A perception is a claim that answer engines make about your brand, for example “expensive”, “best for small teams”, or “hard to set up”. The AI Perception Summary shows what they say now. The monitor lets you manage this claim over time. You do not only read the summary and hope that a claim disappears. You put the claim under observation, and Genezio continues to check it.

Why you monitor a perception

Some claims are much more important than their frequency shows. A single answer that calls your brand “hard to integrate” can cost more than ten answers that call it “reliable”. The reason is that this claim appears exactly where a buyer is not sure. The monitor tells you if your work really changed what the answer engines say. For example, this work can be the documentation that you wrote again or the comparison page that you published. A claim that you want to be true is also worth a monitor. For example, you want your brand to be the option for regulated industries. In this case, you want to know when answer engines start to say this without help.

How to monitor a perception

  1. Open Monitor perceptions for your brand.
  2. Select a perception, or write a perception of your own.
  3. Come back after your next content release, or select Refresh to check again immediately.
If you monitor no perception yet, the page shows the six claims that answer engines state most often about your brand. Select one of them to start. The page uses the filters of your report, such as the topics and the models. A perception can have more than one monitor. Each monitor keeps its own filters. For example, you can watch the same claim for one topic and for all topics. Genezio checks each monitored perception again each day. Thus, you do not have to select Refresh to see new results. Each check reads only the new answers since the last check. Thus, a large brand gets its result fast. Genezio keeps each check. Thus, a tracked perception has a history: what the answer engines said, and when it changed.

Read the result

A tracked perception gives you a verdict, not a general impression. The verdict tells you clearly if the claim agrees with what the answer engines say about your brand now. A summary alone cannot give you these two results:
  • Proof that your work had an effect. For example: “We wrote the onboarding again in March, and the ‘hard to set up’ claim stopped in April.” You can show this sentence to a board.
  • An early alert about a reversal. A claim that you corrected can come back. Usually, the cause is a new source that answer engines start to cite. With the monitor, you find the problem in weeks, not at the next quarterly review.

See the agreement over time

The page of a monitored perception has a chart of the agreement over time. For each period, one band shows the conversations that agree with the claim. A second band shows the conversations that disagree. A period with no disagreement shows no second band. The chart uses days, weeks, or months, depending on the range of the dates. Thus, you see if the models agree more than last month, and if an answer that disagrees is new. When the monitor has more than one model, select a model to see its own chart.

See the sources behind the answers

The page of a monitored perception shows which sources make the models agree and which sources make them disagree. Each side has its own list. The card Sources behind the answers shows sites or pages. Use the switch of the card to change the view. For each source, the list shows how many answers of that side cite it, and their share. Click a source to open a wide panel. The panel shows the pages of the site and the answers that cite it. You can see all the answers, or one group for each answer engine. In the table of the paragraphs, each row also shows the domains that the model cited for it. Click a domain to open the page. Use this information to decide where to work. A site that causes disagreement is the first site to correct or to contact. A site that causes agreement shows you the sources that already help your brand.

Read the perceptions from a different tool

Tracked perceptions and their verdicts are available over MCP and through the API. Thus, a perception check can be part of a larger report. Nobody must remember to open it.