Knowledge
Why aren’t the changes visible straight away?
Improving visibility in AI It takes between 2 and 4 weeks for changes made to your own website, and up to several months or even 12 months for factors outside your website, such as reviews, round-ups and publications. There are four overlapping delays between the publication and the model’s response: the bot must return to the page, the content must be added to the dataset from which the model constructs its responses, and mentions of the brand they have to be generated first. The process of reloading the page alone takes according to Google from a few days to a few weeks. That is why the effect is not visible on the day of publication, and the first meaningful reading is usually the second full scan.
Author: Patryk Pietrzyk · Last updated: 25 October 2026 · Reading time: approximately 6 minutes
Mechanism
What delays are there between a post being published and a response from the AI chatbot?
Four: publication, the robot’s return visit, the inclusion of the content in search results, and the emergence of backlinks from outside your domain. Each of these factors is minor on its own, but together they result in a timeframe measured in months rather than days.
01
Publication
The change is already on your website. The model doesn’t know about it yet, as she doesn’t check it regularly.
Days
02
A return visit
The robot must return to the site and download the new version. How often you do this depends on how frequently the website changes.
Days to weeks
03
Inclusion in the replies
The new content must be sent to the set from which the model constructs its responses. Not every download necessarily means that.
Weeks
04
Signals from outside your domain
Reviews, mentions and rankings must first be created, then discovered, and only then do they begin to carry weight. This is where the biggest delay lies.
Months
The fourth point is the most important. The first three relate to your website and can be completed within a few weeks. The fourth concerns what others say about you, and it is this one that decides on the recommendation.
Dates
How long does it take for vision to improve in AI after each type of work?
From 2–4 weeks for updates to an existing subpage, up to 3–12 months for publications and presentations. The bars show not only the duration but also the start date. Work on your own website starts to take effect immediately, whilst external signals only take effect once they have been generated.
Amendment to an existing subpage
2–4 weeks
New subpage for the query
4–8 weeks
Reviews on independent websites
1–3 months
Publications and presentations
3–12 months
03 months.6 months.12 months.
Indicative range, based on observations from our scans. The actual rate depends on the category and on how often the model bots visit your website. For the lime manufacturer, the first recommendation appeared after 11 weeks.
That is why first report After a month, it usually shows activity only in content-related metrics, whilst visibility in category-related queries remains static. This isn’t a lack of effect, just a different curve. We describe it in the article known, but not recommended.
Earlier indications
What can be seen in the report before visibility improves?
It usually grows first domain citation index and audit rating, and only then does visibility in category-based queries come into play. The work therefore yields results sooner than the single figure at the top of the report suggests. You just need to know where to look.
First scan after implementation: what usually works and what doesn’t
↑
Domain citation index
The model starts to refer to your website, even if the brand doesn’t yet appear in the response.
↑
Audit rating
Technical and editorial changes are immediately apparent, as the audit assesses your website directly.
→
Ranking amongst the brands listed
Sometimes a brand moves up from tenth place to fifth, yet still fails to make it into the top three.
–
Visibility in category-based queries
Main frame. It usually remains there for the first one or two scans, and that’s normal.
This is the practical reason why The report breaks the result down into several metrics rather than a single number. With one number, the first two months look like no progress at all. The Click Made graphic design studio case shows this well: website citations grew in every scan from July, while recommendations only changed after five months. We describe the whole case in the article AI chatbot recommendations after 5 months: the Click Made case.
Practical experience
How can you avoid giving up on improving your visibility in AI too soon?
Measure the same thing every month, analyse the results by metric, and carry out work both on and off the website in parallel. The greatest risk isn’t a lack of results, but giving up in the third month, just before external signals start to carry weight.
Measure the same thing every monthRhythmThe same set of questions, at the same time. Without this, each measurement measures something slightly different and the conclusions are arbitrary.
Look at the breakdown, not just a single figureReadingCitations and the audit gets underway before visibility. It is these that determine whether the work actually reaches the models.
Do two things at onceRangeContent on your own website and one operation outside it. In that case, the shorter and longer curves rise in parallel, rather than one after the other.
In this article, we explain how often it makes sense to take measurements and why once a month is the ideal frequency how often to measure visibility.
Please note
When is the lack of results not simply a matter of time?
When no metrics have changed over four scans, when all the work is carried out on your own domain, when tasks are only half-completed, or when the set of questions has changed.
Four scans with no movement in any of the recordsIt’s not just visibility that’s at a standstill, but also citations and audits. This is a sign that the changes aren’t being reflected in the models, for example, by means of a robot block.
All work on your own domainThree months of articles without a single action taken off the website. In that case, the lack of results is to be expected – it’s no surprise.
Tasks only partially completedStarted but not finished: a profile has been created but is empty; a subpage has been published without any specific details. The model treats this as non-existent.
Revised set of questionsIf the queries have changed between scans, the two results are not comparable and conclusions regarding the absence of an effect are unfounded.
FAQ
Questions about time
The questions we hear most often between the first and second scans.
Why aren’t the changes visible straight away?
This is because there are four delays involved: publication, the crawler’s return visit, the inclusion of content in the results, and the generation and detection of signals from outside your domain. The last of these takes the longest.
How long does it take to see the first results?
Between 2 and 4 weeks for updates to existing subpages. Backlinks from outside your own domain take months to take effect, so the first meaningful reading is the second full crawl.
We’ve done everything we can, and the result stands. What next?
Check the breakdown by metrics. Citations and the audit start earlier than visibility. If they’re also at a standstill, the problem lies elsewhere than in timing.
Will measuring more often speed up the results?
No. Taking measurements doesn’t change anything, and more frequent scans produce noise rather than a trend. We discuss this in the article how often to measure visibility.
Is there any way to speed this up?
To some extent. It’s quicker to work on your own website and gather reviews. There’s no way to speed up inclusion in rankings and publications beyond the pace at which their authors respond.
Why did one model detect the change, whilst the others did not?
The models differ in how often they draw on fresh sources. In our scans, Perplexity usually responds the quickest, as it most frequently uses current search results.
How can I check whether AI chatbots have access to the website?
Check the robots.txt file: ensure it does not block the OAI-SearchBot (ChatGPT), Claude-SearchBot and Claude-User (Claude), and PerplexityBot. Once you have updated the robots.txt file, it may take OpenAI and Perplexity up to 24 hours to reflect the changes.
Will submitting the website to Google speed up the process?
Not directly. Google states that it takes anywhere from a few days to a few weeks for a page to be re-crawled, and repeatedly submitting the same URL does not speed up the process. A sitemap can help when there are many new subpages.
What sources is this text based on?
- Google Search Central: request for re-indexing: It takes anywhere from a few days to a few weeks to re-download the page.
- OpenAI: bots and their user agents: Changes to the robots.txt file take effect after around 24 hours.
- Anthropic: the Claude-SearchBot and Claude-User bots: blocking them may reduce their visibility in Claude’s results.
- Perplexity: how to get started: real-time search with sources.
- Ahrefs: brand mentions and visibility in AI Overviews (May 2025).
- Aelo measurements: monthly scans of clients’ brands in 2026; see the ‘Dates’ section for the time range.
Have a look at what’s moving underneath
Free Account No card required: you get the following to start with 250 AeloCoins and a full report on your brand’s visibility on ChatGPT, Claude, Gemini and Perplexity. The results are broken down into several metrics, so you can see changes even when the main figure remains the same. Plans and modules you’ll find in pricing.
