Share of Voice by Funnel Stage (Awareness / Consideration / Decision)

Why funnel-stage share-of-voice changes the conversation

Most AI visibility tools report a single share-of-voice number. The problem with that number is it averages across very different fights. The prompt "what is generative engine optimization" is fundamentally different from the prompt "best GEO platform for a LATAM SaaS with a 200-person sales team" — yet both contribute equally to a flat SoV score. One determines whether buyers ever discover your category; the other determines whether they choose you when they're ready to buy.

Lumos breaks AI share-of-voice into the three classical funnel stages — awareness, consideration, decision — so you can see which stage you actually own and which one is bleeding. Two brands with identical overall SoV can have wildly different pipeline outcomes if one wins decision-stage prompts and the other wins awareness ones.

What you see after signup

The preview above is Lumos's own funnel-stage SoV this week, run across our full prompt library. Once you sign up, the same view runs against your brand.

• Per-stage SoV. For every prompt in your library, we calculate where you rank in the AI's answer and aggregate by stage. You see three numbers: awareness SoV, consideration SoV, decision SoV.
• Stage-level top prompts. The prompts driving each stage are listed individually with the engine and rank, so you can immediately see whether your decision-stage win is being driven by one prompt or by a healthy spread.
• Engine × stage crosstab. The full view crosses funnel stage with AI engine. You might dominate decision-stage prompts in ChatGPT and lose them in Perplexity — a clue that Perplexity's grounding sources favor a competitor's content.
• Weekly deltas. Stage-level SoV is tracked week over week so you can spot a decision-stage drop before it shows up in pipeline.

How Lumos classifies prompts into funnel stages

Every prompt in your library is tagged at ingestion based on intent. The default taxonomy:

• Awareness — informational queries that signal the buyer doesn't yet know your category exists. Patterns: "what is X", "how does X work", "why does X matter".
• Consideration — comparative or evaluative queries that signal the buyer is shortlisting. Patterns: "X vs Y", "top X tools", "X review 2026", "X alternatives".
• Decision — buyer-intent queries that signal the buyer is ready to choose. Patterns: "best X for [vertical/use case]", "X pricing", "X enterprise", "X for [region]".

You can override any tag per prompt if your category has different patterns. The classification model is exposed in the dashboard so you can audit any decision.

Reading the result

The Lumos demo above shows our own funnel-stage SoV at the moment. A few things to notice:

• Decision-stage rank is #1. Lumos wins the prompt "best GEO platform for LATAM SaaS" — exactly the prompt buyers in our ICP fire when they're ready to choose. That single prompt is worth more than 100 awareness mentions to our pipeline.
• Consideration is mid. We sit at #3 for "GEO tools comparison 2026" — meaning we're on the shortlist but the AI isn't yet leading with us. The fix is publishing comparison content that AI engines can ground on.
• Awareness is weak. We rank #5 for "what is generative engine optimization" — the broad category-defining query. That's a long-term brand asset we haven't built yet.

This shape — strong decision, weak awareness — is typical of a brand that's converting demand it didn't create. It's a perfectly fine quarter, and a worrying year.

Common mistakes when reading funnel-stage SoV

Optimizing the wrong stage for the moment. If your sales team is hitting quota and your AE-to-SDR ratio is healthy, awareness SoV is the lever. If your pipeline is thin, decision SoV is the lever. A brand that pours its content budget into awareness when decision is bleeding will look great in 12 months and miss its number this quarter.

Treating all decision-stage prompts equally. "X pricing" and "best X for [vertical]" are both decision-stage but mean very different things to your funnel. Lumos lets you drill into the underlying prompts so you optimize the ones where the rank is most leverage.

Ignoring the engine × stage crosstab. A 35% decision SoV that's all coming from one engine is fragile. The same number spread across ChatGPT, Gemini, Claude, and Perplexity is durable.

Skipping awareness because it's slow. Awareness SoV correlates with pipeline 6-12 months out. Brands that skip it become commoditized — they show up in comparison answers but never as the default option.

Next steps

• Run the Lumos funnel-stage SoV on your brand: sign up and import your prompt library — first scan in minutes.
• See your overall position: GEO Score Calculator.
• Read the strategy behind funnel-stage GEO: What is GEO?.