AEO is the new hot category. Every SEO agency on the planet added “AEO services” to their landing page last quarter, and half of them are selling schema audits with a relabeled deck. The other half are doing real structural work — and the buying problem is that the pitch decks look identical. This piece is the filter.
Full disclosure upfront — we are Answerly, an AEO agency. We sell this work. The 12-question framework below works regardless of who you pick, and we built it from the inside because we kept losing prospects to vendors who could not have answered three of the twelve. If you read this and decide we are not the right fit, that is a successful outcome too.
When you actually need an AEO agency (and when you don’t)
Three honest filters before you talk to anyone.
If your AI visibility is already fine — meaning your brand shows up cited in ChatGPT, Perplexity and Gemini on the prompts that drive your buyer cycle — do not hire. Maintain what you have, ship one good page a month, and revisit in a year. Most companies that hire an AEO agency would be better off hiring one strong senior content writer instead.
If your category is not yet competitive in AI — small niche, mostly direct sales, buyers do not research in LLMs — wait six months and re-measure. We have walked away from prospects in 2025 whose category was structurally too small to justify the spend. The honest answer was “your buyers are not in ChatGPT yet, come back when they are”.
If you cannot ship content monthly — no content lead, no production cadence, marketing fights for every approval — fix that first. An AEO agency cannot save a content programme that does not exist. We have seen retainers burn for six months because the client’s legal team gated every page rewrite past the point where the citation data could move. Build the production capability before you optimise for the extraction surface.
If you pass all three filters — buyers do research in AI, you are losing share-of-voice to competitors, and you can move content through your org — then the agency conversation is the right next step. Read the 90-day AEO roadmap to calibrate what a serious engagement looks like before the discovery call.
The 12 questions every AEO agency should answer before contract
Print this list. Take it to every discovery call. If the answers are vague, walk.
1. Show three real client citation trajectories with named brands, not anonymous case studies. “Client A in fintech moved citation rate from 8% to 31% in 90 days” is not enough. Who is Client A. What was the baseline prompt set. Which LLMs. If the agency cannot name brands — usually because they signed weak NDAs or the trajectory does not exist — that is your answer.
2. Name the five AI platforms they track and how. ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews — these are the five that matter for B2B in English-speaking markets in 2026. Bing Copilot and You.com are optional. If the agency tracks only ChatGPT and Perplexity, they are running a half-stack. If they say “we track all of them” with no methodology — they are guessing.
3. Their opinionated position on AEO vs GEO. The honest answer in 2026 is “they are 80% the same practice run on one content team against one rewrite cycle”. An agency that sells AEO and GEO as two separate retainers is double-billing for the same structural work. An agency that says “we do both!” with no opinion has not thought about it. See our take in what is answer engine optimization and generative engine optimization.
4. Concrete monthly deliverables, numbered. Three pages restructured. Six FAQs shipped. Schema deployed across the target cluster. Weekly citation report against fifteen tracked prompts. If the deliverables list reads “AEO strategy, content optimisation, schema enhancement” with no numbers — they are billing for slide decks. Demand the count.
5. Reporting cadence and a sample report. Weekly tracking data, monthly written report, quarterly business review is the working pattern. Ask to see a redacted sample report before signing. If they refuse — usually because they do not have a real reporting template and would have to mock one up — that is the answer.
6. Named experts on the brand byline, not ghost-written. Who actually writes the content. Are they bylined on the page with a verifiable LinkedIn and a Person schema with sameAs. Ghost-written content from anonymous writers does not get cited the same way — LLMs weigh authorship attribution heavily. Our named experts and E-E-A-T piece covers why this is a real signal and not E-E-A-T theatre.
7. Their internal scorecard or playbook — if proprietary, ask to see one chapter. Every serious agency has an internal playbook. The 4-layer extraction recipe, the schema deployment checklist, the citation tracking SOP. If they will not share one chapter under NDA — they probably do not have a playbook and are improvising. The chapter does not need to be the secret sauce. It needs to exist.
8. Cancellation terms. Thirty-day notice on a monthly retainer is normal and reasonable. Six-month lock-in is suspicious unless the price is heavily discounted to justify the risk. Twelve-month contracts with no break clause — walk. The agency is protecting themselves against the case where the work does not move citations.
9. Pricing model, with tradeoffs explained. Fixed retainer, performance bonus, hourly — each has tradeoffs. Fixed retainer is the working model for AEO because citation outcomes are probabilistic and a performance bonus on probabilistic outcomes is a dispute machine. If an agency offers a “pay only when cited” model — read the fine print, the threshold is usually unreachable. If they offer pure hourly — they are not running a programme, they are renting consulting hours.
10. Who actually does the work. Strategist? Content team? Junior writer with an AI tool? Ask for the named team on your account before you sign. The pitch deck shows the senior partner. The retainer often gets staffed with a junior plus a content mill. The two outputs do not look the same.
11. How they handle Cloudflare bot fights and crawler blocks. Cloudflare’s bot protection in 2026 will block GPTBot, ClaudeBot and PerplexityBot by default on many configurations. If the agency does not know which AI bots the target site blocks today and what the policy decision is — they are missing the first technical layer. An AEO programme that ships great content to a site that blocks AI crawlers is shipping into a void.
12. What they explicitly will NOT do. This is the most diagnostic question on the list. An honest agency has a list — “we do not write thin listicles”, “we do not deploy invalid schema”, “we do not run programmes under $890/mo because the math does not work”, “we do not take crypto clients without their named experts on the byline”. If the answer is “we do everything!” — they are pitching, not advising.
Twelve questions, twelve concrete answers expected. Most agencies stall on three to five. The ones who answer all twelve crisply are the short list.
5 red flags that mean “skip this agency”
Each of these has cost prospects we work with real money and real quarters of lost time. The list is opinionated and the patterns are real.
- “Guaranteed citations” promise. Citations are probabilistic — they depend on the LLM, the prompt phrasing, the retrieval window, the competitor field that day. Any agency promising guaranteed citations either does not understand the surface or is willing to lie to close. Walk on contact.
- No named team — only the agency brand. “The Acme Team will execute your AEO programme” with no humans listed. LLMs cite named experts with verifiable credentials. An agency that hides its own team behind a brand cannot deliver named-expert work for you either. The pattern compounds.
- $300/mo “AEO package”. The math does not work. A serious AEO engagement runs eight to fifteen hours per week of senior time at minimum — content strategy, schema deployment, citation tracking, reporting. At $300/mo the agency is either reselling someone else’s report or doing nothing. There is no third option.
- Will not show a sample report or scorecard. “Our reporting is proprietary” is not a defense — the format is not the secret, the work is. An agency that refuses to show a redacted sample before contract is hiding either the absence of a real report or the embarrassing quality of the one they have.
- “We use AI to optimise for AI” with no human review. AI-generated content with no editorial layer, no named expert review, no fact-check pass. LLMs increasingly weight content provenance, and unedited AI output is a citation-rate liability — not an efficiency play. If the agency is openly running AI-only writing as their differentiator, they are a generation behind the actual extraction signals.
Two of these are usually enough. Three is a hard no.
AEO agency vs in-house team — when each wins
The comparison nobody on the agency side wants to write honestly. We will write it anyway.
| Factor | In-house wins | Agency wins |
|---|---|---|
| Senior content lead in place | Yes — they already know your voice | No — agency brings the discipline |
| Publishing cadence | 4+ pages/month sustainable | Below 2 pages/month — outsource the cadence |
| Niche regulation | Yes — heavy compliance, in-house counsel | Light regulation, agency can move faster |
| Schema engineering | Engineering team available | No engineers — agency owns the layer |
| Citation tracking discipline | Analyst available, tooling budget | No tooling — agency brings the stack |
| AEO programme age | 12+ months of structured work | First six months — agency teaches the team |
| Cost ceiling | $15k+/mo loaded cost | Under $10k/mo for equivalent output |
In-house wins when you have an experienced SEO or content lead who can absorb AEO as a discipline in two months of self-study and tooling spend. The four-layer extraction recipe is not a secret — it is documented in our pillar piece and in two dozen other places. If your content lead reads the documentation, runs the best AEO tools 2026 stack, and ships against a structured roadmap, they will beat most agencies at half the cost.
Agency wins when you are cold-starting AEO, your content baseline is shaky, or you need a 30 to 60-day learning curve compressed into 30 to 60 days of someone else’s salaried time. The agency has the muscle memory of fifty engagements and will not repeat the same five mistakes on your account that an in-house team would make in their first quarter. After six to twelve months, in-house usually catches up — and the agency that does not see that coming gets churned.
The hybrid model is underrated. Agency runs the first six months as the AEO programme owner. In-house team shadows, learns the methodology, takes over the production cadence by month six. The agency stays on retainer for citation tracking, quarterly strategy and the platform fights that the in-house team will not have seen before. This is how most of our second-year retainers look — reduced scope, deeper specialisation, lower cost per month.
What pricing actually buys you in 2026
The honest breakdown. No price-on-request fog.
$500–1,500/mo — audit-only or single-page rewrite. A starter audit at this range buys you a baseline citation report across the major LLMs, a competitor teardown of three rivals, and a punch-list of fixes. That is it. No ongoing programme, no schema deployment, no monthly cadence. Useful as a diagnostic before committing to a real programme. Below $500/mo, the math does not support real audit work — anyone selling at that price is running automated tooling with no human review.
$1,500–3,000/mo — schema plus 2-3 page rewrites plus monthly FAQ. The minimum viable AEO programme. Suitable for small B2B brands with one priority cluster, an existing content team, and the willingness to be patient about cadence. You will not lead the category at this tier. You will move citation rate on three to five priority prompts over a quarter.
$3,000–5,000/mo — full AEO programme with citation tracking. Where serious work starts. Weekly citation tracking, monthly content output of three to five pages, schema engineering across the cluster, named-expert bylines deployed, quarterly business review. This is where most B2B brands should sit if AEO is one of two or three priority channels.
$5,000–10,000/mo — dedicated team. Strategist plus content team plus citation analyst staffed on your account, not shared across twelve clients. This tier earns its premium through speed — instead of three months to a working programme, you get to a working programme in six to eight weeks because there are more hands.
$10,000+/mo — enterprise category leadership. Multi-cluster work, ten to fifteen tracked prompts becoming forty to sixty, share-of-voice tracking against named competitors, deep schema integration with your CMS, named experts hired and onboarded if you do not have them in-house. This is where you go when AEO is a board-level metric and the question is not “should we be cited” but “should we own the category prompt outright”.
The pricing page shows our exact tiers and the niche multipliers — crypto runs 2.0x, SaaS 1.15x, local services 0.65x — because the underlying work surface and competitive intensity differ that much by category.
Questions to ask on the discovery call (steal these)
Print this section. The agency has a script. You should too.
- “What is the smallest engagement you would take, and what would you NOT deliver at that price?” Tests honesty about scope. The right answer names a tier and lists what the tier explicitly excludes.
- “Show me a case study where the citation score did NOT move.” Every serious agency has at least one. If they claim 100% success across fifty engagements — they are lying or they are filtering out the cases that did not work.
- “How do you split AEO work between in-house staff and freelance?” The pitch deck implies in-house team. The retainer often runs 60% freelance content writers. Both can work — but the answer should be specific.
- “What does month one look like, end to end?” Audit week, taxonomy week, schema week, first content pieces in flight by week four. If the agency talks about “kickoff calls and stakeholder alignment” through month one, they are billing strategy hours instead of shipping work.
- “Which of my competitors are you already tracking, and what did you see last week?” Catches the agencies that have not done basic homework before the call. Five minutes in Perplexity gives any serious agency a working answer.
- “When would you fire us as a client?” Honest agencies have a list. Clients who will not move legal approval, clients whose niche is too small, clients who push for unethical SEO tactics. If the agency cannot name a firing condition — they will take any contract that closes.
- “What is your position on Cloudflare blocking AI bots — do you recommend opening or closing?” Tests technical depth. The right answer is opinionated and depends on the business model. The wrong answer is “we will figure it out together”.
- “Who actually owns the relationship after signing — the partner who pitched, or someone else?” Pitch person should be on the account for at least the first quarter. After that, named account lead is fine. If the partner disappears at signing, you bought a different service than you thought.
Eight questions. Two minutes each. Most discovery calls end before all eight are asked, which is itself a useful filter.
After signing — what should month 1, 3, 6 look like?
Calibration so you can hold the agency accountable. The cross-link is to our 90-day AEO roadmap for the full blow-by-blow.
Month 1. Baseline citation report against fifteen to thirty tracked prompts. Competitor teardown on three to five rivals. Schema audit and validation across the priority cluster. Two to three pages rewritten under the 4-layer extraction recipe. Named-expert bylines deployed if not yet in place. First weekly tracking report by week three at latest. By end of month one you should be able to point at concrete deliverables — not slide decks, not “kickoff alignment”.
Month 3. Citation rate should be moving on at least half the tracked prompts. Not winning, moving. Eight to twelve pages restructured or shipped new. Schema clean across the cluster. Named-expert profiles live and gaining credentials. Quarterly business review with concrete numbers. If three months in you cannot tell whether AEO is working — the agency is not measuring properly.
Month 6. Working programme with predictable monthly cadence. Citation rate at 25–35% on a clean B2B niche from a 0% baseline — that is the realistic 90-day target stretched to six months for slower categories. Share-of-voice climbing against named competitors. In-house team starting to absorb methodology. Decisions to make on whether to scale up tier, hold, or move to hybrid model. Read measuring AI citations for the metrics framework you should be holding the agency to.
If month six looks substantially worse than this — citation rate flat, no clear story on what worked or did not, agency dodging the quarterly review — fire and switch. Time is the most expensive part of an AEO programme and a bad agency burns it cheaply.
Our own pitch (with full disclosure)
We are Answerly Agency. Five service tiers from $890/mo to $8,900/mo, with niche multipliers — crypto 2.0x, SaaS 1.15x, local services 0.65x — because the work intensity is genuinely different by category. Two founders bylined on the work, named content team, citation tracking running on Searchable Agent plus in-house Anthropic API monitoring, schema engineering on Astro and Next sites by default.
We pass our own twelve-question framework. We will tell you on the discovery call if your category is too small, your content cadence too weak, or your budget too low for serious work. We have walked away from five-figure-per-month prospects in 2025 because the fit was wrong, and the reputational cost of taking those engagements would have been higher than the revenue.
If you want to test us cheaply before any commitment, run the free AI visibility audit — you get a baseline citation report across the five major LLMs and a punch-list of fixes ranked by citation potential per hour of effort. Read the report, decide whether the analysis is sharp enough to trust with money. If yes, services overview shows the tier structure. If no, the report is still yours to keep and the framework above still works with any vendor.
Closing
The agencies that will own AEO in 2027 are the ones telling you on discovery call “we might not be the right fit”. If that conversation is not happening — if every pitch ends in “we can absolutely do this for you” — you are talking to a sales operation, not an AEO agency. Walk, run the twelve-question framework on the next three vendors, and pick the one whose firing-condition list is longest. That is the working filter and it is the same one we apply when we hire subcontractors for our own overflow.