The win/loss matrix is the most actionable part of your visibility results. While the leaderboard tells you your overall visibility score, the matrix tells you exactly where you’re winning and losing, broken down by every prompt and every AI engine. This is where you find the specific gaps that GEO content can fix.
How to read the matrix
The matrix is a grid with your tracked prompts as rows and the four AI engines as columns: GPT, Claude, AIO (Google AI Overview), and PLX (Perplexity).
Each cell shows one of three things:
- ✓ — your brand was cited in the AI’s response to that prompt
- ✗ — your brand was not cited
- — — only appears in the AIO column. It means Google didn’t generate an AI Overview for that particular prompt. This isn’t a loss, it just means the query didn’t trigger the feature.
At the bottom of each engine column is a win rate, for example 3/5, showing how many of your prompts that engine cited you for. The colour indicates performance:
- Green means 60% or above
- Amber means 30–59%
- Grey means below 30%
What patterns to look for
Rows with all ✗
A prompt where no engine cites you is your clearest content gap. It means no AI tool is associating your brand with that question at all. These are your highest-priority prompts to write content for. The opportunity is essentially unclaimed as far as your brand is concerned.
Rows with mixed results
A prompt where you win on some engines but not others suggests your content exists and is being picked up, but isn’t strong enough or well-known enough to be consistently cited. This is a refinement opportunity. Improving the depth, structure, or citation-worthiness of an existing article is often enough to convert these ✗ cells to ✓.
Engine columns with low win rates
If one engine consistently ignores you across many prompts while others cite you, it’s worth investigating what that engine tends to prioritise.
- Perplexity often leans heavily on recently updated, well-cited content
- ChatGPT often favours established domains and comprehensive articles
- Claude responds well to content with clear structure and factual depth
Identifying which engine is your weakest helps you focus your content improvements.
Columns where you win consistently
These are prompts where your content is already working. Note what those articles have in common, such as length, format, structure, and how they cite sources. Then apply the same approach to the prompts where you’re losing.
Connecting losses to content action
For every row where you see ✗ across multiple engines, the next step is the same: open the Per-Prompt Citations section below the matrix and expand that prompt. You’ll see exactly who is being cited instead of you, and what content they’re citing. That tells you the bar you need to clear.
From there, take the prompt into the GEO Article Writer and write a piece specifically targeting it. Use the prompt as one of your target prompts in the writer, and the article will be built around covering that question in a way that makes it a strong citation candidate.
Note: If the prompt already has a ✓ on some engines, check whether you already have an article covering it before writing a new one. The issue may be that the existing article needs updating or restructuring rather than replacing.
Tracking improvement over time
The matrix reflects a single run’s results. After you publish new content targeting your ✗ prompts, the following week’s automatic run will show whether those cells have flipped to ✓. This creates a direct feedback loop between your content output and your AI visibility. Each week’s matrix becomes a report card on the previous week’s work.