A finance team I worked with ran the numbers on a quarter of AEO work and found the awkward truth at the center of this whole discipline: roughly a third of the new customers who said an AI engine pointed them to the company had left no trace in the analytics at all. No referral string, no UTM, no click that a dashboard could see. The engine had cited the company, the buyer had read the answer, and then they had typed the brand name straight into a browser or picked up the phone. The work produced revenue. The standard reporting showed almost nothing. That is the problem you have to solve before you can measure AEO ROI honestly, and pretending it does not exist is how most teams either overclaim or give up.
AEO does not behave like the channels finance is used to. Paid search hands you a click and a cost per acquisition. SEO hands you a ranked page and a session count. Answer engine optimization frequently hands you a citation inside a conversation you cannot see, that influences a buyer who then converts through a channel that gets all the credit. Measure it with click-based tools alone and you will conclude it does not work, right up until you lose the citations and watch the unexplained pipeline dry up. The fix is a measurement model built for influence, not just clicks, and it rests on six figures that hold up in a real budget conversation.
Why click attribution breaks for AEO
The core mechanic of AI search is the zero-click answer. A buyer asks an engine a question, the engine synthesizes a response and names a few sources, and the buyer gets what they needed without visiting anyone. When they do act, the action often shows up as direct traffic or branded search, channels that classic attribution treats as if the demand appeared from nowhere. The influence was real and the credit goes to the wrong place.
This is not a flaw you can engineer away with better tracking pixels, because the engine controls the surface and does not always pass a referrer. So the honest response is to stop trying to force AEO into a click model and instead build a model that triangulates influence from several independent signals. No single one of them is airtight. Together they form a picture defensible enough to put in front of a CFO, which is the actual bar. You are not proving a click. You are proving that demand moved in response to the work, using the same kind of triangulation finance already accepts for brand and PR.
Figure one: citation share for priority questions
The first and best leading indicator is how often the major engines cite you when asked the questions your buyers actually ask. Build a list of the twenty or thirty queries that matter, the ones a prospect would type when researching a purchase in your category, and check across ChatGPT, Perplexity, Gemini, and Claude how frequently you appear in the answer. That percentage is your citation share, and it is the closest thing AEO has to a ranking.
Citation share matters because it moves before revenue does. When your share on priority questions climbs from near zero to a meaningful fraction, the pipeline impact follows a quarter or two later. Tracking it monthly gives you an early signal that the work is landing, long before the financial results confirm it. It also localizes the problem. If your share is high on some questions and absent on others, you know exactly where the content and entity work needs to go next.
Figure two: branded search lift
When AI engines start recommending you, more people go looking for you by name. That shows up as a rise in branded search volume, the count of searches that include your company or product name. It is one of the cleanest proxies for AEO influence, because a buyer rarely searches a brand they have never encountered. Something introduced them, and increasingly that something is an AI answer.
Pull branded search volume from your search console and from a keyword tool, plot it over time, and watch what happens as your citation share grows. A correlated rise is strong evidence that the citations are driving discovery. It is not proof in a courtroom sense, but in a business sense a sustained branded-search lift that tracks your AEO work is exactly the kind of leading-to-lagging relationship finance trusts when judging brand investments.
Figure three: direct and referral traffic shifts
Some AI engines do pass referral traffic, and Perplexity in particular often sends visible visits with an identifiable source. Track that referral traffic as a direct, countable slice of AEO impact, the part you can actually see. Then watch direct traffic alongside it, because the invisible citations tend to surface as direct visits when buyers act on an answer without clicking through.
A pattern worth watching: a step-change in direct traffic that lines up with the appearance of new citations, with no other campaign running to explain it. On its own, direct traffic is noisy. In the context of rising citation share and branded search, a direct-traffic jump becomes corroborating evidence rather than a coincidence. The discipline is to read these figures together. Any one of them can be dismissed. The three of them moving in concert are hard to argue with.
Figure four: self-reported attribution
The most honest attribution data often comes from simply asking. Add a short how did you first hear about us field to your demo forms, your signup flow, and your onboarding calls, and let buyers tell you in their own words. A meaningful share will name an AI engine directly, and that self-reported signal catches exactly the conversions the analytics miss.
Self-reported attribution has a bias problem, people forget and misremember, so do not treat it as exact. Treat it as a sample. If one in five new customers volunteers that an AI engine recommended you, that is a number you can extrapolate against your total new-customer count to estimate AEO-influenced revenue. It is rough, and it is also the figure that most directly answers the CFO’s real question, which is whether this work brings in customers who pay.
Figure five: pipeline and revenue influenced
Now connect the influence signals to money. Take the customers who self-report AI discovery, add the ones whose first touch was a trackable AEO referral, and apply a reasonable share of the unexplained branded and direct demand that rose in step with your citation growth. The result is an influenced-pipeline figure: the revenue that AEO plausibly touched, stated as a range rather than a false-precision single number.
Stating it as a range is what makes it credible. A claim that AEO drove exactly $412,000 invites a fight over the decimal. A claim that AEO influenced somewhere between $300,000 and $500,000 in pipeline, built from three independent signals, invites a conversation about scaling the investment. Finance respects an honest range built from triangulation far more than a precise number built from a single shaky source. The goal is a figure you can defend, not a figure you have to retract.
Expect a lag between citation and revenue
One thing trips up teams measuring AEO for the first time: the citation share moves months before the money does, and impatient leaders kill the program in the gap. When your share on priority questions climbs, the buyers who see those answers are often early in their research, weeks or quarters from a purchase. The leading indicator improves, the lagging one stays flat, and a team that only watches revenue concludes the work failed right as it was about to pay off.
Manage that lag by reporting the leading and lagging figures side by side, with the expected delay stated plainly. Show the CFO that citation share rose this quarter, branded search followed a few weeks later, and influenced pipeline is now beginning to move, and frame the sequence as the normal shape of the curve rather than a disappointment. This also protects you in the other direction. If citation share is climbing and branded search is not following after a reasonable lag, that is a real signal something is wrong, maybe the citations are on the wrong questions, or the buyers seeing them are not your buyers. Reading the figures as a sequence with an expected delay, rather than demanding they all move at once, is what keeps a working program funded through the quarter where the leading indicators are the only proof you have.
Figure six: cost against the influenced return
There is one more discipline before the final number, and it is the one that earns finance’s trust permanently: count the full cost honestly, including the parts teams like to hide. The content and the tooling are obvious. The staff hours spent planning, the entity cleanup, the time your subject-matter experts gave to reviews, all of it belongs in the cost line. A ratio built on a lowballed cost looks great for one quarter and destroys your credibility the moment someone notices the missing inputs. An honest cost makes a lower ratio that survives scrutiny, and a number that survives scrutiny is worth more than a flattering number that does not.
The final figure is the one that closes the loop. Total what you spent on AEO over the period, the content, the entity work, the tooling, the agency or staff time, and set it against the influenced revenue range. That ratio is your AEO ROI, and stated honestly with its assumptions visible, it is what justifies the next quarter’s budget or kills it.
Run this same six-figure model every quarter and something useful happens. The picture sharpens as you accumulate self-reported data and watch the leading indicators predict the lagging ones. You stop arguing about whether AEO can be measured and start arguing about how much more of it to fund, which is the conversation you actually want to be having. The teams that win the budget fight are not the ones who claim the cleanest number. They are the ones who show their work, name their assumptions, and present a defensible range built from independent signals that point the same direction. Finance has seen enough inflated marketing claims to distrust precision it cannot verify, and a CFO will fund an honest range with visible logic long before a suspiciously exact figure with no method behind it. Measure AEO ROI this way, with influence triangulated instead of clicks demanded, and the work that was invisible to your dashboards becomes a line finance can see, trust, and grow.