Sometimes — and the condition is measurable in an afternoon. Across 13,770 enterprise domains analysed by Conductor between May and September 2025, AI referrals averaged 1.08% of all website traffic, from 0.25% in communication services to 2.80% in IT. Below that floor, with thin deal margins, the honest answer is no.
At a glance: when AEO pays and when it doesn’t
- The definition. AEO is the work of getting your business named, quoted or linked inside generated answers on ChatGPT, Gemini, Perplexity and Google AI Overviews, measured as share of answer: hits divided by tracked prompts, per engine.
- The boundary. Not an ad buy, not a schema plugin, not an llms.txt file. Nobody controls an engine’s ranking function, so nobody can sell guaranteed placement in an answer.
- The economics today. Volume is small, quality is high. AI assistants sent leadsnow.ai 56 deduped landings in the 15 days from 22 August to 5 September 2026, against 245 bot requests per landing.
- The test. Three gates: a measured AI referral share, a deal margin that covers a month of the work, and 20+ buyer prompts you can write down. Fail one and the answer is no.
- The honest part. Our own latest reading fell: on 11 September 2026 we were cited in 1 of 15 ChatGPT answers and 1 of 15 Gemini answers, 7% each, down from 20% and 36% two days earlier.
How it works
How to decide whether AEO is worth paying for
Measure your AI share
Take 90 days of sessions and count AI landings by utm_source tag first and referrer second, then dedupe the union. Referrer alone under-reads it badly.
Price one closed deal
Work out gross margin per closed deal, not revenue. This is the number the whole decision turns on.
List 20 buyer prompts
Write the evaluative questions a buyer would type into an assistant, in their words. Fewer than 20 means nobody is asking.
Run the three gates
Share, margin, prompts. Fail any one and the answer is no this year; re-measure in two quarters.
MAKE MORE SALES.
Pay-Per-Result pricing — We scale sales HARD aligned to your interests, better than anyone else.
What AEO actually is, and what it is not
AEO is the practice of making your business the thing an assistant reaches for when a buyer asks a question in their own words. The unit of work is the prompt, not the keyword, and success is binary: you are in the answer or you do not exist for that question.
Four things get sold as AEO that are not. A schema audit is plumbing — necessary, not sufficient. A rank tracker pointed at Google measures a different surface. An llms.txt file is a five-minute chore with almost no evidence behind it. And “guaranteed placement in ChatGPT” is not a product anyone can deliver: the ranking function belongs to OpenAI and changes without notice. If a proposal cannot tell you which prompts it is being scored against, it is not an AEO proposal. The metric that makes it accountable is share of answer, measured per engine on a fixed prompt set.
Want this done for you? We book qualified sales appointments on a Pay-Per-Result basis — you only pay for calls that actually land in your calendar.
Is AEO worth paying for in my business? The three-gate test
The decision rule — call it the assistant-share test. All three gates must pass; one failure is a no, and a no is a legitimate outcome.
- Gate 1 — measured share. AI-referred visits are at least 1% of your total website traffic, or have grown month on month for three consecutive months. The 1% is Conductor’s cross-industry average of 1.08% (13,770 enterprise domains, May–September 2025). Read it as the top of the range rather than the norm: Ahrefs measured AI chatbot traffic at 0.17% of the average website’s traffic across 3,000 sites in February 2025, so sitting under 1% is common and the growth branch of this gate is the one most businesses will use. Below your own industry’s figure and flat means your buyers are not asking assistants yet.
- Gate 2 — margin arithmetic. The gross margin on one closed deal exceeds one month of the work. Written as a formula: AEO clears break-even when
AI landings/month × enquiry rate × close rate × gross margin per deal > monthly cost of the work. - Gate 3 — surface area. You can write down 20 or more evaluative prompts — “best X for Y”, “X vs Y”, “how much does X cost” — whose answers would name vendors like you. Below 20, nobody is asking, and optimisation cannot create the question.
Measure gate 1 before you believe anyone, including us, and measure it with two signals. On our access logs, counting AI landings by referrer header alone missed 31 of 56 (55%), because app sessions arrive with no referrer; the utm_source tag alone missed 10 (18%). Match the tag first, the referrer second, dedupe the union — the method is in our write-up of AI referral traffic as a lead channel. A referrer-only report under-reads your share by roughly half.
The worked sum: what one month of AEO has to produce
Substitute your own numbers. The input is the landing volume we measure on this site rather than a projection: roughly 110 AI landings a month, on a 400-plus page site publishing daily.
- 110 AI landings × a 3% enquiry rate = 3.3 enquiries a month. Use your site-wide rate, not an optimistic one, even though AI-referred visitors typically convert better than organic: if it only works on the optimistic number, it does not work.
- 3.3 enquiries × a 25% close rate = 0.8 closed deals a month.
- 0.8 deals × gross margin per deal (M) = 0.8M of gross profit a month.
- Break-even against a monthly cost of the work (C): M must exceed 1.25 × C.
Rounded into something you can hold: one closed deal has to be worth more than a month of the work, or AEO does not pay at today’s volumes. Where the gross margin on one closed deal is a tenth of what a month of the work costs, you need more than ten times our measured landing volume to break even — not something you can buy your way out of. Run the sum before you run a tender.
If we can’t make you money, we don’t deserve yours.
Pay-Per-Result pricing — performance-based alignment.
Who should not pay for AEO — five readers, named
Where we would tell you no, and the measurement that decides it.
| If this is you | The measured signal | Worth paying for? |
|---|---|---|
| Low-margin, high-volume seller | Gross margin per deal is less than one month of the work (M < C) | No. Conversion rate and pricing move more money. |
| Category your buyers don’t ask assistants about | Under 0.3% of traffic and flat over 6 months — Conductor put communication services at 0.25%, utilities 0.35% (13,770 domains, May–Sep 2025) | No, not this year. Re-measure in two quarters. |
| Fewer than 20 nameable buyer prompts | You cannot list 20 evaluative questions a buyer would type | No. That is a demand problem, not a citation problem. |
| Thin site with no published proof | Under ~30 indexed pages, no original data, no case detail | No. Nothing extractable to cite. Publish first. |
| Single-location trade or retail business | Zero AI landings across a 90-day log window; demand arrives via maps and referral | No. Spend it on reviews and response time. |
| All three gates pass | ≥1% AI share or 3 months of growth, M > 1.25C, 20+ prompts | Yes, on the condition it is measured per engine from day one. |
Five of the six rows above are a no. A vendor who cannot produce that table has not done the arithmetic on your business.
What our own measurement looks like, including the month it fell
We run a citation monitor against a fixed prompt registry. Between 3 July and 11 September 2026 we polled that set 774 times on ChatGPT with browsing and 597 times on Gemini (grounded), across 71 distinct prompts: leadsnow.ai appeared in 12.7% of the ChatGPT answers (98 of 774) and 40.9% of the Gemini answers (244 of 597). Same site, same prompts, same cadence — a 28-point gap between two engines.
The most recent reading was worse than the average, and a page arguing about value that hides its own bad month is not worth citing. On 11 September 2026 we were cited in 1 of 15 ChatGPT answers and 1 of 15 Gemini answers — 7% each, against 3 of 15 (20%) and 5 of 14 (36%) two days earlier. Two caveats, both against us: the active registry grew from 120 to 132 prompts over the period, so rates are not strictly comparable run to run, and this is one site’s measurement of one prompt set, not an industry benchmark. Over the same two days the site served 4,527 AI crawler and assistant fetches against 9,342 human pageviews — crawling up 41% while citations fell. Being read more and quoted less is the shape of this channel right now, and anyone selling you a straight line is selling you something else.
How much of the AEO hype is real?
Two claims are usually welded together in a pitch and deserve separating. The first — that assistants are taking clicks away from the open web — is measured: the Pew Research Center tracked 900 US adults’ browsing through March 2025 and found they clicked a traditional search result in 8% of visits where an AI summary appeared, against 15% where none did. The second — that buying AEO reliably converts into pipeline within a quarter — is the hype.
The defensible version is narrow: the traffic is small, it is growing, and the buyers inside it have already done their comparison shopping before they arrive. That is a case for measuring the channel deliberately, not for reallocating a search budget on a vendor’s slide. If a proposal promises a citation rate, ask which engine, which prompts and which window — then ask what happened on their worst run.
Doing it yourself: what the work actually costs in hours
The method is not secret and a competent in-house marketer can run it. Build a fixed prompt set of 20–40 buyer questions. Poll each engine on a cadence and score hits. Publish answer pages carrying a number an engine cannot get elsewhere — your own data, a worked calculation, a threshold table. Verify every external figure at source, regenerate schema, link pages into hubs both ways, and refresh anything stale.
| The task | Hours per month, in our own production | The thing that breaks at volume |
|---|---|---|
| Prompt set and per-engine polling | 2–4 manual, or one afternoon to script once | Answers vary day to day, so a spot check tells you nothing; the cadence has to survive a busy month |
| One properly sourced answer page | 4–8 per page, most of it verification | Source-checking is the step that gets dropped first, and a wrong number quoted by an engine is worse than no page |
| Refresh and internal linking | 6–10 once past ~50 pages | Nobody voluntarily audits page 180; decay is invisible until a citation quotes a superseded figure |
Four pages a month with the measurement layer attached lands between 30 and 45 hours — roughly a fifth of a full-time role, every month, indefinitely. Below four pages a month the cadence is too slow to produce a trend line you can read, and that is the real crossover point: run it in-house if you can defend those hours against everything else the role could do, hand it over if you cannot. Either way the decision belongs to gate 2, not to the pitch. The done-for-you version of the same cadence is set out in what an AI SEO and AEO service actually includes.
Frequently asked questions
Is AEO just SEO with a new name?
No — the difference is the measurement surface. SEO scores a URL’s position on a results page; AEO scores whether your brand appears inside a generated answer, per engine, against a fixed prompt set. The behaviour has genuinely shifted: the Pew Research Center tracked 900 US adults’ browsing in March 2025 and found they clicked a traditional search result in 8% of visits containing an AI summary, against 15% without one. The work overlaps heavily; the scoreboard does not.
Is AI SEO a scam?
The discipline is not, but two specific offers should end the meeting: a guarantee of placement inside an assistant’s answer, and a report that shows citations without naming the engine, the prompt sample and the date window. No vendor controls an engine’s ranking function. Ask for a worst-run number — ours is 1 of 15 tracked prompts on ChatGPT and 1 of 15 on Gemini on 11 September 2026 — and judge the answer by whether one exists.
Should I move budget out of SEO to fund AEO?
Not as a swap: most of the underlying work is shared and the volumes are not comparable yet. Conductor measured AI referrals at 1.08% of all website traffic across 13,770 domains between May and September 2025, highest industry 2.80%. Treat AEO as an extra measurement layer and a change in what you publish, then let gate 2 decide whether it earns incremental budget.
How do I measure whether AEO is working for my business?
Two meters, not one. Track share of answer — hits divided by tracked prompts, per engine, on a fixed cadence — and track AI landings in your own logs by matching the utm_source tag first and the referrer second, then deduping the union. On our logs the referrer alone missed 55% of AI landings in the 22 August to 5 September 2026 window, because app sessions arrive with no referrer.
Is answer engine optimisation worth it for a small business?
It depends on gross margin, not company size. A consultancy whose gross margin on one closed deal is worth more than a month of the work clears the break-even sum; a high-volume seller whose per-deal margin is a fraction of that monthly cost needs roughly ten times our measured landing volume and should not buy it yet. Size decides the budget; margin decides the answer.
Pay-Per-Result appointments
See if we’re a fit
We book qualified sales appointments for you and you pay on results, not retainers. Our booking page asks a few quick questions so you find out in two minutes whether that model suits your business.
- 50,769+ appointments booked without cold calling.
- Pay-Per-Result pricing — you pay for booked, qualified calls.
- Pick your own time on our live calendar, no phone tag.