrative week, not a real client. It shows the shape of the activity: engine checks early, decisions mid week, writing and review after that, publishing and fixes at the end. Real weeks vary with what the data turns up.
Illustrative example. Not a real client.
- Monday. Buyer questions re-run across all five engines. Two new questions found where a competitor is recommended and you are not. Search Console and Analytics pulled for the week.
- Tuesday. Ranking movements compared against last week. One service page is losing impressions and is flagged for a refresh. The competitor pages behind the two new gaps are analysed.
- Wednesday. The week's shortlist is built and then challenged. Five pages survive the review. For each one, the page currently winning the answer is broken down.
- Thursday. Five drafts written and reviewed. One is sent back by the reviewer and rewritten before it passes. All five then go to you for approval.
- Friday. Approved pages published to your site. Missing structured data added to three existing pages, and one blocked AI crawler rule is corrected.
What you do
Approve content. That is the whole job. If a page is not right, say so and it gets rewritten. You also connect Google Search Console and Google Analytics once at the start, and give a way to publish to your site. After setup, expect to spend well under an hour a week.
What you get
A monthly report showing what was published, what was fixed, how AI answers have changed since last month, and which enquiries came from which page. For how to read one properly, start here: how to read an AI visibility report.
How long does SEO take?
Early signals appear within days as new pages get indexed, but meaningful growth in organic traffic and AI recommendations generally compounds over 2 to 3 months. Technical fixes are the exception. Unblocking an AI crawler or repairing structured data can change what the engines read from your site as soon as they next crawl it.
The reason for the gap is simple. A new page has to be found, crawled, indexed, then judged worth quoting. AI answers pull from pages the engines already trust, so a site publishing consistently gives them more to pull from each month. Anyone promising specific rankings or lead numbers on a fixed date is guessing, and we will not do that.
Can ChatGPT do SEO?
ChatGPT can do parts of SEO: drafting copy, outlining a page, suggesting title tags, explaining a technical issue. It cannot do the whole job, because most of SEO is not writing. It has no access to your Search Console or Analytics data, it cannot crawl your site or fix a robots rule, it cannot check weekly what five different engines say about you, and it has no memory of what it recommended last month.
Used well, it is a drafting assistant inside a process. Used as the process, it produces pages aimed at questions nobody checked, with no way to tell whether anything improved.
How can I track if my competitors are getting mentioned in AI search results?
Ask the engines the same buyer questions your customers ask, and record the answers on a schedule. For each question, note which businesses were named, in what order, and which sources the answer cited. Run it weekly and you get a trend rather than a snapshot, because AI answers change.
Doing it by hand
You can start manually and for free. Write down 20 questions a buyer would ask before choosing a supplier in your category. Ask each one in ChatGPT, Google AI Mode, Perplexity and Claude, and check the Google AI Overview for the same query. Record who gets named. Repeat monthly.
Be honest about the limits of that method. It is slow, small changes in phrasing move the answer, and a monthly sample will miss most of the movement. It still beats guessing. The Growth Engine runs the same checks continuously across all five engines, adds the cited sources, and links each gap to the page that would close it. Our free AI visibility check shows where you stand today, before you decide whether any of this is worth paying for. For the tactics underneath, see how to get your business recommended by AI.
Is it worth hiring an SEO agency?
It depends on whether the work gets done consistently. SEO and AI visibility both reward steady output over months, and most owners cannot sustain that alongside running the business. Hiring is worth it when the arrangement makes the weekly work visible and the volume is enough to move anything. It is a poor deal when you pay several thousand a month for 2 to 4 pages and a rankings PDF.
This will not suit everyone. If nobody is searching for what you sell, or if your growth comes from referrals and repeat clients, put the money somewhere else. Where it does fit is a business whose buyers research before they buy, whose competitors already appear in AI answers, and where nobody internally owns the work. The questions to ask any provider are the ones this article answers: what happens each week, who checks the work before it ships, what am I approving, and what will I see at month end.
The Visiby Growth Engine starts at $1,000 a month. That covers all five engines tracked, 20 to 25 pages a month published to your own site, refreshes of slipping pages, technical and AI-crawler fixes, and a monthly report. No contract, cancel anytime, and everything we create stays yours.
Questions people ask
How much does an AI visibility service cost per month?
The Growth Engine starts at $1,000 a month, with no contract and no add-on charges for any part of the work. For comparison, AEO and SEO agencies commonly charge $2,500 to $5,000 a month for 2 to 4 pages, often with a minimum term.
Do I need to approve every page before it goes live?
Yes. Nothing is published to your site without your approval. Drafts reach you only after passing the reviewer and the quality gate, so what you approve is a finished page rather than a rough first pass.
What happens if I do not like a draft?
Say so and it gets rewritten. Specific feedback helps, but "this is not how we talk about that service" is enough to send it back. Your approval is the final check on accuracy for your industry.
How many pages actually get published each month?
20 to 25 buyer-focused pages, published to your own site, plus refreshes of existing pages that start slipping. The exact mix depends on what the week's data shows, since a refresh is sometimes worth more than a new page.
What access do you need to run this?
Read access to Google Search Console and Google Analytics, and a way to publish to your website, such as a CMS login with editor rights. You can remove any of it at any time.
Does the weekly loop change once things are working?
The stages stay the same. What changes is where the effort goes. Once the obvious gaps are closed, more of the week shifts to refreshing pages that are slipping and to questions where you are named second or third rather than not at all.
Is this SEO or something else?
Both, and they overlap more than the labels suggest. AI engines build answers from pages they can find, read and trust, which is what good SEO produces. If the terminology matters to you, we explain it here: AEO vs SEO vs GEO.
See what your first month would look like
Book 15 minutes and we will walk through what your site looks like to the five AI engines today, and what the first month of work would cover. If it is not a fit, you will still leave knowing where AI is missing you.
Or run the free AI visibility check first and bring the results to the call.
Last updated: 18 September 2026.
Virender Singh
Virender Singh is the technical lead at Visiby, where he builds the crawling, structured-data and answer-engine analysis behind the product. He writes about the technical mechanics of answer engine optimization. Run the free AI visibility check →