New in Scrunch: Find your prompt-tracking Goldilocks zone with Topic Prompt Optimizations

TL;DR - Track fewer prompts without sacrificing AI search coverage. - Identify which prompts deliver unique value and cut the ones that don’t to reduce cost, noise, and redundancy. - Customize your strategy with flexible configurations and see the impact with data-backed visualizations.

We built a feature that might shrink your bill—and that's the point. Most AI search platforms (including us) price by prompt. This can nudge teams toward over-tracking. More prompts don't always mean better insight. They often mean overlapping results, higher costs, and noisier signals.

Topic Prompt Optimizations is designed to right-size your monitoring so every prompt pulls its weight.

With Topic Prompt Optimizations, Scrunch helps you: - Identify which prompts contribute unique value and which are redundant. - Develop the optimal set of prompts to monitor for each key topic. - Tune your strategy to your KPIs and risk tolerance—with transparent, data-backed recommendations.

How it works

Step 1: Open Topic Prompt Optimizations

Go to the Insights tab in Scrunch and select Topic Prompt Optimizations. Choose a topic (surfaced from your AI Context) and the AI platforms to analyze. By default, Scrunch evaluates the past 14 days across all active models.

Step 2: Run the optimization

Click Run optimization. Scrunch analyzes your prompts for that topic and returns:

The full list of tracked prompts mapped to the topic
Which prompts to keep vs. cut
Anticipated impact on coverage, resilience, brand presence, answer position, and sentiment

Step 3: Review the rationale

Recommendations balance two core signals:

Coverage: the percentage of unique URLs cited by at least one prompt
Resilience: the percentage of URLs cited by two or more prompts

If new prompts introduce new URLs, you are expanding visibility. If they return the same URLs, you are paying for overlap.

Step 4: Act with confidence

Archive suggested prompts directly from the results page. Drill into any prompt for personas, tags, and funnel-stage context.

Export a CSV containing:

Keep vs. cut recommendations
Closest replacement prompts for any removals
The configuration used to generate your results

If your setup is already efficient, you will see "Pruning not recommended." That is a good outcome—no changes needed.

Advanced controls and transparency

Under the hood, recommendations are generated via a Pareto frontier. This is the set of optimal prompt combinations where you cannot improve coverage without hurting resilience—and vice versa.

Configuration Option	What It Does	Default
Smart defaults	Balances coverage and resilience equally	On
Coverage floor	Minimum coverage threshold; below this, core metrics (brand presence rate, citation share, sentiment) fluctuate meaningfully	95%
Advanced Configuration	Adjust constraints, change the coverage floor, or apply different pruning strategies; click Apply changes to update results instantly	Off
Data Deep Dive	Interactive visualizations showing how different prompt sets affect coverage and resilience	Available on demand

Practical guidance to maximize ROI

Overshoot at first. Start with more prompts than you think you need. More initial data makes optimization more accurate, and trimming is easy.
Let data accrue. Run prompts for at least two weeks before pruning. You need enough signal to separate high-value prompts from redundant ones.
Watch the coverage curve. A flattening coverage curve means new prompts are no longer discovering new URLs—a strong signal it is time to prune. Avoid aggressive pruning if every prompt still adds unique URLs.
Reallocate budget to blind spots. Archive prompts that do not add meaningful new coverage, then reinvest that budget into under-covered topics.

What you’ll gain

Lower costs with maintained (or improved) coverage
Less noise and faster analysis for your team
More stable KPI tracking across brand presence, citation share, and sentiment
A defensible, auditable prompt strategy you can explain to stakeholders

Get started fast

Try it free. Start a 7-day free trial of Scrunch's self-serve Core plan and begin monitoring immediately. Start your free trial. Trial users can set up monitoring, analyze brand presence and sentiment, and access Insights to uncover optimization opportunities.
Quick time to value. Implementation is fast—many teams, including large enterprises, are deployed and collecting data within one day. Learn more on the Monitoring & Insights overview.

If you’re evaluating for an enterprise: free trials are limited to a single brand and don’t include certain Enterprise features (SSO, enterprise data API, etc.). If you need a broader evaluation, book a demo with our team.

Ready to right-size your monitoring and boost ROI? Request access to Topic Prompt Optimizations via support or your CSM, start your 7-day free trial, or book a demo to see it in action.