SIGNALS
Buying AI visibility

Should you sign a 12-month AEO contract or go month to month?

The short version

Sign for about six months with a break clause, rather than twelve with none or one month at a time. The reason is arithmetic, not preference. A baseline has to be run twice before any work starts, to measure your own variance. Structural changes take two to six weeks to register in citation frequency. A movement only counts when it holds direction across three consecutive runs. And published decay rates are fast enough that a single reading describes a brand for about a month, with Trakkr Research's study of 108,650 citation URLs finding a median brand half-life of 31 days (2026). Add those up and a three-month contract ends in the month the first readable signal arrives, while month to month guarantees you will cancel before anything is legible. A twelve-month term is fine if it buys a break point and a price, and not if it only buys the agency certainty.

Should you sign a 12-month AEO contract or go month to month?

Sign neither as offered: a 12-month AEO contract is worth signing only with a break clause at the halfway point, and month to month is worth signing only if you accept that you will never read a trend. The useful question is not how long the agency wants, it is how long the measurement needs before it can tell you anything, and that length is knowable.

The industry answer to this question is usually a preference dressed as a principle. Agencies say results take time. That is true, and then they set the term at whatever suits their forecasting. Buyers counter with month to month, which feels prudent and quietly guarantees a decision made on activity rather than evidence, because the evidence is not available yet at month two.

There is a third way to set the term, which is to derive it. Work out how long the measurement takes to produce a signal you could defend in a board meeting, add the time the page work takes to register, and set the term at that number plus a review point. That calculation lands at around six months for most engagements, and the sections below show the arithmetic rather than asserting it.

One thing to settle before the term: what this is. A one-off project and an ongoing retainer are different purchases with different right answers on length, and our guide on whether AEO is a project or a retainer separates them. The length question below assumes you have decided you are buying a programme.

Why do most AEO agencies ask for 6 to 12 months?

Most AEO agencies ask for 6 to 12 months because that is the category's inherited norm from SEO retainers, and because the work genuinely does not produce a readable result sooner. The norm is explicit in the published guidance. One agency guide states that "Retainers typically require 6-12 month minimum commitments" and that "6-12 months is standard, but negotiate flexibility", adding that some agencies accept 90-day trial periods before longer commitments, in Stackmatix's guide to AEO services contracts (12 March 2026).

Treat that as market description rather than evidence. It is an agency writing about agency terms, which is the commercial interest you would expect, and the same page recommends 30 to 60 days notice with prorated refunds and flags harsh termination penalties as a red flag. The useful part is that the norm is openly stated, so you know what you are negotiating against rather than guessing.

The defensible half of the agencies' reasoning is that page work is slow to reach the engines, and the better evidence for that is not an agency blog. Chen and Koudas found that which pages the engine's own search retrieves explains about 32% of the variation in citations against 2.8% for page properties, in "What Drives AI Citations?" (arXiv 2610.15033, 10 October 2026). A page edit has to be crawled, re-indexed and then win a retrieval step you do not control, which is a slower path than a ranking change and a less certain one.

What is not defensible is a long term with no exit. The length that the work requires and the length that protects the agency from a bad first quarter are two different numbers, and a contract should name the first one. Our guide on how long AEO takes to work covers what each month realistically produces.

What is the shortest term that can show whether the work is working?

The shortest term that can show whether the work is working is about six months, and it comes from stacking four waiting periods that cannot be run in parallel. Each one is published, so you can check the arithmetic rather than take anybody's word for it.

Start with the baseline. A single reading cannot be a starting position, because the engines sample rather than rank once and hold, so the first measurement has to be run twice to size your own variance. How much variance there is depends on how often you are cited at all: BrightEdge put the stability inflection near 50 citations, where weekly volatility drops from 50% to 8%, and found frequently cited domains moving about 0.7% week to week against more than 50% for sporadically cited ones, in "AI Search Engine Citation Volatility: The 70x Stability Gap" (23 October 2025, read 17 November 2026). Below that threshold, two baselines are the minimum and three are better. Call it three to four weeks.

Then the work has to register. Our published methodology puts the lag between a structural change and a shift in citation frequency at two to six weeks, and treats a real move as the same direction across three consecutive monthly runs rather than any single rise. That is the step most contracts ignore, and it is three months on its own.

Then decay sets the floor under the whole cadence. Trakkr's citation decay study logged 108,650 citation URLs and 857,138 daily reports covering 10,991 brands across eight models, and found 73.5% of citation URLs appearing exactly once and not returning, with the median brand falling to half its highest observed day in 31 days, in "The Half-Life of AI Citations" (22 April 2026, read 17 November 2026). A reading describes a brand for roughly a month, so a quarterly report is three readings late. Baseline, lag, three confirming runs: six months, with the first two spent measuring rather than winning.

What can each contract length actually prove?

Each contract length buys a different class of evidence, and matching the term to the evidence you need is the whole decision. The table below sets out what is available at each length, on the cadence above.

Term What it can prove What it cannot
Month to month That the agency does the work it said it would Anything about citations. You will cancel before the second baseline
90-day pilot Where you stand, what is blocking the pages that matter, and whether the agency can diagnose Whether the fixes moved anything, which has not happened yet
Six months A baseline, a verified implementation, and a direction held across three runs Revenue attribution, and whether the gain survives
Twelve months Whether a win holds, how fast it decays, and the maintenance cost Little more than six months does, if there is no break clause to make month seven a decision

The honest reading of the fourth row is that twelve months earns its keep on durability rather than on results. Given a median brand half-life of about a month, the question of whether a page holds its place is only answerable over two or three quarters, and that is a real thing to buy. It is also the thing most likely to be sold to you as "results take a year".

The second row is the one most buyers should start with. A pilot scoped to produce a baseline and a diagnosis is a complete deliverable in 90 days, and our page on piloting AEO before signing a retainer covers what to ask for in it.

What should the exit terms say, whatever length you sign?

The exit terms matter more than the length, and they should cover three things: notice, money and data. Of the three, the data terms are the ones that decide what your next engagement costs, and they are the ones almost nobody negotiates.

On notice, 30 to 60 days is ordinary and workable. On money, ask for a prorated refund of anything prepaid and resist a clause that charges the remainder of the term on early termination, which converts a commitment into a penalty. On data, name the four artefacts in the contract: the prompt set as a file, the full run history with the prompt, engine, date and cited URLs for every run, the page change log by URL and date, and the written record of what was tried and did not work.

The reason to be specific about the run history is the decay figure above. Your measurement series is the only asset in the engagement that compounds, because a baseline you already own is six weeks a new vendor does not have to spend. A platform that shows you reports while the subscription is live and exports nothing when it ends has quietly made your history theirs. Our guide on switching AEO agencies without losing your progress covers the handover in detail, and what access an agency needs covers the revocation side.

Ask too what evidence you will have at the review point, and accept that in this category the published proof is thin. Ours is two anonymised early case studies, where one client's audit score went from 17 to 68 and another more than doubled, with the page stating that the audit score is an input and share of voice is the outcome. Hold any vendor to that distinction at the break clause: an input metric is the one that will have moved by then.

When is month to month the right answer?

Month to month is the right answer in three situations, and in all of them you are buying something other than a citation trend. Knowing which one you are in stops you paying for a measurement programme you will cancel before it reports.

The first is a defined piece of work with an end: a set of pages brought to a structural standard, a technical fix to rendering or crawl access, a migration checked before and after. That is a project, it has a deliverable, and a long term adds nothing. The second is a maintenance arrangement after a programme has finished, where the work is a quarterly check and a response when something slides. The third is a trial of the people rather than the method, which is a legitimate thing to want and should be described honestly as that.

Where month to month is the wrong answer is the common case: a first engagement bought in monthly increments by a buyer who intends to judge citations. On the cadence above, month two is the second baseline, month three is the implementation, and month four is the first reading that means anything. A buyer three months in with no movement and a monthly invoice will cancel, and the thing they cancelled was the measurement rather than the work.

If you want the first two months of that cadence done before you commit to anything, a free visibility assessment produces the baseline from outside your systems: your buying questions sampled across ChatGPT, Perplexity, Google AI Overviews and Claude, plus a read of your public pages as a crawler sees them. What the work costs after that, and what moves the figure, is on our pricing page rather than guessed at here.

What else do people ask about AEO contract length?

Buyers negotiating an AEO contract length ask these five questions most often.

What is the shortest AEO contract worth signing?

Around six months, and the reason is arithmetic rather than preference. Structural changes take two to six weeks to show up in citation frequency, a real move is a direction held across three consecutive monthly runs, and the baseline has to be measured twice before any work starts. Add those together and a three-month term ends in the month the first readable signal arrives.

Is a 12-month AEO contract unreasonable?

Not unreasonable, but it should buy you something. A twelve-month term is defensible when it comes with a break clause at six months, a price held for the term, and a stated review point where either side can walk. A twelve-month term with no break and an early termination charge is the agency's cash flow dressed as a methodology.

Should you ask for a 90-day trial instead?

Ask for a scoped pilot rather than a trial of the retainer. A pilot with its own deliverable, a baseline on your buying questions and a diagnosis of what is blocking the pages that matter, is worth running in 90 days. Three months of a retainer is not, because the measurement cannot report a trend in that window and you will be judging it on activity instead.

What should the exit clause say?

Notice period, prorated refund of anything prepaid, and above all what you take with you: the prompt set, the full run history behind every report, the page change log and the record of what was tried and failed. The data terms matter more than the notice period, because a vendor who keeps the measurement history makes your next engagement start from zero.

Does a longer contract get you a better price?

Often, and it is a fair trade if the break clause survives the discount. The thing to avoid is paying for a year in exchange for a rate and discovering the discount was conditional on there being no way out. Ask for the price with and without the commitment, so the discount is a number you can weigh rather than an implication.

Related guides

The Assessment

Get the baseline before you commit to a term.

A free visibility assessment samples your buying questions across ChatGPT, Perplexity, Google AI Overviews and Claude, and reads your public pages the way a crawler sees them.

Request a free visibility assessment →
SIGNALS · A BlackSig Systems company