Skip to main content
This page explains how DevTune calculates each Visibility metric and how to read it: what a number covers, its range, and when a change or a lead is clear.

Core metrics

The Dashboard scorecards show Presence Rate, Average Position and Sentiment. Share of Voice appears as the SoV column of the Market comparison table and as a metric in the Trend chart.

Presence Rate

Your primary metric for AI search visibility. The Trend chart calls it Overall Presence Rate. Definition: The percentage of completed AI answers with citation data that cite at least one URL tracked to your brand. An answer counts once, whether it cites one tracked URL or several. Completed answers without citation data are left out of the denominator. Calculation:
Example:
  • 100 completed answers with citation data across all platforms
  • Your tracked URLs were cited in 45 of them
  • Presence Rate: 45%
Each Presence Rate comes with a 90% range, and the Dashboard says whether you lead the most visible competitor. To see whether you are ahead, read the lead line rather than a fixed threshold. See Ranges, leads and clear changes.

Share of Voice

Your citations as a share of your citations plus your tracked competitors’ citations. Definition: The proportion of citations to you or your tracked competitors that are citations to you. Mentions without a link are not counted, and neither are citations of any other source, so this is your position against the competitors you chose, not your share of everything cited. Calculation:
Example:
  • Your citations: 120
  • Competitor A citations: 80
  • Competitor B citations: 50
  • Total: 250
  • Your Share of Voice: 48%
Interpretation:
  • Compare to your actual market position. If your market share is 30% but your AI search Share of Voice is 15%, there is room to improve AI visibility.
  • Track Share of Voice trends to see whether you are gaining or losing ground relative to competitors.

Sentiment

How positively or negatively AI platforms describe your brand. Definition: The average tone of AI answers that mention your brand, on a scale from −1 (very negative) to +1 (very positive). The card shows the score, such as +0.35, a label, and how many answers it rests on, such as 30d · 212 scored answers. Sentiment isn’t available with platform or topic filters, and always counts every topic. Sentiment history starts on 20 September 2026: earlier days are withheld, so compare only periods from that date on. Interpretation:
  • Negative sentiment may indicate known issues, common complaints, or unfavorable comparisons
  • Sudden sentiment changes warrant investigation

Average Position

Definition: Your average position among cited sources, counted only in answers that cite you. A position of 1 means you are the first-cited source. Lower numbers are better. The Dashboard card shows how many citations the average rests on, and the Market comparison shows it as Avg #. Usage: Average Position shows how prominently your content is featured among sources. A low average means AI platforms cite you early.

More metrics

These metrics add detail behind the core metrics. Hover the Presence Rate ring on the Dashboard to see Presence Breakdown by Home, Docs, Blog, GitHub, Packages and Other.

Docs Presence

Definition: The rate at which citation sources tagged as “docs” appear in AI responses. This measures how well your documentation is recognized and cited by AI platforms. Usage: Compare Docs Presence to overall presence. If your docs are cited less than your marketing site, it may indicate documentation quality or discoverability issues.

Blog Presence

Definition: The rate at which citation sources tagged as “blog” appear in AI responses. This measures how well your blog content is surfaced by AI platforms. Usage: Blog presence indicates whether your content marketing is reaching AI-mediated audiences. High blog presence suggests your blog content is authoritative and well-indexed.

Brand Mentions

Definition: The rate at which AI responses contain your brand terms (name mentions) regardless of whether citation URLs are present. Calculation:
Unlike Presence Rate, Brand Mention Rate divides by all completed answers, including those without citation data. Usage: Brand mentions capture visibility that citation-based presence intentionally does not include. If an AI response says, “You could use DataForge” without linking to dataforge.io, the Brand Mention metric records it, while Presence Rate does not.

Top 3

Definition: The share of your citations that sit in the top three cited positions of an answer. The Market comparison shows it as Top 3 for each brand. Calculation:
Usage: A high Top 3 means AI platforms cite you among their first sources, not just somewhere in the list.

Primary Citation Share

Definition: The proportion of all citations that point to your primary citation sources versus competitor citation sources and other sources. The Prompts table shows it per prompt as Citation. Calculation:
Usage: Primary Citation Share measures how much of the citation real estate belongs to you. This is complementary to Share of Voice, which leaves out citations of sources other than you and your tracked competitors.

Ranges, leads and clear changes

Every visibility rate rests on a sample of AI answers, so DevTune shows how far each one can be trusted. A Presence Rate from 10 answers is much less certain than one from 200. This section explains the range under a rate, the Directional and Unmeasured labels, the lead line, and the change labels.

The 90% range

A rate’s range is where its true value most likely sits: DevTune is 90% confident the rate lies inside it. The Presence Rate card shows the range under the rate, followed by the number of answers the rate rests on:
The answer count is the rate’s denominator. For Presence Rate, it is the completed answers with citation data in the selected window. In the Market comparison, each brand’s range sits under its Presence, and the Answers column gives the count.
  • A narrow range means the rate is stable. A wide range means more answers are needed.
  • Ranges narrow as runs accumulate.
  • To compare two rates, read the lead line or the change label rather than comparing ranges by eye. They apply the test described below.
DevTune computes ranges with the Wilson score interval, and compares two rates with Newcombe’s interval for the difference between them.

Directional

A rate that rests on fewer than 10 answers is marked Directional, and the Presence Rate ring turns grey. Treat it as a hint about direction, not a figure to report.

Unmeasured

Unmeasured means the window holds no answers with citation data for that rate. The Presence Rate card adds “No answers with citation data · N days” for the selected window. Widen the date range, or wait for the next runs to complete.

The lead line

Under Presence Rate, the lead line compares your Presence Rate with the competitor whose Presence Rate is highest. DevTune makes this one comparison only. A lead is clear only when the 90% range on the difference between the two rates excludes zero. N is that difference in percentage points. The readout under the scorecards uses the same rule for its title. A raw gap of several points can still read Level with, when either rate rests on few answers. There is no lead line when no tracked competitor has a measured Presence Rate in the window.

Change against the previous period

Where a rate shows its change against the previous period of the same length, the change is in percentage points:
  • An arrow and colour appear only when the change is clear at 90%: the range on the difference excludes zero.
  • Otherwise the points are shown muted, and the tooltip reads No clear change.
  • Not compared means one of the two periods has no answers for that rate. Its reason reads Not measured in both windows.
Accuracy’s Match your site change and your brand’s column in the Positioning statements matrix use this rule. The API and MCP server return the same ranges and change verdicts. See Visibility ranges and unmeasured values.

Date ranges

Every rate covers the date range you select, ending today. Each view offers its own ranges: A wider range rests on more answers, so its range is narrower. Use the change label, not single runs, to judge whether a rate moved.

Filters and per-prompt metrics

On the Dashboard, the Platforms and Topics filters apply to Presence Rate, Average Position, the Trend chart and the Market comparison. Two cards behave differently:
  • Sentiment isn’t available with platform or topic filters.
  • Positioning appears only with no topic filter and no more than one platform.
To compare platforms, select one platform at a time. See Platform Comparison. The Prompts table shows each prompt’s Presence, Mention (Brand Mention Rate), Citation (Primary Citation Share) and SOV. Use it to find the prompts with the strongest presence and the ones where competitors lead.

Understanding Metric Relationships

High Brand Mentions, low Presence Rate

AI answers name your brand often but rarely cite your URLs. This may indicate:
  • AI platforms know about your brand but do not reference specific pages
  • Content is recognized but not structured for citation
  • Brand awareness is stronger than content depth

High Primary Citation Share, low Presence Rate

The answers that cite you cite you many times, but few answers cite you at all. This indicates:
  • Strong content quality for covered topics
  • Gaps in topic coverage
  • Focus on expanding prompt coverage

Platform Divergence

Large differences between platforms suggest platform-specific content opportunities. A platform where you are weak may respond to different content approaches than your strong platforms.

Advertising

Advertising is measured separately from organic Presence. The Advertising card on the Dashboard shows three rates over the responses where ad capture was measured:
  • Any advertiser - the share with any ad
  • Your brand’s name - the share with an ad for your brand
  • Competitors - the share with an ad for a tracked competitor
Each response counts once in each rate, even if it contains several ads. Ad capture covers Google AI Mode and ChatGPT. Responses without ad data, and other platforms, are unmeasured and left out; the line under the rates says how many of the window’s responses were measured. These rates describe ads detected by our provider, which may not capture every ad displayed by a platform. Tracking starts when ad capture is enabled. Earlier responses are not repaired or counted as ad-free. The card uses the last 30 or 90 days, as labelled. Retained response views show the captured cards under Sponsored ads. Paid links are excluded from organic measurement.

Read these metrics over the API

The API and MCP server return the same metrics, ranges and change verdicts as the app. To inspect individual answers rather than rates, use List recent AI responses, List AI response history and Get a retained AI response. See Visibility ranges and unmeasured values for how the API reports ranges.

Next Steps