Topics

Group prompts into a buying question and cross the noise floor sooner

A topic is a named group of prompts that belong to the same buying question. Grouping changes nothing about how a prompt runs — topics are a reporting structure, not a scheduling one. What changes is the evidence: a topic pools every successful run of every prompt inside it.

Get Started Book A Demo

By Karl-Gustav Kallasmaa, founder of Attensira · Last updated 3 September 2026

In short

Topics roll per-prompt numbers up into one set of rates for a theme. The rollup is a pooled rate, not an average of averages, so a topic's sample is the sum of its prompts' successful runs. That is the point: a topic can prove a change in a week that a single prompt would take a month to demonstrate.

Reporting, not scheduling

Grouping a prompt changes nothing about when or how often it runs. The cadence is untouched.

Pooled, not averaged

A topic weights all runs together rather than averaging prompt-level rates, so a thin prompt cannot distort it.

Bigger n, sooner signal

Topic rates cross the 95% significance gate before the prompts inside them do.

Retroactive by design

Rollups are derived from stored runs, so moving a prompt recomputes both topics over existing history.

The problem it solves

One prompt on Starter gives you one reading a day. A single reading can only ever be 0% or 100%, so a prompt-level rate swings wildly for reasons that have nothing to do with your market — models are non-deterministic, and the same question asked twice can produce two different answers minutes apart.

The instinct is to look harder at the prompt. The right move is to look at a coarser cell. Ten prompts in one topic give you ten readings a day, and the same swing that was noise at the prompt level becomes a signal you can act on. Grouping is the cheapest way to buy statistical power in the product: it costs nothing, issues no extra runs, and needs no plan change.

It is also how a measurement becomes a piece of work. A weak topic points at a page you need to write, in a way a single weak prompt never does.

How it works

You name a topic, assign prompts to it, and Attensira rolls the per-prompt numbers up into a single set of rates for the group: mention rate, citation rate, and competitor rates computed across every successful run of every prompt in the group.

That rollup is a pooled rate, not an average of averages. A topic with two prompts — one that produced three successful runs and one that produced nine — weights the twelve runs together rather than averaging two percentages. This is what you want, because the topic is answering a single question: across the questions in this theme, how often were we named?

A prompt belongs to one topic at a time. Assigning it to a new topic moves it rather than copying it.
Individual prompt rows do not disappear. Grouping adds a level of reporting above them; you can still open any prompt and see its own rates, citations and source domains.
A prompt without a topic still runs and still reports. Topics are optional; ungrouped prompts simply appear only at the prompt level.
Deleting a topic does not delete its prompts. They become ungrouped and keep reporting individually.
On the wire a topic is group_id, while the get_analytics area enum value is topic and POST /v1/prompts files a prompt under one with a topic field. The published vocabulary and the enum agree; only the identifier field kept its original name.

Regrouping is retroactive. Moving a prompt between topics recomputes both rollups over the existing runs, because the rollup is derived from stored runs rather than stored per topic. Historical topic charts will therefore shift when you reorganise — so do a reorganisation in one sitting rather than drifting prompts between topics week to week, or your trend lines describe your bookkeeping instead of your market.

A worked example

Two prompts in one topic. The first produced three successful runs in the window and named you in one of them. The second produced nine and named you in four.

Cell	Runs naming you	Successful runs	Mention rate
Prompt A	1	3	33%
Prompt B	4	9	44%
Topic rollup	5	12	42%

The rollup is 5 over 12, not the midpoint of 33% and 44%. Averaging the two rates would give the three-run prompt the same weight as the nine-run one and quietly move the answer. Pooling keeps every run worth the same amount, which is the only weighting that can be defended when the sample sizes differ.

Three structures hold up in practice. Group by stage — discovery prompts, evaluation prompts, objection prompts — because discovery answers name many brands and evaluation answers name few. Group by segment, asking the same question for a solo operator, a twenty-person team and an enterprise, because models often recommend entirely different products across them and a blended number hides it. Or group by use case, one topic per problem you solve, which is the structure that maps most cleanly onto content work.

Keep topics large enough to be stable — five to fifteen prompts is a workable band — and resist a topic holding one prompt, which reports exactly what the prompt already reported.

What it does not do

Grouping does not change scheduling. Each prompt, model and country combination still runs at most once per workspace-local calendar day, at your plan's sampling depth. A topic does not buy extra readings.
A topic does not create data. A brand-new topic reports a null value until the prompts in it have successful runs — not measured, never zero.
It does not make competitor rates sum to 100%. Competitor rollups share the same denominator as your own rate, exactly as at the prompt level.
It does not stabilise a bad prompt. Pooling ten prompts that ask the same question in ten ways gives you ten times the runs on one question, not coverage of ten questions.
Regrouping rewrites history in reporting. That is deliberate, and it is a reason to reorganise deliberately rather than continuously.
There is still no visibility score. A topic reports the same honest shapes as everything else: a rate with its sample count, and a delta only when it clears a two-proportion z-test at 95%.

Which plans include it

Topics are on every plan, and grouping prompts costs nothing — no extra runs, no extra slots. Your plan sets the number of prompt slots you can fill and the sampling depth behind each one, which together decide how large a topic's n grows per day.

Plan	Per month	Prompt slots	Sampling depth	Model surfaces	Retention
Starter	$129	50	n=1	ChatGPT, Perplexity, Claude, Google AI	3 months
Growth	$349	150	n=3	ChatGPT, Perplexity, Claude, Gemini, Google AI	12 months
Business	$699	350	n=3	ChatGPT, Perplexity, Claude, Gemini, Google AI	Unlimited

Yearly billing is charged at ten months' price. Full plan detail is on pricing.

Frequently asked questions

Does putting prompts in a topic change how often they run?

No. Topics are a reporting structure, not a scheduling one. Each prompt, model and country combination still runs at most once per workspace-local calendar day at your plan's sampling depth, whether it is grouped or not.

Is a topic rate the average of its prompts' rates?

No, it is a pooled rate. All the successful runs of all the prompts in the group are weighted together, so a prompt with three runs does not count the same as one with nine. Averaging percentages would let a thin prompt move the topic as much as a well-sampled one.

Why do my historical topic charts change when I regroup?

Because rollups are derived from stored runs rather than stored per topic, so moving a prompt recomputes both topics over the existing history. This is deliberate — it keeps a topic's number consistent with its current membership. Reorganise in one sitting rather than drifting prompts week to week.

Can one prompt belong to two topics?

No. A prompt belongs to one topic at a time, and assigning it to a new topic moves it rather than copying it. Prompts left ungrouped still run and still report at the prompt level.

How many prompts should a topic hold?

Five to fifteen is a workable band. Large enough that the pooled sample is stable, small enough that the prompts genuinely ask the same buying question. A topic holding one prompt reports exactly what that prompt already reported, so it buys you nothing.

What happens if I delete a topic?

The prompts survive. They become ungrouped and keep reporting individually, with their history intact, because the readings belong to the prompts rather than to the topic.

You set the destination. It handles the route.

Attensira measures what assistants answer, finds what is holding you back, and opens a pull request you review before anything ships.

Get Started Book A Demo

Group prompts into a buying question and cross the noise floor sooner

In short

Reporting, not scheduling

Pooled, not averaged

Bigger n, sooner signal

Retroactive by design

The problem it solves

How it works

A worked example

What it does not do

Which plans include it

Frequently asked questions

Does putting prompts in a topic change how often they run?

Is a topic rate the average of its prompts' rates?

Why do my historical topic charts change when I regroup?

Can one prompt belong to two topics?

How many prompts should a topic hold?

What happens if I delete a topic?

Keep reading

Prompt tracking

Mention rate

Competitor tracking

Model surfaces

You set the destination. It handles the route.

The problem it solves

How it works

A worked example

What it does not do

Which plans include it

Frequently asked questions

Does putting prompts in a topic change how often they run?

Is a topic rate the average of its prompts' rates?

Why do my historical topic charts change when I regroup?

Can one prompt belong to two topics?

How many prompts should a topic hold?

What happens if I delete a topic?

Keep reading

Prompt tracking

Mention rate

Competitor tracking

Model surfaces

You set the destination. It handles the route.