For about 15 years, “share of voice” in search meant one thing. Pull a list of your category keywords. Pull your competitors. Compare visibility scores. The bigger your share, the more of the demand you were capturing. Rough but useful.
In 2026, that calculation gives an answer that is increasingly disconnected from reality. Half of the impressions on a typical commercial query are now AI Overviews, ChatGPT replies, Perplexity answers, or Google answer boxes. Your traditional share of voice can be growing while your actual share of attention falls. We have been working on a more honest framework for our clients. Here it is.
The four surfaces, weighted
A query in 2026 is answered across up to four surfaces.
- The classic SERP (blue links, local pack, ads).
- The Google answer surface (featured snippets, AI Overviews, knowledge panels).
- Standalone answer engines (ChatGPT, Perplexity, Claude, Gemini).
- In-app and vertical AI tools (Reddit answers in Google, in-app shopping assistants, vertical search).
Each one has its own visibility metric. Each one has its own share. The honest measure of share of voice is a weighted blend of all four, not just the first.
How we weight
Weights vary by industry, but a useful starting point for most B2C local service businesses:
- Classic SERP visibility: 50%
- Google answer surface citations: 25%
- Standalone answer engine citations: 20%
- In-app and vertical: 5%
For B2B and enterprise, the weighting shifts. Standalone answer engines move up to 30 to 35%, because that is increasingly where buying committees do early research. Classic SERP drops to 40%.
If you are a B2B brand and you are not measuring how often ChatGPT mentions you when a buying committee asks for vendors in your category, you are flying blind on the channel that is increasingly producing the shortlist.
How to actually measure each surface
Classic SERP
Same as it has always been. Pull a category keyword list. Score visibility in tools like Ahrefs, Semrush, or our internal stack. Compare to a fixed competitor set.
Google answer surface
Track which queries in your set are returning AI Overviews, snippets, and knowledge panels, and how often you are cited inside them. We use a combination of GSC data and a daily SERP scrape.
Standalone answer engines
Run your category questions through ChatGPT, Perplexity, Claude and Gemini on a weekly schedule. Capture which sources are cited. Score yourself vs your competitor set. This is mostly a custom job in 2026. The tooling is improving but not mature.
In-app and vertical
Channel by channel. Reddit citations inside Google. Industry specific AI tools. Track what matters in your category and ignore the rest.
What this changes operationally
A few things, depending on what your blended score reveals.
If your classic SERP share is healthy but your answer engine share is weak, you are about to lose ground. Your traffic numbers will hold for another 6 to 12 months while AI engines eat the easy queries. You should be investing in AEO now.
If your answer engine share is healthy but your SERP share is weak, you are punching above your weight in the new world but you are vulnerable to anyone with deep classic SEO who learns to write for the LLM. Defend the position by hardening the foundations.
If both are weak, you have a real problem and the right place to start is usually classic SEO, because it remains the floor that everything else builds on.
The bottom line
Impressions are not going to be a useful metric for much longer. Sessions either. The marketers who are going to look smart in 18 months are the ones who started measuring share of attention across all four surfaces in 2026, not the ones who waited for their tooling vendor to catch up.