One AI chatbot instead of four: the Click Made case, October

Below, we describe the fourth report clickmade.pl and what visibility based on a single model means. We described the July measurement in the article High score, zero mentions: the Click Made case, July 2026, and the whole case starts with the article A graphic design studio in AI chatbots: the Click Made case, May 2026. We describe the late-October result in the article AI chatbot recommendations after 5 months: the Click Made case.

What happened in August and September?

We were refining the set of questions and the website. The 2 September report had no new scan, because new questions had been added to the list and were to be included in the next run. It did show the result of the technical audit: the website got an A rating both for AI readiness and for classic SEO. Brand recognition did not change, so neither did the direction of work: independent confirmation off-site.

During this time we were implementing further on-site tasks from the plan and creating directory listings for the company. What an audit is and what its rating says is described in the entry visibility audit.

What did the 3 October measurement show?

Full presence in the topic. Click Made appeared in answers to all the questions examined. Average visibility was 22%, the citation rate 32%, and 95% of the models’ answers gave a source. Local questions were the strongest, with a score of up to 37.5%. General questions, in which a customer compares different ways of getting a service done, turned out to be the weakest: there the brand appeared in only 13% of answers.

22% is lower than July’s 47%, but it is not a drop. In July the high average came from two models, and the brand was missing from most answers. In October the brand was present for every question, just rarely in the top positions. That is why we compare mentions and citations, not the average alone. Why individual results fluctuate is explained in the article Why does AI give a different answer every time, and what does this mean for measurement?.

Why was Perplexity the only one to see the brand?

Because Perplexity searches the web for every question and builds its answer from current pages. Once clickmade.pl had service pages with specific information, Perplexity could find and cite them straight away. The other models use search less often or differently, and rely more on what they already know about the brand. How we scan models with web search on is described in the article Measurement with web search: how we scan AI chatbots.

On top of that came a change at OpenAI: on 7 and 8 October, a few days after this measurement, ChatGPT switched to GPT-6. ChatGPT results from before and after that date therefore have to be read separately, as we describe in the article GPT-6 in ChatGPT: what it means for brand visibility.

The report rated this situation as a risky concentration. If three of the four main AI chatbots leave the brand out, most customers will not see it, even when the result in one model looks great.

What do 32% citations mean?

That the website had become a source. In seven questions, the model cited clickmade.pl as the source of the information, whereas in May and early July it did not do so even once. As well as clickmade.pl, the models cited major websites offering tools for creating web pages and graphics, as well as local directories and service comparison sites.

A citation is not yet a recommendation. The report stressed that the problem is not the number of citations but the lack of independent rankings and comparisons that would lead models to recommend the brand. What works when earning recommendations is described in the article How to get AI to recommend your business: what works in 2026.

What are we implementing following this report?

All the on-site tasks. Existing service pages get sections answering general and comparison questions, and the page about the author and the sales page are being strengthened. We do not create new subpages on topics that already have their own page, because that scatters the signals. The concept these improvements relate to is described in the entry search intent.

Off-site, we keep creating and tidying up directory listings for the company. Rankings and media publications are longer-term work, spread over the coming months.

What does the fourth measurement tell us?

• The average can drop while the situation improves. Presence for every question is worth more than a high average from two models.
• Models with web search react first. Perplexity noticed the changes on the website sooner than the other AI chatbots.
• A citation is a stage, not the goal. The website is a source, but the models do not recommend the brand yet.
• General questions are the weakest: in them the customer compares ways of getting the service done, not companies.

Methodological caveats

We refined the set of questions along the way, and in September it changed enough that the report did not compare the results with July. The number of models checked also changed: in October these were the four main AI chatbots. That is why we compare the direction of change and which models mention the brand, not differences in points. Details are on the page Methodology.

How does Aelo help you move beyond one AI chatbot?

Aelo shows the result separately for each model, so you can see right away when all the visibility rests on one AI chatbot. The measurement is done by Visibility in AI, the work plan is drawn up in Tasks, and content is prepared with the help of Studio. We describe the whole process step by step in the article How Aelo works: measurement, diagnosis and a step-by-step plan.

Small businesses will find more on the page Aelo for small businesses, personal brands on the page Aelo for personal brands, and agencies that measure their clients, on the page Aelo for agencies. You can take your first measurement on the Free account, without a card.

Try it for free

FAQ

Sources

• Aelo, visibility reports
• OpenAI, GPT-6 and Intelligent UI for everyone (7 October 2026): https://openai.com/index/gpt-6-for-everyone/
• Aelo, a methodology for measuring visibility in AI: https://aeloapp.io/wiedza/metodologia/