Share of Model

AI & Generative Search

Also: Share of Model Attention · LLM Share of Voice · AI Share of Voice

What it measuresHow often AI answers mention your brand

Descends fromShare of voice, moved to LLMs

MaturityMeasurement is still rough

Watch forVendors selling certainty they don't have

Quick definition

Share of model is your brand's presence inside AI-generated answers relative to competitors. When someone asks ChatGPT, Gemini or Perplexity a question in your category, share of model asks how often your brand gets named, cited or recommended versus the alternatives. It's the large language model (LLM) version of share of voice.

What it actually means

Share of voice measured how much of the category conversation you owned across advertising and media. Share of model asks the same question about a new surface: the answers large language models (LLMs) generate when someone asks about your category. If a buyer asks an AI assistant to recommend a mortgage broker or a project management tool, does your brand come up, and how often relative to competitors?

The logic is sound. As more buying research shifts from Google search results to AI answers, being named inside those answers matters the way ranking on page one used to. This is the frontier that generative engine optimisation is trying to influence, the same way search engine optimisation (SEO) chased rankings.

Here's where we get sceptical. The measurement is immature. LLM outputs vary between the same prompt run twice. They shift when the model updates. They change with the phrasing of the question and the region of the user. A share of model figure quoted to a clean percentage is describing a moving target with false precision.

Treat it as a directional signal about brand presence, not a dashboard number you defend in a board meeting.

Share of model is a real question dressed in a metric that isn't ready yet. Track the direction, distrust the decimal.

How it shows up

Share of model shows up when a marketing team runs a set of category-relevant prompts across ChatGPT, Gemini, Claude and Perplexity, then counts how often each brand appears in the answers. Some tools automate this at scale and report a percentage. Others sample manually.

It also shows up as a gap. A brand with strong share of voice in traditional media can be nearly invisible inside AI answers if it lacks the third-party citations, structured content and authoritative mentions that LLMs draw on. That gap is the useful part. It tells you where your brand's authority hasn't translated to the new surface yet. The single number is noise. The gap between channels is signal.

The Australian context

Australian brands face a specific disadvantage here. LLMs lean heavily on the volume of authoritative English-language content, and that corpus skews American. Ask an AI assistant for a category recommendation and it will often surface global or US brands before Australian ones, even when the user's context is clearly local.

That means share of model for an Australian business is frequently understated by geography, not by weakness. The practical response is investing in local authority signals, Australian citations, local review presence and clear regional relevance, rather than chasing a number that a US-weighted model was never going to hand you fairly.

Where people get this wrong

Quoting share of model as a precise percentage.LLM outputs are non-deterministic and shift with model updates, so any single figure describes a snapshot that won't reproduce reliably.

Treating it as a replacement for share of voice.AI answers are one discovery surface among many. Abandoning proven brand measurement for an immature metric trades a rough map for a blank one.

Buying tools that promise to guarantee AI answer placement.No vendor controls what an LLM outputs, and anyone selling guaranteed placement is selling certainty the technology cannot provide.

Share of Model vs Share of Voice

	Share of Model	Share of Voice
Surface measured	AI-generated answers from LLMs	Advertising, media and category conversation
Measurement maturity	Emerging and unstable	Established and repeatable
Reproducibility	Varies between identical prompts	Consistent within a measurement window
How to read it	Directional trend only	Trackable number over time

Common questions

How do you actually measure share of model?

You run a set of category prompts across the major AI assistants, then count how often each brand appears in the answers relative to competitors. Because outputs vary between runs, the honest approach is many prompts over time, reported as a trend rather than a fixed percentage.

Is share of model worth tracking yet?

The trend is worth watching, the precise number is not. As buying research shifts toward AI answers, knowing whether your brand shows up matters. Just treat it as directional. Anyone quoting a clean figure is describing a moving target with false confidence.

How is share of model different from share of voice?

Share of voice measures your presence across traditional media and category conversation and is established and repeatable. Share of model measures your presence inside AI-generated answers and is still immature and unstable. Same underlying question, very different reliability.

Can I improve my share of model?

Indirectly. LLMs draw on authoritative third-party mentions, structured content and citations, so the work overlaps with generative engine optimisation and strong brand presence. No one can guarantee placement in an AI answer, so be wary of any tool promising to.

Debrief

Get the next one

No spam. No fluff. Just the next article, straight to your inbox.

Keep exploring

About New Rebellion

New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.

How we think →