# How Long Does AEO Take to Show Results? > AEO results do not arrive on a fixed schedule. What the evidence says about how long new pages take to be cited, why early numbers mislead, and what to track. Page: https://trylectern.com/blog/how-long-does-aeo-take By Edgar Li, October 30, 2026. There is no published timeline that holds across brands, assistants and tactics. What the evidence does support is narrower: a brand-new page is cited less often than an older one for roughly its first month, answers shift on their own from day to day, and [the only field study with untreated comparison pages](https://arxiv.org/abs/2606.04362) found an effect that its authors called "suggestive, not conclusive". Plan in quarters, not weeks, and set up the measurement before the work so you can tell a result from noise. ## Key Takeaways - Do not promise a number of days. Nobody has a controlled dataset that supports one. - New pages are cited less in their first month. [One ChatGPT study](https://www.airops.com/report/the-fan-out-effect-what-happens-between-a-query-and-a-citation) found 25.3 percent of pages under 30 days old were cited, against 32.8 percent at 30 to 89 days. - Fixes you control, such as unblocking a crawler, can show up within days. Authority built through third-party coverage takes longer and is less predictable. - Platforms change underneath you. A movement that starts on a platform change date is not a result. - Baseline first, then change one thing at a time, then compare against pages you did not touch. ## Three Kinds of Work, Three Clocks "AEO" bundles work that runs on very different clocks. **Access fixes: days.** If an assistant's crawler is refused by robots.txt or a firewall, removing the block can change what the assistant can read as soon as its cache turns over. [OpenAI says](https://developers.openai.com/api/docs/bots) robots.txt changes can take about a day to be picked up, and [Perplexity gives a similar figure](https://docs.perplexity.ai/docs/resources/perplexity-crawlers). A fix like this is also easy to verify, because you can check the cause directly. See [AI crawler blocking](https://trylectern.com/blog/ai-crawler-blocking). **On-site content: weeks.** A new or rewritten page has to be crawled, indexed and then retrieved for a question. In one ChatGPT study from [AirOps](https://www.airops.com/report/the-fan-out-effect-what-happens-between-a-query-and-a-citation) (April 2026), pages under 30 days old were cited 25.3 percent of the time and pages 30 to 89 days old 32.8 percent. The rate then stayed flat out to two years (see [does updating old content help AI citations](https://trylectern.com/blog/does-content-freshness-affect-ai-citations)). One reading: give a new page about a month before you read its citation rate. **Third-party authority: months.** Assistants mostly cite sites you do not own. In [Profound's 2026 analysis of 11.84 billion citations](https://www.tryprofound.com/blog/where-do-ai-citations-come-from), roughly 57 percent went to what the study calls brand sites, which includes any company's site rather than only the brand being asked about. In other datasets the brand's own pages are a small share of citations on category questions (see [mention vs citation in AI answers](https://trylectern.com/blog/mention-vs-citation-in-ai-answers)). Getting named on pages that assistants already read means earning or arranging coverage, and that depends on other people's calendars. ## Why Early Numbers Mislead **Answers vary on their own.** [SparkToro and Gumshoe](https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/) found in January 2026 that two runs of the same prompt almost never returned the same brand list. A rise after two weeks can be nothing but sampling. By the standard 95 percent binomial formula, the margin for a mention rate built on five answers is roughly plus or minus 44 points near a 50 percent rate, and it narrows slowly as you add questions. [Evertune](https://www.evertune.ai/resources/insights-on-ai/how-many-repetitions-of-a-single-prompt-does-it-take-to-understand-your-ai-visibility) shows the same shrinking margin on real prompts. **Platforms change.** On 8 August 2026 ChatGPT's search behaviour changed in a way two vendors, [Searchable](https://www.searchable.com/blog/chatgpt-research-depth-study) and [Promptwatch](https://promptwatch.com/data/chatgpt-site-operator-fanouts), measured independently, with `site:` queries going from under 1 percent of ChatGPT's internal sub-queries to about 17 to 18 percent. Reddit's share of ChatGPT citations moved sharply across the same weeks (see [does Reddit affect ChatGPT answers](https://trylectern.com/blog/does-reddit-affect-chatgpt-answers)). If your numbers shift on a date like that, the platform moved. **Everything rises together.** The one field study with untreated pages ([Watanabe and Nakayashiki, arXiv 2606.04362](https://arxiv.org/abs/2606.04362), June 2026) found that ChatGPT referrals to the treated pages grew 6.1 times between January and May 2026. Untreated pages on the same site grew 3.5 times. The net effect was 1.82 times, with a confidence interval from 1.31 to 2.54, and a placebo test gave p = 0.16. The authors call it suggestive. A vendor case study would have quoted 6.1 times. ## A Reasonable Schedule This is a way to structure the reading, not a promise of what you will see. - **Week 0: baseline.** Fix the question list. Record how often each assistant names and cites you, and your competitors, over enough answers to read. [AI share of voice](https://trylectern.com/blog/ai-share-of-voice) explains how. - **Weeks 1 to 2: access and facts.** Remove blocks, correct wrong facts on pages you control, and re-check. These changes are cheap and verifiable. - **Weeks 3 to 6: new and updated pages.** Publish, then leave them alone for a month before judging. - **Day 60 to 90: first honest read.** Compare against the baseline using the same questions and assistants, and compare pages you changed with similar ones you did not. - **Quarter two: authority work.** Third-party coverage and directory presence need longer, and are read on the same panel. ## What Would Be Reason to Worry - Nothing moves at 90 days and the access checks were never run. Fix access first. - Mentions rise in one assistant and fall in another. That is a panel finding, not a failure. Read each assistant separately. - You changed five things at once. You will not know which one worked, if any. ## Setting Expectations Honestly AI assistants do not publish a webmaster console that reports when they re-read your page, so progress is inferred from repeated sampling. A few lab studies, such as [Vishwakarma et al.](https://arxiv.org/abs/2605.25517) and [Aggarwal et al.](https://arxiv.org/abs/2311.09735), say content edits can change how a page already in a model's context is used. Nobody has shown a technique with a stable, repeatable effect on being found across platforms. Be wary of anyone who gives you a date. Lectern runs the sampling on a schedule so the baseline exists before the work starts. The free Hobby plan checks 10 buyer questions weekly across ChatGPT, Gemini, Perplexity and Google AI Overviews, and the Lead plan refreshes daily; see [pricing](https://trylectern.com/pricing). For the broader approach, read the [AI visibility guide](https://trylectern.com/blog/ai-visibility-guide).