A good citation-change alert should be structured around what changed, how important it is, and what to do next. For answer engines, I’d recommend organizing each alert into these parts:

1) Alert header

Include the essentials first:

• Query / topic
• Engine / source (e.g. Google AI Overviews, Perplexity, ChatGPT Browse, Bing Copilot)
• Detected change type
  • citation added
  • citation removed
  • citation replaced
  • citation order changed
  • answer text changed with same citations
  • source domain changed
• Severity / priority

2) Change summary

A short, human-readable summary:

• “Top citation changed from example.com to competitor.com”
• “3 citations removed, 2 added”
• “Answer now cites an internal doc instead of public source”

This should be one or two sentences max.

3) Diff details

Show the before/after clearly:

• Old citation list
• New citation list
• Added / removed / moved
• Anchor position if relevant
  • top citation
  • supporting citation
  • footnote/reference order

If possible, include a compact diff view:

• - sourceA.com
• + sourceB.com
• ~ sourceC.com (moved from #2 to #4)

4) Impact context

Not every citation change matters equally. Add fields that help rank urgency:

• Business importance of query
• Brand / competitor mention
• Traffic or conversion impact
• Authority of replaced source
• Recency of change
• Historical volatility for this query

Example:

• “High impact because this query drives product-intent traffic”
• “Competitor domain gained top citation”
• “Citation changed after 18 stable days”

5) Evidence snapshot

Store enough evidence to debug:

• Timestamp detected
• Exact answer text
• Citations with URLs
• Screenshot or HTML snapshot
• Prompt/version/locale/device if available

This is important because answer engines can be personalized or dynamic.

6) Classification / reason

If you can infer it, label the likely cause:

• content update on cited page
• ranking shift
• model answer rewrite
• SERP/source availability change
• localization/regional variation
• user-agent / logged-in state difference

Even if uncertain, a “likely cause” tag is useful.

7) Recommended action

This turns an alert into something actionable:

• verify whether the new cited source is accurate
• check if your page lost support/authority
• update content to regain citation
• investigate competitor coverage
• archive if low priority

8) Alert metadata

Helpful operational fields:

• alert ID
• monitored query ID
• monitoring interval
• baseline version
• confidence score
• suppression/deduplication key
• severity rules applied

---

Recommended alert schema

Here’s a practical structure you can use:

{
  "alert_id": "cit-20260924-0001",
  "timestamp": "2026-09-24T10:15:00Z",
  "query": "best project management software for agencies",
  "engine": "perplexity",
  "locale": "en-US",
  "change_type": ["citation_replaced", "top_source_changed"],
  "severity": "high",
  "summary": "Top citation changed from projectmanagepro.com to competitorhub.com.",
  "diff": {
    "added": ["https://competitorhub.com/reviews/pm-tools"],
    "removed": ["https://projectmanagepro.com/agency-pm-guide"],
    "moved": ["https://saasreview.com/pm-tools (2 -> 4)"]
  },
  "current_answer": "…",
  "current_citations": [
    {"rank": 1, "title": "...", "url": "..."},
    {"rank": 2, "title": "...", "url": "..."}
  ],
  "previous_citations": [
    {"rank": 1, "title": "...", "url": "..."},
    {"rank": 2, "title": "...", "url": "..."}
  ],
  "impact": {
    "business_priority": "high",
    "query_type": "commercial",
    "historical_stability_days": 18
  },
  "confidence": 0.92,
  "likely_cause": "source ranking shift",
  "recommended_action": "Review the new cited page and update your content if needed."
}

---

Best practices

• Deduplicate repeated alerts for the same query/source pair.
• Group changes into one alert when multiple citations change at once.
• Track both:
  • citation-level change
  • answer-level change
• Use severity rules like:
  • top citation replaced = high
  • citation order changed = medium
  • low-priority source changed = low
• Include historical trend context to reduce noise.

---

Simple severity model

You can score alerts using something like:

• +5 top citation changed
• +4 brand/domain removed
• +3 competitor added
• +2 answer text changed
• +2 high-value query
• +1 after long stability window
• -3 low-importance query
• -2 known volatile query

Then map:

• 0–3 = low
• 4–7 = medium
• 8+ = high

---

If you want, I can also give you:

1. a normalized database schema for these alerts, or
2. a Slack/email alert template with exactly what fields to show.