Prompt Tracking Alert Badges — Reading the Warning Signals

What Are Prompt-Tracking Alert Badges?

Each row in the Prompt Tracking table can carry a small alert badge to the right of the prompt text — for example, ⚠️ Competitor Intercept +3 or ⚠️ Low Visibility +7. These badges are the at-a-glance signal of what's wrong with each tracked prompt: AI engines aren't recommending you, or they're recommending competitors instead, or your brand is being described incorrectly.

This article explains exactly what each badge means, how the system decides which one to show, and what to do when you see one. If you're new to Alerts in general, start with the AI Alerts overview first — this guide focuses specifically on the per-row badges in Prompt Tracking.

The Two Most Common Badges

For 95% of users, the two badges that matter are:

The number after the badge (e.g., +7) is the count of additional alerts on this row beyond the one being shown — typically because the same prompt has issues on multiple AI platforms.

Competitor Intercept — When You're Being Replaced

This is the badge to triage first. It means: a real user could type this prompt into ChatGPT (or Claude / Gemini / etc.) and get an answer that names another tool, while ignoring yours.

Click the badge to open the full alerts panel, which shows:

• Which platforms intercepted you — e.g., ChatGPT and Perplexity flagged this; Claude is fine
• Which competitors appeared — the actual brand names the AI mentioned
• Suggested actions — written-out playbook steps for getting your brand into the answer

The competitor extraction is now stricter than earlier versions of Citadex. Section headings, verb-led imperatives ("Provide context"), and lowercase multi-word labels ("press releases") are filtered out of competitor lists automatically — so when you see a Competitor Intercept badge today, the listed competitors are almost certainly real brand names, not extraction garbage.

What to do

1. Read the snippet. Click into the evidence view to see the actual AI answer. Understand why competitors got named: did they have a better positioning page? Did the AI cite a specific third-party review?
2. Add the relevant source to your Knowledge Base. If the AI is citing the competitor's docs / pricing / case studies, your equivalents need to exist publicly and be high-quality.
3. Generate a Strategy Article targeting this exact prompt cluster. Use Deep Research mode — the Strategist agent specifically uses the competitor names as positioning context.
4. Earn citations on third-party listicles. AI engines pull from "best of X" articles. If you're not in those listicles, AI won't pull your name. See the outreach guides for how.

Low Visibility — When the AI Doesn't Say Your Name

This badge replaces what was called "Rank Drop" in earlier Citadex versions. It now covers two scenarios:

1. Always-zero coverage: the prompt has been tracked for a while and never crossed into "your brand mentioned" territory. Nothing dropped — you just never were there. The badge accurately says "low visibility" instead of misleadingly saying "rank drop."
2. Actual decline: previously you were mentioned on this prompt and recently you're not. The badge still says Low Visibility, but the alert detail page distinguishes "first time at 0%" from "fell from X% to 0%."

Click into the alerts panel to see which case applies.

What to do

• For always-zero prompts: This is a content-gap problem. The AI hasn't been trained or doesn't have a current source to associate your brand with this query. Fix by creating content that targets this prompt cluster and ensuring it's indexable (sitemap, schema, internal links).
• For decline-from-presence: Investigate what changed. Did your relevant page go offline? Did a competitor's content overtake yours on key signals? Did the AI engine update its training cutoff and lose the old data?

Why Some Rows Have No Badge At All

If a prompt row shows no badge, one of these is true:

• You ARE mentioned on enough platforms that no platform-level alert was generated
• No platform has scanned this prompt yet (status shows "checking…" or "queued")
• Coverage is mixed but above the alert threshold — Citadex doesn't fire alerts when the picture is acceptable

This is intentional. The badges are designed to direct attention to real problems, not paint every row with warnings.

Severity Ordering and the +N Counter

When a prompt has multiple alerts across multiple platforms, Citadex picks the most severe one to display and shows +N for the rest. Severity order:

1. Competitor Intercept — most urgent (someone else is winning the prompt)
2. Low Visibility — second (you're absent, but at least nobody else clearly won)
3. Entity Clarity — third (you're mentioned, but described wrong)

So a row showing ⚠️ Competitor Intercept +3 means: at least one platform intercepted you (the displayed alert) and three more alerts of either equal or lower severity also exist on this prompt. Click to see all of them.

What Changed in the Recent Revision

If you've been using Citadex for a while, you may have noticed the badges change wording and behaviour recently. Three changes:

1. "Rank Drop" → "Low Visibility." The old label was misleading for prompts that were at 0% coverage from the start — there was no "rank" to drop from. Low Visibility describes both the never-covered and actually-declined cases honestly.

1. Competitor extraction filtering was tightened. Earlier, the evidence panel sometimes listed article section headings ("Ask directly in your prompt", "Press releases", "Pro tip") as if they were competitor brand names. The system now filters those out via three layers: a strict regex at extraction time, a heading-shape filter, and a stopword filter that catches phrases containing "your / the / for / to / in".

1. Intent gate was removed for alert classification. Earlier, only "best of" and comparison prompts could trigger Competitor Intercept alerts; how-to and FAQ prompts would always fall to Low Visibility even when real competitors appeared in their answers. That was over-aggressive. Today, any prompt where real brand-name competitors appear in the answer fires Competitor Intercept — regardless of the prompt's surface intent.

If you're seeing different behaviour from what you remember, this is why.

Reading the Evidence Panel

Click any badge to open the Evidence view. The structure:

• Top: Platform selectors. Each AI platform is listed; the one with the most severe alert is selected first. Click another to switch.
• Middle: The actual snippet of the AI's answer for this prompt on this platform. Highlighted: your brand name (if mentioned), competitor names (if extracted).
• "Competitors found in this answer": the filtered competitor list. Headings and topic phrases are excluded.
• Detected at: timestamp of the scan that produced this evidence.

The evidence is what's actually being said about you in AI right now. Read it carefully — most actionable AEO insights come from reading these snippets, not from looking at the badge alone.

Frequently Asked Questions

Q: I see a Competitor Intercept badge but the listed "competitors" don't make sense.

Tell us — this should be rare. The filtering is designed to reject anything that isn't a real brand name. If a topic phrase still slips through, send the prompt + platform + snippet to support and we'll tighten the filter. (See also: the "What changed in the recent revision" section above — the most common source of bad competitor extraction has already been fixed.)

Q: My prompt has had 0% coverage for weeks. Why am I only seeing Low Visibility now?

The system fires alerts when prompts have been scanned across enough platforms to be statistically meaningful. A prompt that's only been scanned once won't get a Low Visibility alert; once it's been scanned a few times on multiple platforms with the same 0% result, the alert fires.

Q: How do I clear an alert?

You can't dismiss alerts manually — they clear automatically when the underlying problem is fixed. Specifically:

• Competitor Intercept clears when your brand starts appearing in the platforms where competitors used to be
• Low Visibility clears when your brand appears in answers for that prompt on any tracked platform

A re-scan after you publish new content / update existing pages will surface fresh evidence in 1–24 hours depending on the AI engine.

Q: Why does one platform's badge persist after I fixed the others?

Each platform's alert is independent. Some AI engines update their effective training data faster than others (Perplexity is fastest, ChatGPT and Claude are slower). A fix that takes 24 hours to register on Perplexity might take 1–2 weeks on Claude. The +N badge will shrink as each platform clears.

Q: Do the badges fire on prompts I've never scanned?

No. Alerts are based on actual scan data. A prompt that's only been added to your tracking list but never scanned won't generate badges. Wait for the first auto-scan to complete (status will move from "queued" to a real coverage %).

Q: What's "Entity Clarity"?

A third (less common) badge that fires when your brand IS mentioned, but described in the wrong industry or with wrong attributes — e.g., AI says you're a CRM tool when you're an AEO platform. This signals a brand-entity confusion problem (the AI has a wrong mental model of your business). Fix by updating your structured data (Organization schema), Wikidata entry, and any third-party listings that miscategorise you.