Prompt Funnel sorts your tracked prompts by buyer intent across eight stages, from Discovery to Agentic, so you can see whether you only measure early questions or also the prompts where an assistant recommends a product, a shopper buys, or an agent acts for them. Click a stage and the table narrows to the prompts in it.
In the new Opttab it lives in the Prompts area as the Funnel View. Prompt tracking belongs to Discoverability, the first of our four pillars: Discoverability, Visibility, Optimization, Agentic Commerce. This post walks through the screen, the stages, and what we suggest you do with an empty one.
Why we built the Funnel View
Prompt lists tend to start at the top of the funnel. Early, exploratory questions are the first ones a team thinks to write, so they pile up. A brand that tracks mostly those will not see a Comparison gap, a weak Recommend position, a Reputation problem, or Transaction and Agentic prompts where it does not appear at all.
Every tracked prompt in Opttab already carries an intent tag. We wanted a view that shows the mix of those tags at a glance and lets you jump straight into one intent. The Funnel View does that one job.
Where to find the Funnel View
Open Prompts from the main navigation. The tabs along the top read Prompts, Responses, Search Queries, Visibility Signals and Funnel View. Click Funnel View.
The page splits into two parts. On the left, a panel called Intent Funnel draws the eight stages as a narrowing funnel, in order: Discovery, Education, Comparison, Recommend, Reputation, Operational, Transaction and Agentic. A green + at the top of the panel adds prompts. On the right is your tracked-prompt table.
The Intent Funnel narrows from Discovery to Agentic on the left. With no stage selected, the table lists all your tracked prompts, each in its market's language with its visibility, country, volume and scope.
The table has the same controls as the main Prompts tab: date range, Platforms, a filter button, Search Prompts, sort by Date added, delete, Generate and Add. The footer shows how many prompts are listed and lets you change the page size.
The eight stages, read plainly
The stage names come straight from the screen. We do not publish a longer definition for each, so the notes below are plain readings of the names, not product rules.
Discovery. Early, exploratory questions. The buyer is looking at a need or a category and may not have a brand in mind.
Education. Questions asked to learn: how something works, or what to look for.
Comparison. Options weighed against each other, product against product or brand against brand.
Recommend. The buyer asks the assistant to pick. The answer names a brand or product as the choice.
Reputation. Questions about trust in a brand or product, such as reviews and quality.
Operational. Practical questions about buying or using a product rather than choosing one.
Transaction. The buyer is ready, and the question is about buying.
Agentic. The shopper wants the assistant to act on their behalf, not only to answer.
Click a stage to filter the table
Each stage in the panel is a button. Click Comparison, for example, and the stage turns green while the table on the right filters to the prompts tagged with that intent. Click another stage to switch. The page address picks up the stage as well (it ends in ?stage=comparison), which makes a filtered view easy to bookmark.
With Comparison selected, the table shows only comparison prompts. The Scope column tells non-branded questions apart from prompts that name a competitor.
The filtered table has five columns. Visibility shows how often your brand appears in answers to that prompt. Prompts holds the full question as we send it. Country shows the market, Volume is a dot meter of estimated demand, and Scope labels the prompt.
Scope is worth a second look. We use labels such as Non-branded for open questions and Competitor-branded for prompts that name a competitor. A comparison prompt that names a competitor tells you how you fare when a buyer brings up a rival by name. A non-branded one tells you whether you make the shortlist at all.
Prompts are written in each market's own language. If you track a Dutch-speaking market, its comparison prompts are in Dutch, next to English prompts for an English-speaking one. Comparison prompts take forms such as where to buy a product from one of two named retailers, which premium brands to compare for a need, or which products under a price point offer a given feature.
Where the intent tags come from
The stage names match the labels in the Intent column on the main Prompts tab, and clicking a stage filters on that intent. So the funnel is a different way to look at tags your prompts already carry. If a prompt sits in the wrong stage, the place to look is its intent tag.
On the main Prompts tab, every tracked prompt carries an intent label in the Intent column. The Funnel View groups prompts by those same labels.
Fill a stage that looks thin
Once you can see the mix, the next step is filling the gaps. Click Generate above the table and describe what you need in plain words, such as comparison prompts for one product line in one market. The Prompt Generator writes the questions in the language of each market you pick.
Generated prompts arrive tagged with an intent, so you can tick the ones for the stage you are filling and add them in one go.
Pick 5, 10 or 20 prompts, review the list, tick the ones you want, and click Add Prompt. Since the stages filter on intent, a prompt you add with a Comparison tag shows up when you click Comparison. You can also write and edit your own prompts with Add above the table. Prompt Coverage shows how much of the relevant prompt set you track overall, which pairs well with the stage view.
How to read the mix
Start with the distribution. Note which stages are crowded, which are empty, and which matter most to how you sell. A subscription business and a brand selling one-off purchases will weight the stages differently.
The mix should then steer the work. Content written for an Education prompt is not the content that wins a Comparison prompt. Ads and agent readiness follow the same logic: put effort where the intents you need to win actually sit, including the late stages where an assistant has to act on clean product data.
When a stage is covered, measure the answers. The funnel tells you what you track. Answer Engine Analytics tells you how assistants answer those prompts, with mentions, citations, sentiment and sources.
What the Prompt Funnel does not show
The Funnel View shows the eight stages and the prompts in each. It does not show stage-to-stage conversion rates, drop-off percentages or revenue per stage, so do not read it as a sales funnel. The narrowing shape is a way to order the stages, not a measure of how many buyers move between them.
Keep the method in mind too. We send our own tracked prompts to AI models and record the answers. We do not see real users' private conversations, and prompt volumes are modeled estimates, not exact counts.
Prompt Funnel is also separate from the Brand Agent, our selling agent. The Brand Agent has its own Journey Orchestrator for shaping what the agent does at each step, and it is plan-gated to Enterprise for brands, and Agency Scale or Agency Enterprise for agencies. Prompt Funnel is a measurement view in the Prompts area. Our Brand Agent guide covers the selling side.
Getting started
If you already use Opttab, open Prompts in the account you already use at app.opttab.com, click Funnel View, and click through the eight stages one by one. The launch post on the new Opttab shows how Prompt Funnel fits with the other modules.
New to Opttab? Start your agentic journey or Book a demo.
