Profound vs Scrunch: Historical Data and Prompt Volume Compared
Profound is the better fit for researching historical AI prompt demand at the individual-question level; Scrunch is better suited to monitoring a broader set of selected prompts and reviewing recent answers. Those are different jobs, and a larger prompt allowance does not mean better estimates of how often people ask each question. At Frizerly, we recommend evaluating both against the work that follows measurement: identifying useful customer questions, publishing clear answers, and tracking whether search and AI visibility improve.
There is a second distinction buyers often miss. Historical prompt demand describes past estimates of what users asked AI assistants. Historical brand visibility describes answers collected while a tracker ran your selected prompts. Profound offers a historical demand research product, but its visibility tracking does not retroactively show how your brand performed before you began tracking a prompt. Scrunch likewise builds a history from monitored activity, with the available reporting window depending on how the data is accessed.
Our assessment: Choose Profound when the decision hinges on discovering exact questions and studying demand over time. Choose Scrunch when you already have a prompt portfolio and need to inspect answers, citations, and competitor appearances regularly. Choose a content execution workflow when your main challenge is turning those findings into pages customers can find.
Historical data and prompt volume: the terms that matter
AI visibility platforms can use the word prompt for two very different things. One is an actual question a person asked an assistant. The other is a question your company configures for repeated testing. Understanding which one a dashboard measures prevents costly mistakes in content planning.
- Prompt demand: An estimate of how often people ask an AI assistant a question or related questions. This is market research, not a count of your site's traffic.
- Tracked prompt: A question selected for recurring tests against an answer engine. It tells you what the assistant returned when the test ran, not how many consumers asked that question.
- Response volume: The number of AI answers a monitoring platform collects. It can rise when a prompt is tested across several engines, markets, or configurations, even though the underlying set of questions stays the same.
- Historical demand: Past estimates from a prompt-research dataset, which can help reveal changing interests and question wording.
- Historical visibility: Your recorded brand appearances, citations, or positions across repeated tests. This history normally starts when monitoring starts.
Suppose a team monitors the question, “Which software helps a small clinic answer patient questions online?” across several AI assistants. Its response count describes how many test answers were collected. It does not tell the clinic how many prospective patients asked that question. To assess demand, the team needs a separate research method. To assess results, it needs to inspect the answers and the business outcomes that follow.
Profound vs Scrunch at a glance
| Question | Profound | Scrunch |
|---|---|---|
| What is its strongest use here? | Researching individual prompts and historical estimates of AI question demand. | Monitoring selected prompts, inspecting answers, and grouping visibility findings. |
| Does it estimate prompt demand? | Prompt Volumes estimates demand for individual prompts and shows related research dimensions. | AI Search Trends provides broader topic-level demand signals; tracked prompts measure answer visibility. |
| What historical distinction matters? | Historical demand research and accumulated tracking history are separate datasets. Adding a tracked prompt does not backfill earlier brand visibility. | Historical visibility reflects monitored responses. Its change-over-time reporting describes a window of up to 12 months or since account creation, whichever is shorter. |
| How much can a team monitor? | Published plan information describes 50 tracked prompts on Starter and 100 on Growth, with tailored enterprise capacity. | Published Core plan information describes 125 active unique prompts and 5,000 responses per month, with custom enterprise capacity. |
| How do routine tests refresh? | Tracked prompts run daily. | Its collection FAQ describes daily checks for newly added prompts for an initial period, then a default refresh every 72 hours. |
| Who should pay closest attention? | Teams sizing opportunities and studying exact question wording. | Teams organizing a larger recurring monitoring set across their existing priorities. |
Platform details: Profound Prompt Volumes, Profound's prompt management guide, Scrunch's change-over-time FAQ, Scrunch's collection FAQ, and the platforms' published plan pages. Limits and availability can vary by plan and configuration.
Profound: historical demand research versus tracked visibility
What Prompt Volumes contributes
Profound's Prompt Volumes is a research tool for examining questions people ask AI assistants. It uses anonymized consumer conversation data and modeling to estimate demand. Its value is not merely showing that “AI SEO software” is a topic. It is helping a marketer distinguish specific questions within that topic, such as how a business tracks mentions, how it creates answer-ready pages, or how it compares competing approaches.
That distinction changes editorial decisions. A broad topic can hide different intents: someone defining a term, someone troubleshooting a workflow, and someone evaluating a purchase may all use similar words. Prompt-level research can help a team select questions whose phrasing and intent correspond to the products it actually sells. Profound also describes related prompts, intent categories, and trends as part of its research environment.
For US research covering ChatGPT, Gemini, and Perplexity, Profound describes historical Prompt Volumes coverage beginning in January 2025. That start point refers to its demand research dataset, not a retrospective record of every brand's visibility. It is particularly useful when a team wants to ask whether interest in a category changed before and after a launch or whether buyers shifted from general education toward comparisons.
What its tracking history does not mean
Profound separately lets teams configure prompts for recurring answer-engine monitoring. Its prompt tracking feature examines answers and visibility on the selected prompts. That monitoring history begins after a prompt is activated; it is not the same as querying an existing archive of every historical answer to an arbitrary question.
This matters when someone asks, “What did ChatGPT say about us before we signed up?” A historical demand chart can show what kinds of questions people were asking during an earlier period. It cannot establish whether your brand appeared in past answers that your organization never monitored. For a credible before-and-after analysis, begin monitoring before a major content change whenever possible, keep the prompt wording consistent, and record the publication dates of relevant pages.
Where prompt limits affect the analysis
Profound's published plan information describes 50 tracked prompts on Starter and 100 on Growth, with enterprise configurations tailored to larger requirements. The number of tracked prompts is a capacity limit for recurring testing. It is not a ceiling on the historical demand questions a researcher can investigate through Prompt Volumes.
A team with several products can exhaust a tracking allowance quickly if it treats every wording variation as a separate must-track question. Start with a compact set representing distinct buyer intents: discovery, comparison, purchase, and implementation. Keep the rest as research candidates until they justify a place in recurring reporting. Profound's prompt management guidance describes how prompts are created, managed, and tagged.
Scrunch: monitored answers and topic trends
What a larger prompt portfolio buys you
Scrunch's Core plan describes 125 active unique prompts and 5,000 responses per month. Those figures represent different constraints. The first limits the distinct questions you can keep active; the second limits collected answers. If your team tracks the same questions across multiple supported answer engines, each question can lead to multiple responses. A buyer should assess both allowances against the monitoring design, rather than reading “125 prompts” as a complete measure of coverage.
That monitoring approach suits a company that already knows its important customer journeys. Prompts can be organized around topics, personas, and stages of consideration, making it easier to spot where a competitor appears and where your brand is absent. Scrunch's monitoring guidance explains this kind of answer and citation analysis.
How far back can you look?
Scrunch describes change-over-time reporting with up to 12 months of visibility in its interface, or the period since account creation if that is shorter. Its API FAQ describes a 90-day history window for its Query and Responses APIs. Interface history and API availability therefore should not be treated as interchangeable. If your organization relies on a data warehouse for long-range reporting, plan its collection and storage process at the start.
Scrunch's collection FAQ describes a faster initial schedule for newly added prompts and a default refresh every 72 hours afterward, with on-demand refresh available. That cadence matters when interpreting small changes. If a brand publishes a new comparison page today, a lack of change in tomorrow's monitoring view is not a sound judgment on the page. First ask when the relevant prompt last ran, which engine returned the answer, and whether the page has had time to become discoverable.
Why a topic trend is not an exact-prompt estimate
Scrunch's AI Search Trends helps teams notice broader patterns of interest. A topic such as “contact lens problems” can be relevant even when no single question accounts for the whole trend. For article planning, however, a topic trend does not tell you which precise patient question to answer first. Use it to identify a theme, then examine customer language and the actual answers being returned before assigning a page.
There is no universal winner between topic signals and exact-prompt estimates. Topic signals are useful for deciding where to investigate. Exact-prompt estimates are useful for selecting language and separating intents. Tracked responses are useful for checking whether AI assistants mention a brand when a selected question is asked. The strongest content process keeps those functions separate.
How we compare prompt coverage without confusing the numbers
At Frizerly, we use a simple sequence to assess any AI search measurement plan: demand, coverage, evidence, action, outcome. Each stage answers a different question, and none should be substituted for the next.
- Demand: Are people asking a relevant question? Examine available prompt research or broader topic signals, then check whether the question expresses a problem your business can solve. An estimated large audience for an unrelated question is not a content opportunity.
- Coverage: Are you testing the questions that matter? Separate a core recurring prompt set from a research backlog. Include a mix of early research and decision-stage questions. Count active prompts, engines, markets, and response runs before selecting a monitoring configuration.
- Evidence: What did the assistant actually return? Read the answer, check whether the brand is named, and distinguish a brand mention from a cited source. Also note which competitor appears and whether the assistant's description is accurate. For more on this distinction, see our guide to AI citations, mentions, and competitor recommendations.
- Action: Which page can address the gap? Match the missing answer to a useful page: a direct FAQ, an evidence-led comparison, a product guide, or a specific troubleshooting article. Avoid producing several near-identical pages for minor wording variants.
- Outcome: Did visibility and business performance change? Compare repeat AI observations with search impressions, organic clicks, and relevant inquiries. Do not infer sales from a prompt-volume estimate or assume one AI citation guarantees traffic.
This framework also explains why Frizerly can be the practical next step for a team with plenty of monitoring data but a publishing bottleneck. Its SEO and AEO workflow connects to a business website, uses website and Google Search Console information to inform topics, publishes relevant articles, refreshes older content when rankings decline, and tracks brand mentions across ChatGPT, Claude, Gemini, and Perplexity. The point is to connect a question worth answering to an actual page, then revisit performance.
Which platform fits each decision?
Frizerly: content execution
Best when the team needs to turn customer questions into published, answer-focused pages and monitor search and AI brand visibility. Website learning, scheduled publishing, Search Console integration, and content refreshes connect insight with action.
Profound: demand research
Best when exact prompt phrasing and historical demand estimates drive the decision. Its research product and recurring visibility tracking serve separate purposes, so plan for both if you need both.
Scrunch: prompt monitoring
Best when a defined set of questions needs regular answer inspection and competitive review. Assess both active-prompt and response allowances, plus the history available through your preferred reporting route.
These are choices by primary workflow, not claims that a single score can rank unlike products. If your most important question is “Which exact AI questions merit investment?”, Profound's prompt-level research is particularly relevant. If it is “What do assistants currently say about us across our chosen questions?”, Scrunch's monitoring workflow is relevant. If it is “Who will keep publishing useful answers and updating them as we learn?”, Frizerly is our recommendation.
What real search results tell us about question specificity
Prompt research is useful only when it guides content that helps someone. Our published client examples show why specificity matters, although their reported results are Google organic clicks, not AI prompt volumes or AI citation counts.
In our optometry case study, a practice published 360 articles answering concrete patient questions. In the reported last 30 days, 72 blog pages received clicks and blog content generated more than 600 organic clicks per month. The top five articles accounted for 425 clicks. They addressed contact lens add powers, prescription conversion, progressive lenses, cloudy contacts, and toric lens rotation rather than repeating broad “eye care” advice.
Our building-materials retailer case study shows a similar pattern in a different category. After 276 posts over approximately 12 months, 136 posts received clicks during the reported 30-day period. The top 25 generated more than 3,600 organic clicks per month. Strong performers answered project-sizing, material comparison, installation-depth, and regional climate questions.
Neither example proves that a particular AI prompt-volume estimate will produce the same outcome. They do show a useful editorial principle: precise, consequential questions give a page a clear job. Whether a team discovers those questions through prompt-demand research, monitored answers, Search Console, or customer conversations, the page still needs to answer the question accurately and make the next step clear.
Best practices for historical AI visibility reporting
Keep demand and brand performance in separate charts
A rising estimate of category interest is not proof that your brand gained visibility. Label demand as estimated audience behavior and tracked responses as observed test results. When presenting results, show the question, engine, market, observation period, and any change to prompt wording. That context makes a trend more interpretable than a single aggregate score.
Preserve a stable comparison set
If your tracked prompts change constantly, you cannot cleanly compare this month's visibility with an earlier period. Maintain a stable core set for longitudinal reporting and a separate exploratory set for new questions. When a product launch changes the language buyers use, add relevant prompts, but keep their start dates visible in your reporting.
Read the answer, not just the dashboard
An assistant may mention your business while recommending someone else. It may cite your article without naming your company prominently. It may also answer from a page that no longer reflects your current services. Reviewing the full answer helps identify whether the problem is missing coverage, inaccurate positioning, or an outdated source. Our guide to citation rate versus brand mention rate explains why those measures deserve separate attention.
Treat small changes as signals to investigate
Generated answers can vary, and monitoring tools sample selected questions at particular times. A change on a single prompt is a reason to inspect the response and timing, not to announce a durable win or loss. Look across a related group of questions, compare the actual language of the answers, and check whether the same issue appears in search performance.
Prioritize pages that can answer a real buying question
A useful page states the answer early, explains the conditions under which it changes, and gives readers enough detail to evaluate the advice. For commercial queries, include meaningful comparisons and limitations rather than repeating claims about being the best. For technical queries, address the specific term, product, or situation the customer named. Then review the page as customer questions evolve.
Frequently asked questions
Does Profound have more historical data than Scrunch?
For historical prompt-demand research, Profound offers a separate dataset with US coverage for ChatGPT, Gemini, and Perplexity described from January 2025. For tracked brand visibility, both platforms accumulate observations from monitoring activity. Profound's older demand data should not be mistaken for backfilled historical brand appearances. Scrunch describes up to 12 months of change-over-time visibility in its interface, limited by account age.
Does Scrunch's higher tracked-prompt allowance mean it has better prompt-volume estimates?
No. A tracked-prompt allowance describes how many questions can be monitored repeatedly. Prompt volume describes estimated user demand. Scrunch's Core information describes 125 active prompts and 5,000 responses monthly, while its trends focus on broader topics. Profound's Prompt Volumes focuses on demand for individual questions. Those numbers answer different questions.
Can either tool tell me exactly how many people saw an AI answer mentioning my brand?
Tracked answer collections show what an assistant returned when selected questions were tested. Demand estimates describe modeled interest in questions or topics. Neither figure is, by itself, a count of people who saw your brand in an assistant's response. Use them as distinct inputs to a broader measurement process.
What is the difference between a prompt and a response?
A prompt is the question submitted to an assistant. A response is the answer returned. One tracked prompt can generate multiple monitored responses when it runs repeatedly or across different engines. This is why a response allowance should not be read as a count of unique customer questions.
How should a team choose its first monitored prompts?
Start with questions that represent different customer decisions, not just spelling or wording variations. Include the problems buyers need to solve, the alternatives they compare, and the details they ask before choosing. Keep that core set stable enough to assess change over time, then use a separate research backlog for emerging topics.
Can historical AI prompt demand predict organic search traffic?
No. AI prompt demand and Google search clicks measure different behaviors. A valuable AI question may inspire a useful article, but its demand estimate does not predict that article's rankings, citations, clicks, or conversions. Review search performance and actual customer inquiries alongside AI visibility.
What should we do after finding a question where a competitor appears and we do not?
Read the full answer and inspect the sources. Then determine whether your site lacks a relevant page, has an incomplete answer, or describes the offering unclearly. Publish or update the page that best resolves that gap, and continue monitoring the same question to see whether the answer changes.
Conclusion
Profound leads this comparison for historical, individual-prompt demand research. Scrunch is a strong fit for managing a broader recurring prompt set and reviewing monitored answers. The crucial distinction is that research history, monitoring history, prompt limits, and response limits are four different measures. Compare them separately before choosing a workflow.
We recommend beginning with the customer questions that matter, building pages that answer them, and checking both AI appearances and organic results over time. For teams ready to make publishing and measurement part of the same routine, Frizerly connects SEO and AEO content creation with visibility tracking.