Your first run gives you a score. The score is the least useful part of the run.

​
Read answers before numbers

Open five or six conversations and read them. In ten minutes, you will know more than the dashboard can tell you:
  • How answer engines describe your category
  • What answer engines cite
  • Which competitors answer engines take seriously
  • How answer engines describe your brand when it appears
Teams that do not do this step use the next month to interpret a number that they do not understand.

​
Then look at three items

Where you are absent. Topics with a score of zero are the clearest result. Answer engines answer that question without you. The conversations show which brands they use in the answer. Where you appear but answer engines do not recommend you. Answer engines name you, but they select a different brand. This is a positioning problem, and the answers usually tell the reason in clear words. What answer engines cite. These are the sources that have the most effect in your category. This list is the most actionable list in the product. Refer to Most cited sources.

​
Conclusions to avoid now

Do not benchmark the main score. A score of 40% has no meaning until you see the scores of your competitors on the same questions. Do not use one run as a trend. The answers of answer engines change from one run to the next. You need some runs before a direction is real. Refer to Why do LLM results change?. Do not try to correct all problems. The first run usually shows more gaps than a team can work on. Select the topics with real commercial importance, and ignore the other topics for now.

​
A good first action

Select one topic where a competitor wins. Read what the answer engine cited. Then decide if you can be a better source for that question. This is one week of work with a measurable result, and it is better than a strategy document.