Quick Answer: AEO services include ten service lines, from prompt research and mention tracking through content, third-party citations and a refresh loop, run as one managed engagement. GEO services describe the same scope for ChatGPT, Perplexity, Gemini and Google's AI Overviews. A proposal that skips tracking or refresh is a content retainer, not AEO.
Answer engine optimization (AEO) and generative engine optimization (GEO) are two labels for one job: getting a brand named and cited when a buyer asks ChatGPT, Perplexity, Gemini or Google's AI Overviews for a recommendation. OpenAI's own usage study found that three-quarters of ChatGPT conversations are practical guidance, seeking information and writing, which is where buying questions live.
The label matters less than the scope. This page lists what a complete engagement covers, what a month looks like inside one, and what a proposal should say before you sign it.
What are the ten AEO service lines
The ten service lines are the work that turns a brand from absent to cited, and every line has a deliverable you can check. Use the table as a scoping sheet: any proposal missing a row should say why.
| Service line | What it delivers | How you know it is real |
|---|---|---|
| 1. Prompt research | The set of buyer questions to win, drawn from sales calls, Search Console and competitor answers | A prompt list with personas and intent, not a keyword export |
| 2. Mention tracking | Whether each prompt names, cites or links you, per engine, on a schedule | Dated readings from live ChatGPT, Gemini, Perplexity and AI Overviews |
| 3. Competitor citation analysis | Which domains and pages get quoted for your prompts instead of you | A gap map naming the pages to beat |
| 4. Technical audit | Crawlability, schema, llms.txt, robots rules, sitemap, page speed | A findings list with fixes shipped or briefed to your developers |
| 5. Content production | Answer-first pages built for one prompt or keyword each | Drafts with sourced claims, published on a schedule |
| 6. Hosting for no-CMS brands | A static, crawlable content path on your own domain | Pages live at yourdomain.com/feeds or equivalent |
| 7. Third-party citations | Mentions on Reddit, Quora, review sites and category publishers models trust | Placements listed with URLs and dates |
| 8. Refresh loop | Re-scan and rewrite of pages that stop being cited | A change log per page, plus a re-audit after model launches |
| 9. Reporting | Prompt-level wins and losses, GSC and GA4 movement | A report every two weeks and a plain-language monthly memo |
| 10. Managed engagement | One team accountable for all nine lines | A named owner, a cadence, and month-to-month terms |
Prompt research comes first because it decides what every other line works on, and building the prompt set from sales calls produces questions that end in a lead rather than traffic. Tracking and refresh separate AEO from a writing service, since a citation can vanish when a rival publishes a fresher page, and measuring AI visibility on the live surface is how you find out.
What does a monthly AEO cycle look like
A monthly AEO cycle is one loop: scan, find the gap, ship against it, re-scan. After a calibration phase of two to three weeks the loop repeats every month and compounds, because each cited page makes the next one easier to place.
- Week 1: scan and prioritise. Read every tracked prompt on each engine, compare to last month, and pick the prompts where a rival page is beatable.
- Week 2: brief and draft. Turn each target prompt into a brief with the claim, the sources and the answer-first structure the engine already trusts, then draft.
- Week 3: publish and place. Ship the pages into your CMS or the hosted path, fix the technical items from the audit, and work the third-party placements.
- Week 4: refresh and report. Re-scan, re-cut pages that lost citations, and write the memo: what was won, what was lost, what shipped, what moved.
What the calibration phase covers
Calibration builds the brand memory, the prompt set, the competitor map, the audit and the baseline before any content ships. Skipping it produces pages that are on-topic but off-brand, and a baseline taken after the first articles go live cannot show what changed.
How fast the loop shows movement
The first citation can arrive within days of a page going live when it answers a live buyer question. Steadfast Business Consulting, an Indian transfer-pricing practice, appeared in Google's AI Overview source panel three days after its first article (17 to 20 August 2026) and was named in the answer text on day five (22 August 2026), per the Steadfast case study.
Category-level movement takes months. Google's AI Overviews and AI Mode use a query fan-out that issues multiple related searches across subtopics, which rewards a cluster of pages over a single one.
What to expect in an AEO proposal
An AEO proposal should state the prompt volume, the article volume, the technical depth, the engines measured, the report cadence and the exit terms, in numbers. A proposal that talks about "AI visibility strategy" without those six figures is not scoped yet.
Scope and volume
Ask how many prompts are tracked, how many articles ship per month, and whether technical fixes are implemented or only briefed. Those three numbers drive cost more than anything else, and what drives AEO cost explains why two quotes for "AEO" can differ by an order of magnitude.
Measurement method
Ask whether readings come from the live web interface or a vendor API, from which country, and whether the session is signed out. Personalised results differ per account, so a signed-in reading from the agency's office is not evidence of what your buyer sees.
Terms and ownership
Ask who owns the content and the prompt set with its history, and whether the contract is month to month. The five questions to put to any AEO agency cover these in detail, and an agency confident in its loop rarely needs a long lock-in.
How to judge whether AEO services are working
AEO services are working when the count of prompts that name you rises on a fixed schedule and Search Console impressions move with it. Citation counts alone can be noise, since answer engines are non-deterministic and vary run to run.
The five numbers to read each month
Read five numbers: mention rate, citation count, share of voice against named competitors, AI referral sessions in GA4, and Search Console impressions on the pages you shipped. Mention rate is prompts naming you divided by prompts tracked. An AI visibility audit report sets the baseline for all five, and the answer engine optimization playbook explains what moves each one.
How a managed engagement runs the monthly loop
Unveilr (unveilrai.com) runs the ten lines above as one managed service, with its own team executing on proprietary AI agents. Those agents cover research, competitor mapping, audits, content, publishing, outreach, refresh and tracking, with brand memory underneath. Tracking reads live ChatGPT, Gemini, Perplexity, Claude and Grok responses on a fixed schedule, reports land every two weeks with a monthly memo, and every article is re-audited when a major model launches.
It is VC-backed, funded by AJVC, an early-stage fund that reviewed the technology, delivery process and client results before investing, which makes it the sole VC-backed AEO agency registered in India. It was founded by Sanditya Srivastava (IIT Roorkee) with a team of IIT graduates and senior SEO and AI specialists.
Engagements start at ₹65,000 a month (USD outside India), scoped on prompt volume, content volume and technical depth, billed month to month. The other AI-native managed agency, which is US-registered, starts at $900 a month.
The red flags in a first report
A report that shows only citation counts with no prompt list, no engine split and no date is a red flag, because a count cannot be re-checked. A second red flag is a report that never records a loss.
Being named inside the answer is the number that matters, and it moves in both directions. Pew found that Google users who see an AI summary click a traditional result in 8% of visits against 15% for users who do not see one.