Why is my visibility dropping on a specific topic?

A drop in visibility on one topic is usually signal, not a bug — here's how to read it.

First, expect some movement. AI models are non-deterministic — they give different answers to different people at different times. GetMint runs repeated simulations and reports your visibility as a probability, so small week-to-week swings on a topic are normal. Always read the trend over several weeks, not a single refresh.

A real, sustained drop usually means one of these:

• A competitor gained ground. Another brand is now being cited more often for those questions, pushing you out of the answer. Check your Share of Voice for that topic to confirm.
• Your source content went stale. AI engines lean on fresh, authoritative pages. If your content on that topic is thin or outdated, models pull from stronger sources instead.
• The questions themselves shifted. AI answers evolve as the models update. A phrasing that surfaced you last month may now surface someone else.

What to do:

1. Open the topic and look at which sources the AI cited — that's your most actionable clue.
2. Compare against the competitors now appearing. What do their cited pages cover that yours don't?
3. Strengthen or publish content that directly answers the question, then track the trend over the next few refreshes.

Improving measured visibility is a content-and-authority effort: expect a few weeks for early movement and a few months for durable gains.

Read by model — where exactly did the drop land?

A drop is rarely uniform across AI engines. Filter the topic by model (ChatGPT, Perplexity, Gemini, Google AI Overview) and look at each line on its own.

• Drop on one model only — almost always a source problem for that model. The other models still cite you because their source mix is different. Open the Sources view filtered to the affected model and look at which third-party domains now dominate.
• Drop across every model — broader: usually content gone stale, a question that shifted in phrasing, or a competitor that genuinely gained authority everywhere.
• Drop on a model that uses live web search but not on the others — fresh-content signal. The web has new pages on the topic that don't include you yet.

Identifying the model that moved is half the work — it tells you where to focus content and outreach.

A cadence that turns the dashboard into a plan

The right cadence keeps you ahead of drops instead of reacting to them.

• Weekly — anticipate and observe. Open the topic, scan the trend, note any model that moved. No action yet; this is the early-warning loop.
• Monthly — plan and act. Look at the month of weekly readings together. Pick the one topic × one model with the clearest drop, decide what content to write or which source to pitch, and ship it. One thing per month, done well, beats five things half-done.
• Quarterly — report to leadership. Roll up the GetMint Score, Share of Voice trend, and the actions you took into a single deck. Show what moved, what didn't, and the bet for the next quarter.

Still stuck? Email support@getmint.ai with your project URL and a screenshot of the topic.

Key takeaways

• A drop on one topic is usually signal, not noise — but always read the trend, not a single refresh.
• Real, sustained drops trace to: a competitor gaining ground, stale content, or a shift in how the question is asked.
• The cited sources view is your most actionable clue — look at what competitors' cited pages cover.
• Filter by model to localize the drop — one-model drops are source problems; cross-model drops are broader.
• Cadence: weekly to anticipate, monthly to act, quarterly to report.