AI NativeNot every question a buyer might ask an AI carries the same commercial weight. A question asked by ten buyers a month at the moment they are comparing providers before a purchase is worth more to measure than a question asked occasionally by people at the start of a category exploration. Demand and value scoring is how AI Native turns that commercial difference into a number that drives which gaps to close first.
Why conversational prompts need head-term anchoring
AI assistant questions are long and conversational. A question like "which provider should I choose for a personal loan if I need funds within 48 hours and have an existing relationship with my bank" does not appear as a keyword that a search-volume tool can return a number for. It returns zero or near-zero, not because nobody asks it, but because the exact phrasing is unique to each asker.
The solution is head-term anchoring. For each prompt in the tree, AI Native identifies the short-tail head keyword that the conversational question belongs to: the one-to-three word phrase that carries the demand for that question's cluster. The head keyword "personal loan" might carry 200,000 monthly searches, and the conversational prompt is a specific expression of that demand. The prompt inherits the head term's volume as its base demand signal.
Demand is then supplemented with AI-specific keyword signals. Some questions that generate meaningful AI assistant traffic do not appear prominently in traditional search volume. The platform layers AI keyword data alongside search volume and takes the higher of the two signals, so prompts that are emerging as AI-native question patterns are not undervalued by a volume tool that is calibrated to a different behaviour.
How value is calculated
The value score for a prompt is a product of three inputs:
- Demand: the normalised head-term volume, capped at 1.0 for the highest-demand term in your set and scaled proportionally for every other prompt.
- Intent weight: a multiplier for the commercial intent class of the question. Transactional and comparison queries score highest because they are asked by buyers closest to a decision. Informational queries score lower because the gap between an informational impression and a conversion is longer. The multiplier values, highest to lowest, are transactional, comparison, commercial, evaluative, informational.
- Funnel weight: a multiplier for where in the buyer journey this question sits. Decision-stage questions are weighted most heavily; awareness-stage questions least. This reflects the fact that influence at the decision stage has a more direct line to revenue than influence at the awareness stage, even if awareness-stage volume is higher.
For branded prompts, a floor is applied to the demand normalisation. Branded questions like "is this provider reliable" or "what are the complaints about this brand" may not have high search volume on their own, but they are asked at high-intent moments by buyers who already know your brand and are deciding whether to proceed. The floor ensures branded prompts are not zeroed out by low keyword volume when the intent and funnel signals are strong.
The final value score is multiplied by the persona's business weight before it feeds into the opportunity score. A need-state persona that your product team has identified as high-revenue carries a higher weight than a peripheral persona, so the value of a prompt is anchored to how much the buyer segment matters to the business, not just to search volume.
People Also Ask expansion
After the Layer-1 prompt tree is built and grounded, the platform expands it with real questions pulled from People Also Ask results for each head term. These are questions that real buyers are asking in the context of your category, sourced from live search data. Each PAA question is mapped to a persona-stage-class cell and deduped against the existing tree by topic similarity, so the expansion adds genuine new coverage rather than near-duplicate phrasing.
PAA questions inherit the head term's demand signal from the same anchoring step. This means a real buyer question discovered through PAA expansion is measured against the same demand baseline as the structured Layer-1 prompts it sits alongside, and its value reflects that real demand rather than a proxy.
What the value score tells you
Value is not a proxy for how hard an opportunity is to capture, only for how much it would matter if you captured it. A high-value prompt with a low current visibility score is an opportunity worth prioritising. A high-value prompt where you already lead is a position worth defending. A low-value prompt where you are absent might still matter for completeness but should not come before the high-value gaps.
The value score feeds directly into the opportunity score. See The opportunity score for how value, gap, and winnability combine.
Questions
Why is my branded prompt's value high even though the search volume is low?
Branded prompts receive a demand floor in the normalisation step. This floor is applied because branded questions are asked at high-intent moments by buyers who have already narrowed to your brand. Even if the raw search volume for a branded query is modest, the intent and funnel weights, combined with the floor, reflect the commercial importance of those answers. You want to be well-represented when a buyer is checking you specifically.
Can I see which head term a prompt is anchored to?
Yes. The prompt detail view shows the head term that was extracted for each prompt and the demand volume that head term carried. If the extraction produced a head term that does not match the question well, you can adjust the prompt or the anchoring through the Brand Truth Studio.
Does demand data come from AI Native directly?
Demand is sourced through our data partners rather than estimated by AI Native. The platform queries real search volume and AI keyword signals from external data sources and uses the higher of the two values for each head term. AI Native does not fabricate volume numbers or estimate them from internal signals.
How is the intent multiplier assigned?
Intent is classified when the prompt is generated. The generation step asks the model to label each question with one of the intent types (informational, commercial, comparison, transactional, evaluative). That label is stored with the prompt and used in the value calculation. Prompts sourced from PAA expansion are assigned intent through a fast heuristic classifier that checks for transactional and comparison signals in the text.
What happens to value when demand data is unavailable?
If a head term returns zero from both the search volume and AI keyword sources, the prompt's demand is recorded as zero but the prompt is not dropped from the tree. Its value will be low, which means it will rank below higher-demand prompts in the opportunity queue, but it is still measured and scored. In that case, intent and funnel position still differentiate prompts relative to each other within the zero-demand set.
Where can I read about how value feeds the opportunity score?
See The opportunity score for how value combines with gap and winnability to rank the cells that matter most.