What are good benchmark numbers?
Benchmarks are highly industry and company-specific, and they can shift with seasonality, news cycles, model changes and perhaps more importantly, the types of prompts you're measuring. So treat any “good number” as a starting point, not a goal.
Looking for the exact Explorer name or API field for a metric? The Scrunch Metrics Reference lists every metric, breakdown and filter with the API field it maps to.
Here are a few general ranges that often show up for strong brands. Each one is labeled with its Explorer picker name and API field so you can find it in your own views:
Mentions Rate (API field: brand_presence_percentage): Top brands often appear in 50%+ of relevant answers in a topic (assuming prompts are high-intent).
Position Average (API field: brand_position_score): Leading brands often average a 50+ position score, where top=100, middle=50, and bottom=0 (depending on how your scoring is defined and normalized).
Sentiment Average (API field: brand_sentiment_score): Leading brands often average a sentiment score in the 80s or higher, since sentiment is scored positive=100, mixed=50, negative=0, so mixed responses only count as half credit. Track the negative share closely, since small increases can matter.
Citations Share of Voice (API field: brand_citation_share_of_voice): 2% to 3% of total citations can be solid, especially if the citations are on high-intent prompts.
Explorer's rate and average charts show the sample size behind each number in the tooltip, for example "43% (37 of 86 responses)". Check this before comparing a benchmark against a small prompt set.
Better approach (recommended):
Use benchmarks like meteorologists use “normals.” Build a stable prompt set, keep it consistent, and track:
Change vs. your own baseline (rolling 8 to 12 weeks works well)
Peer-relative movement (competitors on the same prompts, same week, same surface)
Anomalies (what moved beyond what is typical for this topic and time)
This gives you a read on real improvement, without overreacting to one-off spikes or broad platform shifts. When you build these views in Explorer, keep in mind it supports up to 5 metrics, 3 breakdowns, and 8 filters in a single view.
"How do we explain low presence rates?"
Low Mentions Rate (API field: brand_presence_percentage) below 10-20% can indicate:
Prompts are too broad or not relevant to brand
Brand is legitimately small/new in the space
Competitive landscape is very saturated
Content optimization needed
Use as diagnostic, not just reporting metric.
"Should we track brand positioning heavily?"
No, it's useful to monitor for unusual trends, but don't over-index. Similar to SEO, drop-off is high for lower positions. Most users don't read entire lengthy LLM responses.
"How often is data refreshed?"
Prompts are monitored daily for the first two weeks, then collected every 72 hours (3 days) thereafter. This explains why you may see date gaps in exported data. You can manually trigger a new response collection anytime when viewing a prompt variant in Explorer.
"What's the difference between presence and citations?"
Presence (Mentions Rate, API field: brand_presence_percentage) means your brand is mentioned anywhere in the response text. Citations (Citations Rate, API field: brand_citation_rate) mean your domain is explicitly referenced as a source with a URL. Think of citations as your "authority score," AI trusts your content enough to link to it directly. See Understanding Citation Metrics in Scrunch for the full breakdown.
"How is position calculated (top/middle/bottom)?"
Position reflects where your brand is first mentioned within the AI response, top, middle, or bottom. That bucket feeds into Position Average (API field: brand_position_score), a 0-100 score where top=100, middle=50, and bottom=0. This matters because users typically focus on recommendations at the top of responses.
"How far back can I access historical data?"
Up to 1 year of data is available in the platform and via API.
"Why do I see gaps in my data when I export?"
After the first two weeks, prompts are collected every 72 hours (3 days) rather than daily. The platform's trend views smooth this data, but raw exports will show these gaps. Use weekly or monthly aggregation for cleaner trend analysis.
"Why are my dashboard metrics different from Explorer's prompt-level view?"
Dashboard metrics show rolling 7-day averages by default and aggregate across all active prompts. Explorer (using a Prompt breakdown) shows individual prompt performance. Both are accurate, just different views of the data.
"What does 'competitive presence' measure exactly?"
Competitive presence (API field: market_presence_percentage) measures the pooled, volume-weighted share of responses that mention your competitors by name, across your brand plus every active competitor. It helps identify where competitors are winning name-mention visibility that you're not. Citations of a competitor's domain don't count toward this metric; that's tracked separately by Citations Rate (API field: competitor_citation_rate).
"Is high third-party citation percentage bad?"
No, 96%+ third-party citations is completely normal and healthy. AI models heavily favor independent, authoritative sources for credibility. Focus on getting your brand mentioned in those high-authority third-party sources.
One caveat: Grok no longer contributes citation data, so it won't show up in citation metrics. Grok's presence, position, and sentiment metrics are unaffected.