When a potential customer asks ChatGPT or Gemini about your industry, the AI builds its answer from specific phrases and concepts it has learned to associate with your space. Most organizations have no visibility into which phrases the AI actually uses about them. Lenses in Citate’s platform changes that. It clusters every phrase AI models use across hundreds of responses into semantic topics, ranks them by how often they appear, and lets you build saved analytical views around the topics that matter most to your strategy.
Lenses runs against a campaign. A campaign is a set of AI responses Citate has already collected for your prompts. Two summary tiles sit at the top of the page: the total number of responses sampled, and the total number of distinct phrases the AI used across them. Below those, the main topics table lists the clustered phrases ranked by their coverage percentage.
The main Lenses view. The Window selector sits at top-left, the RESPONSES and PHRASES tiles summarize the campaign, the Topic Size slider controls clustering granularity, and the Tracking panel on the right is where saved lenses are built.
Coverage percentage is the central metric. It shows what share of the sampled responses contain a given phrase or cluster. A topic with “436 matches · 83%” appears in 83% of all sampled responses for that campaign. It is a phrase the AI almost always reaches for when answering questions in your space. A topic with 12% is something the AI mentions occasionally. Together, the ranked list is your concrete map of what AI actually says about your subject, ranked by prevalence rather than by guesswork.
Unlike the Links tab, which catalogs the websites and URLs AI cites, Lenses focuses on the language itself. That means the words, phrases, and concepts the AI reaches for. The two views answer different questions: Links tells you which sources AI trusts; Lenses tells you what AI is actually saying.
The AI’s language doesn’t naturally arrive as neat topics. It arrives as thousands of overlapping phrases. The Topic Size slider at the top of the topics table controls how aggressively Citate clusters those phrases into groups. Sliding toward “fewer topics” merges related phrases into broader themes; sliding toward “more topics” splits them apart into narrower distinctions. There is also an Auto setting that picks a reasonable middle ground.
Clicking a topic row drills into its sub-topics. Here “safety” expands into “safety reports” at 62% coverage and “safety” at 100% coverage, showing how the same parent topic resolves into narrower phrases when you look closer.
The right clustering depth depends on what you are trying to do. If you are scoping a comprehensive pillar page that needs to cover an entire domain, broader clustering surfaces the high-level themes you need to address. If you are scoping a focused blog post about one narrow subtopic, finer-grained clustering reveals the specific phrasings you should weave into the copy.
Most users find the Auto setting works for a first pass, then adjust the slider once they know whether they are zooming out or zooming in.
The right-hand Tracking panel is where the feature gets its name. You drag topics from the main table into named groups in this panel, and that set of grouped topics becomes a lens. A lens is a saved configuration you can return to over time. The hint built into the UI puts it plainly: “All groups together = a lens.”
This matters because raw clustering on its own can be overwhelming when a campaign surfaces hundreds of distinct phrases. A lens lets you curate the view. Pull out the ten phrases your sales team needs to address objections, or the fifteen phrases your competitive intelligence team is monitoring, or the cluster of phrases that signal a specific buyer concern. Each becomes a named group inside the lens.
Below the topics, the Track a custom phrase field lets you add phrases the clustering did not surface on its own. This is useful for proper nouns, brand names, or specific terms you care about. Examples include competitors’ product names, your own brand mentions, and particular regulatory or technical terms. Once tracked, custom phrases appear in the daily trend chart alongside the auto-clustered topics.
You can save multiple lenses per campaign with the Default / Save / Save as controls at the top right, then return to any saved view whenever you need it.
Two views below the topics table turn coverage numbers into something more concrete.
The Activity over time chart plots daily mentions for the top phrases in the selected window. You can toggle the chart between “By groups” (showing your saved tracking groups) and “By topics” (showing the auto-clustered topics). This is where you see whether the phrases AI uses are stable, growing, or fading. The chart does not surface a momentum percentage. Instead, it shows the actual line, and you read the trajectory yourself. A flat line means stability. A rising line means the AI is starting to associate that phrase with your space more often. A falling line is the opposite.
The Activity over time chart plots daily mentions for the top extracted phrases. The “By topics” and “By groups” toggle controls whether you are watching the auto-clustered topics or your own saved tracking groups, and the “Show all 50 topics” button expands the chart beyond the default top ten.
The Response Viewer below the chart shows the individual AI responses that produced the data, with the tracked phrases highlighted directly in the text. Each response is dated, paginated, and sourced. This is the layer where you move from statistic to evidence: if “fiduciary duty” sits at 83% coverage, you can read the actual responses that contain it, see how the AI is framing the concept, and use that framing to inform your content. It is the difference between knowing what gets mentioned and knowing how it gets mentioned.
The Response Viewer with focus filtered to responses that contain “safety”. Tracked phrases are highlighted directly in the response text, and each response includes a date stamp and citation links so you can trace every claim back to its source.
At the top of the page, a Window selector lets you compute Lenses over a chosen time range: 7d, 14d, 30d, 90d, All, or Custom. Switching windows recomputes the snapshot. Each window is its own computed view rather than a comparison against another window. A “Last computed at” timestamp and a Recompute now button make the data freshness explicit.
Shorter windows (7d, 14d) show what AI models are saying right now, and they respond quickly to new training data or freshly indexed web content. Longer windows (30d, 90d) smooth out short-term variance and reveal the durable phrases AI consistently associates with your space. These are the topics that warrant deep, hub-level content investment. The “All” window draws on every response Citate has collected for the campaign, giving you the maximum-sample view at the cost of timeliness.
A reasonable workflow: anchor your strategy in the 30d or 90d view to identify what is durable, then check the 7d view weekly to spot phrases newly emerging. When the daily trend chart shows a rising line for a phrase in a short window, confirm it in a longer window before committing real content budget. As the Citate GEO Guide notes, single-point observations of AI behavior are statistically meaningless. The value comes from sustained, multi-window measurement.
A query run against Claude surfaces different phrases than the same query run against ChatGPT or Gemini. This is not a quirk. It reflects fundamental differences in how each model was trained, what data it emphasizes, and how it reasons about your domain.
Citate handles this structurally rather than with an in-page filter. The left sidebar groups campaigns under each LLM: Gemini, ChatGPT, AI Overview, Perplexity, and Meta.ai. You select which model’s perspective you want to analyze by choosing the campaign that runs against that model, then open Lenses on that campaign.
The practical implication is that you cannot build one generic content strategy from a single LLM’s Lenses view and assume it covers all of them. If Claude citations matter to your business, run a Claude campaign and analyze its Lenses on its own terms. If multiple models matter, open Lenses on each model’s campaign and compare the top-ranked topics. The topics that appear in all of them are the most stable optimization targets. They represent points where multiple AI systems independently converge, which reflects agreement across different architectures and training data.
The straightforward workflow is this: open Lenses on a campaign, identify the highest-coverage topics, check the daily trend chart to see whether they are stable or moving, read a sample of responses in the Response Viewer to understand how the AI is framing each topic, then turn that intelligence into content.
The Content Brief tab picks up directly from here. The topics you identify in Lenses as central are the natural inputs for hub-and-spoke content planning. These are topics with high coverage, a stable or rising trend, and language the AI is already using fluently. A pillar page should address every high-coverage topic in the lens. The supporting spoke articles should address the lower-coverage but still-relevant phrases that fill out the question space the AI is generating.
For competitor displacement, build a lens around phrases that include competitor names. When competitors appear at high coverage in AI responses about your category, those are the contexts where your brand should also appear. The Response Viewer shows you the exact framings where competitors are winning citations, so you can produce content that earns the same mentions.
For content gaps, look at the topics with low coverage that are nonetheless relevant to your audience. These are subjects where the AI is reaching for whatever it can find because no source has dominated the conversation. Quality content in those gaps tends to get cited quickly because there is little to compete with.
AI models are probabilistic. The same query run twice can produce different responses. If you ask ChatGPT a question once and base your entire strategy on that single answer, you are building on a sample size of one. That is guessing, not strategy.
Lenses solves this by aggregating data across hundreds of responses, multiple prompts, and multiple time windows. The snapshot you see is the product of statistical sampling, not a single anecdotal answer. The “Last computed at” timestamp and the Recompute now button at the top of the page make the data freshness explicit. You always know how current your view is.
AI models evolve. The phrases that defined your space last quarter may not carry the same weight this quarter. A campaign you set up six months ago will surface different topics today because the underlying models, search indexes, and discourse have all shifted. Check your topic data at least once a month, and recompute your saved lenses on a regular cadence. This keeps your strategy aligned with what AI considers important right now, rather than what mattered six months ago.
The organizations that get the most out of Citate use Lenses systematically. Build a saved lens for each strategic question you care about. Check it monthly. Watch the daily trend chart for early signal on emerging phrases. Read the Response Viewer when the numbers surprise you, so you understand the why behind the shift. Feed the high-coverage topics straight into the Content Brief tab for content planning. Visibility in generative AI responses is not accidental. It is the result of deliberate, evidence-based decisions about which phrases to compete on.
