XLR8 AI now measures brand resilience across repeated AI answers
AI models rarely give the same answer twice. A brand that appears in one response can be missing from the next, so a single check only tells you what happened on one run. XLR8 AI now runs each prompt up to five times, so you can see how consistently your brand and your sources appear, not just whether they showed up once.
What's new
Multiple runs per prompt: run every prompt between one and five times. Each run is analyzed, so you get a pattern instead of a snapshot.
Brand resilience: for each prompt and model, see how many of the runs included your brand. Five out of five means it showed up every time. Two out of five means it usually did not.
Brand stability: a percentage showing how much the set of brands changes between runs. If the same brands came back every time, stability would be 100%.
Source stability: the same measure for cited sources, so you can tell which sources are cited consistently and which appear once.
Shaded brand tags: brands in the results are shaded by how many runs returned them. The darker the tag, the more runs it appeared in.
Why it matters
A single run can make a brand look visible when it is not. If your brand appears in two of five runs, a customer asking that question would usually not see you, even though one lucky run would have counted as a win.
We saw this in our own test: 25 prompts across ChatGPT, Microsoft Copilot and Perplexity, run five times each, for a well-known apparel brand. It appeared in at least one run for 28% of prompt and model combinations, but in all five runs for only 8%. Depending on which single run you looked at, its visibility read anywhere from 13% to 20%.
How to use it
Find prompts where you are close. Brands that appear in one to four runs out of five are the nearest wins.
Let source stability guide your strategy. A source cited in every run, including a third-party site, is doing something that works. Study it and pursue similar placements.
Re-run and compare. After you publish or earn new placements, run the experiment again and see whether your brand moves toward five out of five.
Availability
Set the number of runs per prompt, from one to five, in your experiment's settings. More runs give a steadier read but use more compute, so use five for the brands and prompts that matter most.
See it for your brand
Want to see how consistently AI models mention your brand? Request a free AI visibility report.
