How Long Does AEO Take to Work?

Most practices see AI crawler activity rise within weeks of the structural work going live, and movement in whether AI assistants actually name the practice within two to four months. The exact timing depends on the engine, the size of the market and how much needs fixing at the start. Here is what happens at each stage, with real timelines from MaxConnex engagements.

Weeks 1 to 2: the baseline

Every engagement starts with an audit. We ask ChatGPT, Google AI Overviews, Perplexity, Gemini and Claude the questions your patients ask, record whether your practice is named and who is named instead, and keep those exact questions as the benchmark for every later report. The other job in these weeks is access to the website, Google Business Profile and analytics. In practice, access is the single most common delay in the first month.

Weeks 2 to 6: the structural work

This is the work that makes a practice verifiable: structured data for every location, practitioner and service, a complete Google Business Profile, listings on directories corrected so they agree with each other, and pages that answer patient questions in their first sentences. For a single-location practice with a usable website this usually takes a few weeks. A multi-site group, or a website AI cannot read, takes longer.

Weeks 4 to 8: the crawlers arrive

The first measurable change is usually AI crawlers fetching the site more often, because there is now more for them to read. For a fertility group across Melbourne and Sydney, AI crawler fetches went from 32 a month in June 2026 to 2,241 in August, and crawler activity rose 528% from July to August on a like-for-like basis. Crawler activity is not the goal, but it is the earliest sign the work has been picked up. Read the fertility case study.

Months 2 to 4: named in the answer

Being named in the answer takes longer than being read. Assistants that search the live web for each answer, such as Perplexity and Google AI Overviews, tend to move first. Answers drawn from a model’s built-in knowledge change only when the model is updated.

The commercial result follows the recommendations. For a single-site GP and skin cancer clinic, AI-attributed bookings went from 10 a month in May 2026 to 109 a month in September, with 120 bookings attributed to ChatGPT specifically over that period. Read the GP clinic case study.

After that: holding the position

AI assistants keep re-checking what they can verify. New reviews, current hours, newly published practitioners and fresh answer-led content all keep a practice in the answer, and stale information can cost it the position. That is why AEO is reported monthly against the same questions rather than treated as a one-off project.

What changes the timeline

• Market size. A small market such as Hobart, where an answer names one or two practices, can move faster than a contested Sydney specialty.
• Starting point. A website that AI can already read needs structure added. One it cannot read may need a refresh first.
• Consistency of listings. Directories that disagree with your website take time to correct, and each correction has to be picked up.
• Number of locations. Every location is its own entity with its own listings and tracked questions.

How to find out where you stand

The free AI visibility audit gives you the baseline: whether the major AI assistants name your practice today, and who they name instead. If you want to see what the work costs at each size of practice, our pricing is published.