Score Methodology Update — Engine Weight Recalibration

On May 2, 2026, we recalibrated how your AI Visibility Score weights each AI engine. The new weighting reflects what the score is actually measuring — your practice's presence across AI surfaces a dental patient might use — and corrects an anchor point that no longer matches how generative AI search has evolved over the past 12 months.

1. What changed

The composite AI Visibility Score combines visibility signals from four AI engines. Each engine gets a weight that determines how much it contributes to the final score. We changed those weights:

Per-engine sub-scores (SIR by engine, KPC accuracy rate, PAWC positioning, DAIRA-18 citation diagnostics) are unchanged. Only the composite weighting shifts.

2. Why we changed it

2.1 The score now measures presence, not clicks

The old weighting was anchored on ChatGPT's 77% share of AI-driven referral traffic. That number is real, but it answers a different question than the one the score is for. Your AI Visibility Score isn't about which engine drove a click — it's about whether your practice gets mentioned at all when a patient asks AI about local dentists. That's a presence question, and presence works differently across engines than referral clicks do.

About 80% of local searches now resolve in-SERP without any click — the patient reads the answer, sees your rating, and calls you (or doesn't). A score weighted by clicks would systematically under-count visibility on engines that show patients answers without sending them off-site, which is exactly what AI Overviews and Gemini do. The new weighting puts presence at the center.

2.2 The market shifted

ChatGPT's share of the AI chatbot market dropped from about 87% to about 65% over the 12 months ending January 2026 (Similarweb / First Page Sage). It's still the leader and still the heaviest weight. But weighting it at 70% — the original level — implies ChatGPT is the only engine that meaningfully matters, which the data no longer supports.

At the same time, Google AI Overviews trigger on roughly 88% of healthcare-related searches, the highest of any industry tracked. For a dental practice, AIO is a critical surface that the old 10% weight under-represented. Gemini grew from about 5% to 21.5% chatbot market share over the same period, which justifies doubling its weight from 5% to 10%.

2.3 Perplexity carries citation transparency

Perplexity is the only engine where we can attribute every visibility claim to a specific source — its citations are explicit and machine- readable. That makes Perplexity our most measurement-trustworthy engine, which justifies a heavier weight than its raw market share would suggest. Perplexity's Yelp Fusion API licensing and heavy Zocdoc indexing for healthcare also make it especially relevant for local dental visibility.

2.4 Per-engine directory citation rates

US dental, branded queries, 90-day rolling sample as of 2026-05-11.

• Zocdoc: cited in 23.9% of Perplexity dental responses. Citation volume on ChatGPT, Gemini, and Google AI Overview is too low in the current sample to report a reliable rate.
• Healthgrades: cited in 10.2% of Gemini dental responses; 3.8% of Perplexity dental responses. Citation volume on ChatGPT and Google AI Overview is too low in the current sample to report a reliable rate.

Source: Seeniq internal audit data from Pro + Agency-tier customers. Citation rates vary by practice geography, query volume, and competitor density.

3. What you'll see in your score

• If your practice is over-indexed on ChatGPT (mentioned a lot on ChatGPT, less on the others), you may see a meaningful score drop. This is the recalibration working as intended — the new score reflects that single-engine presence is a less reliable signal than presence across multiple engines.
• If your practice is mentioned across all four engines, you will see a modest score gain. The increased weight on Perplexity, AIO, and Gemini compounds in your favor when you have presence on those engines.
• If you have Perplexity citation misattribution issues (your practice is cited but linked to a competitor on Zocdoc), the γ-attribution penalty now hits noticeably harder than before — about 1.7× — because Perplexity now claims a larger share of the composite. Use the corrections list to fix these — they're the highest-leverage Perplexity fixes you can make.
• Per-engine sub-scores (the per-engine cards on your overview page) are unchanged. Only the composite changes.

4. What is NOT changing

• The factual-check pipeline. Your practice ground truth is compared against AI claims using the same 12-field schema, the same Haiku-based extraction, and the same deterministic comparison logic.
• The correction lifecycle. Detect, attribute, guide, notify, re-verify — same per-engine, per-data-path workflow.
• Re-verification windows. Same 2/30/30/45-day thresholds for Perplexity / Gemini / AIO / ChatGPT.
• Historical scores. Audits run before May 2, 2026 keep their stored composite scores. The new weighting applies only to audits run on or after May 2.

5. When this might change again

The current weights are documented as the binding decision for at least the next 6 months. We'll re-open the question only if one of two specific triggers fires:

• Gemini chatbot market share crosses 30% (currently 21.5%). At 30%+ Gemini becomes a peer of ChatGPT for general use, and the Gemini weight needs to climb above 0.10.
• We ship N=8 Stochastic Consensus Polling for AIO. This closes the variance gap that currently caps AIO at 15%. Once AIO sample stability matches the other engines, AIO can defensibly move to 20-25%.

Either trigger would produce a new methodology page like this one. We won't silently re-weight scores — every methodology change ships with a customer-facing record.

6. Questions or concerns

If you're seeing a score change that doesn't match what we've described here, contact us at support@getseeniq.com with your practice ID and we'll walk through your specific audit with you. Score changes are always derivable from the per-engine sub-scores, the SIR breakdown, and the engine weights — there is no opaque blending.