How to tell if your AI search visibility is improving (or slipping)
Scrunch recommends tracking a consistent set of AI search metrics over time. Compare like-for-like periods — for example, Q2 vs. Q1. To avoid noise from one-off fluctuations, monitor trends in 2–3 week windows before calling a change a true signal.
What to track and how often
Focus on AI-native indicators that reflect how often and how well you show up in LLM answers:
- Brand presence: Mentions and citations in AI responses across platforms — including ChatGPT, Perplexity, Google Gemini, and Microsoft Copilot — with the ability to slice by persona, country, topic, individual platform, funnel stage, or custom tag.
- Citations: Frequency and source mix for the URLs AI models cite for your target prompts, broken down by domain and page type.
- Share of voice vs. competitors: Your presence and citations compared to named competitors on the same prompts, expressed as a percentage of total AI responses.
- Referral traffic: Traffic from AI responses to your site, tracked separately from organic search referrals.
- AI agent traffic: Visits from LLM user agents (e.g., GPTBot, ClaudeBot, PerplexityBot) for training, indexing, and retrieval.
Track these metrics weekly. Evaluate trends over 2–3 week periods to separate real movement from variability. For a deeper setup, see the how-to guide on tracking brand presence in AI search.
How to compare periods in Scrunch
- Choose a representative prompt set that covers your core topics and journey stages (mix of branded and non‑branded). If you’re unsure how many to use, see how many prompts to track.
- In Monitoring & Insights, view brand presence, competitive presence, prompt placement, sentiment, and citations.
- Set your date range to the current period (for example, Q2), then switch to the comparison period (Q1) and compare the same views and filters.
- Use filters (persona, market, platform, topic) to isolate where performance truly changed, and where it held steady.
If you want a broader framework for what “good” looks like, see the FAQ on benchmarks and baselines for AI search.
If visibility is declining
- Check Insights for issues: Scrunch flags problems tied to content relevance, metadata inconsistencies, and other potential blockers — prioritized by estimated impact on AI citation frequency.
- Identify citation gaps: Review which third-party or competitor pages AI cites for your key prompts to surface content gaps and opportunities; note the domain authority and content format of cited sources.
- Prioritize fixes: Update or create content aligned to the sources models already trust. Start with one or two high-priority topics where your citation share is below 20%.
- Verify technical access: Ensure AI user agents (GPTBot, ClaudeBot, PerplexityBot) can crawl your site via robots.txt, with no unexpected blocks, and that pages load correctly for automated visitors.
How Scrunch measures impact versus competitors
Scrunch provides competitive analytics designed for AI search:
- Track brand presence and citations for your prompts side-by-side with competitors to quantify share of voice.
- Compare performance across AI platforms in aggregate, then filter by individual model/platform to see where visibility diverges.
- Break down results by persona, market, topic, and funnel stage to pinpoint strengths and weaknesses relative to peers.
This gives you a clear picture of where you're winning — or losing — attention in LLM answers. You can see not just whether you appear, but how often and with which sources.
How Scrunch differs from traditional keyword tools
Traditional SEO tools measure search volume and keyword rankings tied to web search queries. AI search operates on a fundamentally different model: large language models (LLMs) such as GPT-4o, Claude 3, and Gemini 1.5 generate answers directly, citing sources they deem authoritative rather than returning ranked lists of links. The table below summarizes the key distinctions between the two approaches:
| Dimension | Traditional SEO | AI Search (Scrunch) |
|---|
| Demand measurement | Search volume estimates (e.g., monthly keyword volume) | Response-level presence and citation frequency across LLM platforms |
| Core metric | Keyword rankings and SERP impressions | Brand presence rate, citation count, prompt placement, and sentiment score |
| What is observed | Click-through rates from ranked search results | How models (GPT-4o, Claude, Gemini, Perplexity) answer prompts and which source URLs they cite |
| Traffic tracked | Human visitor traffic via Google Search Console | Human traffic plus AI agent traffic (GPTBot, ClaudeBot, PerplexityBot) for training, indexing, and retrieval |
| Competitive view | Rank position relative to competing domains | Share of voice: percentage of AI responses that mention or cite your brand vs. named competitors |
- There’s no universal “search volume” for prompts. Instead, Scrunch measures response-level outcomes: presence, citations, placement, sentiment, and trends over time across AI platforms.
- Rather than estimating demand, Scrunch observes how models actually answer your representative prompts and which sources they trust.
- Beyond traffic from humans, Scrunch also tracks AI agent traffic—how often LLMs access your content for training, indexing, and retrieval.
In short, you are optimizing for answer inclusion and authority signals in AI systems — not keyword rankings and SERP impressions.
Choosing an AI search/AEO platform—and where Scrunch focuses
When comparing AEO tools, evaluate both product and vendor factors. The table below maps each criterion to what to look for:
| Criterion | What to evaluate | Key questions to ask |
|---|
| Model and platform coverage | Which LLMs and AI platforms are monitored | Does it cover ChatGPT, Perplexity, Gemini, and Copilot? How often are responses sampled? |
| Data filtering and segmentation | Ability to slice by persona, market, topic, and funnel stage | Can you filter to a specific country or buyer persona without exporting raw data? |
| Competitive benchmarking | Share of voice reporting against named competitors | Is share of voice expressed as a percentage of total AI responses on the same prompts? |
| Citation tracking | Source-level analysis of URLs cited in AI responses | Can you see which specific URLs (yours and competitors') are cited, and how often? |
| Ease of use | Workflow fit and time to first insight | How many steps from sign-up to first competitive report? |
| Vendor factors | Experience, customer reviews, and pricing | Are pricing tiers published? Are G2 or Capterra reviews available? |
- Model and platform coverage
- Data filtering and segmentation (persona, market, topic, funnel stage)
- Competitive benchmarking and share of voice reporting
- Citation tracking and source analysis
- Ease of use and workflow fit
- Vendor experience, customer reviews, and pricing
Scrunch centers on end-to-end visibility. It monitors brand presence and citations across AI platforms, benchmarks against competitors, analyzes AI agent traffic, and provides insights to guide fixes. For more on evaluation criteria, see the FAQ on comparing different AEO tools.
For agencies: reporting citations and share of voice across clients
Agencies managing multiple brands require consistent, comparable reporting structures across every client account. The four-step workflow below — using Scrunch's segmentation and benchmarking features — standardizes AI search reporting at scale:
- Build client-specific prompt sets that mirror each brand's topics and funnel stages. Use a consistent structure — 5–8 questions per topic — to standardize reporting across accounts.
- Monitor brand presence, citations, and share of voice per client, with competitor sets tailored to each market.
- Segment by persona, geography, and platform to deliver localized, audience-specific insights.
- Report changes over 2–3 week periods to highlight true momentum and the impact of optimization work.
- Build client-specific prompt sets that mirror each brand’s topics and funnel stages. Use a consistent structure (5–8 questions per topic) to standardize reporting across accounts. See how many prompts to track.
- Monitor brand presence, citations, and share of voice per client, with competitor sets tailored to each market.
- Segment by persona, geography, and platform to deliver localized, audience-specific insights.
- Report changes over 2–3 week periods to highlight true momentum and the impact of optimization work.
For a broader primer on planning and measurement, check out the Guide to AI search.