Model Impression Share

Model Impression Share (MIS) estimates how much of the available organic visibility you capture across a keyword set. In plain terms, it answers a better question than average position: what share of the market are we actually getting?

The usual model is simple enough: search volume or impression potential multiplied by expected CTR at your current rank, then divided by total available impressions in the set. If your MIS is 22% on a 300,000-impression topic cluster, you are modeling that roughly 78% of the opportunity sits with competitors, SERP features, or both.

Why SEO teams use it

Average rank is weak on its own. A move from position 8 to 5 on a 20-search keyword is noise; the same move on a 40,000-search keyword is budget-worthy. MIS fixes that by weighting rankings by opportunity.

• Forecasting: If a cluster moves from 18% to 26% MIS, you can estimate incremental clicks and revenue with a straight face.
• Prioritization: Keywords sitting in positions 4-10 often produce the biggest MIS gains per content update or link acquisition sprint.
• Competitive reporting: It is easier to explain “we hold 31% of category visibility” than dump 500 keyword positions into a slide.

How to calculate it properly

Most teams build MIS from rank tracking data in Ahrefs, Semrush, STAT, or a SERP API, then calibrate with Google Search Console. Screaming Frog is useful here too, not for the model itself, but for mapping keywords to URLs and spotting cannibalization that distorts the output.

A practical formula looks like this:

MIS = sum(keyword impression potential × expected CTR at current rank) / sum(keyword impression potential)

Use your own CTR curve if possible. GSC query and page data is usually the best starting point because generic CTR studies age badly. A 2022 curve is not reliable in a 2026 SERP full of ads, AI Overviews, video packs, and People Also Ask.

Where MIS breaks down

This metric is only as good as its assumptions. That is the caveat people skip.

• CTR curves are unstable: Brand bias, SERP features, device mix, and query intent can wreck a blended model.
• Search volume is approximate: Ahrefs, Semrush, and Moz all model volume differently, and low-volume terms are often wrong by a lot.
• Rank tracking is not reality: Personalization, localization, and volatile SERPs mean your “position 3” may not be what users actually see.

Google's John Mueller has repeatedly said rankings are not fixed, universal positions, and that matters here. MIS is a directional planning metric, not an accounting metric. Treat it like a forecast model, not ground truth.

Best use cases

MIS works best for non-branded topic clusters, category-level reporting, and quarterly planning. It is especially useful when you need to compare content hubs, countries, or product lines on the same scale.

It is less useful for tiny keyword sets, news-driven SERPs, or anything dominated by SERP features that steal clicks. If AI Overviews suppress organic CTR by 15-30% for a query class, your old MIS model will overstate opportunity unless you adjust for that explicitly.

Bottom line: MIS is one of the better growth metrics in SEO because it connects rankings to market share. Just do not pretend the model is cleaner than the data feeding it.