Reading your report | Contentstack Canoe

Find your way around

Open Brands, pick a brand, then a row in GEO Monitors. The dashboard opens on Overview.

Header: the brand, Persona and Product chips and Download Report (PDF).
Context strip: location, the engines that ran and Trend window (Last 5, Last 10, Last 15 or Last 30 runs).
Left menu: Measure (Prompts, Competitors, Sentiments), Diagnose (Citations, Site Audit), Act (Recommendations) and History (Execution History).

A first read in ten minutes

Overview. Note the four headline numbers and the lowest engine on the trend chart.
Prompts, filtered to Gaps. Open two or three misses and read what the engines said instead.
Competitors. Find who wins most of what you lose. That's your benchmark.
Citations. The biggest third-party domains are the sources you need to be on.
Recommendations. Assign the first one or two. The rest can wait a run.

Overview

The latest run, with trends over your window. Metrics defines each tile.

The four headline tiles, each with the change since the last run.

AI Visibility Score Trends

One line per engine plus the total. Hover a point for every engine's score.

Read the shape first. Lines moving together mean the engines agree. One engine well below the rest usually leans on different sources: check it in Citations by Engine.

Brand Ranking: your rank among the competitors. Competitor trends opens the Competitors tab.

Score Anatomy

A heatmap of Visibility, Competition, Mention rank and Sentiment, one column per run, latest outlined, in four color bands.

Tip

Score dropped? Find the row whose shade changed. Visibility sends you to Prompts; Sentiment sends you to Sentiments.

Visibility by Funnel Stage

Each ring is the share of that stage's questions that name you. They don't sum to 100%.

Full trend opens the stage trend on Prompts.

Top domains and top recommendations

Top Domains by Citations and Top Recommendations (the first three fixes) link to their full tabs.

Prompts

Every question, which engines named you and who they named instead.

A check means that engine named you. The last column lists the competitors it named.

Stat strip: prompts, engines, runs and "You appear in" (visibility for the selected run).
Stage tiles: visibility and question count per stage.
Filters: EXECUTION picks the run, STAGE a stage and STATUS Appears or Gaps.
Expand a row for each engine's full answer, with your brand and competitors highlighted and cited links inline.

An expanded row: two competitors named first, and the reason why.

Below the table, Visibility by Model shows each engine's mentions by stage (ChatGPT shows as OpenAI), and Visibility by Funnel Stage tracks each stage and calls out the biggest gap.

Tip

Read the winning answers, not just the checks. Why a competitor gets named ("longer range", "better reviews") is usually right there. That's your missing content.

No response captured from this engine for this prompt in this run means the engine returned nothing. It isn't a content gap.

Competitors

How you compare with the brands the engines put in your market, as they perceive it.

Your rank #3 of 10Share of voice 16%Sentiment Positive 72

Share of Voice vs Sentiment

Top right is winning: a big share of the conversation, framed positively. Illustrative data.

High sentiment with low share: the engines like you but rarely bring you up (a visibility problem). High share with low sentiment: you're talked about but framed poorly (a sentiment problem).

The Brand Ranking table

Columns #, COMPETITOR, SHARE OF VOICE, SENTIMENT, VISIBILITY, OVERALL and AVG POSITION. Rows sort by overall score. AVG POSITION is perceived, from the engines' market leaderboard, with a dash where an engine didn't rank a brand.

Competitor trends

Brand Ranking Trends: rank over time, #1 at the top.
Mention Trends: every appearance of each name, passing mentions included.
Share of Voice Trends: each brand's slice. Yours can dip when a rival rises, even if your numbers held.

Sentiments

How the engines describe you, in their own words. Sentiment is a model's read, so check it here.

Overall tone, a score per engine and the topics the engines raise. Illustrative data.

Overall sentiment: the blended label and score, plus one tile per engine.
Sentiment Trends: overall and per engine, 0 to 100 where 50 is neutral.
Sentiment Drivers: per engine, its tone, score and the verbatim strengths and concerns.
Topics by Sentiment: themes the engines raise. Expand one for quotes and your share of voice on it.

Note

A concern on three or four engines is a pattern worth fixing. On one engine, check its cited sources first: it often traces back to a single review or forum thread.

Citations

The sources the engines cite, and who owns them.

Citation share by owner and the most cited domains. Illustrative data.

Filters: Buyer Stage and a scope of This run, Last 5, Last 10, Last 15 or Last 30. The stat strip counts citations and distinct sources.
Citation Share by Owner: your domain, competitors and third-party.
Citations by Engine: how much each engine cites, and whose sources.
Sources by Engine: up to 200 cited pages by engine, with owner and total. Filter by domain or URL; darker cells mean more citations.
New and lost sources: compared with your previous run.

Tip

Make three lists: your pages that already earn citations (keep them current), third-party pages that cite competitors but not you (your outreach list) and engines that barely cite you (focus there first).

Site Audit

Whether AI crawlers can reach and read your site, on every plan. It has its own AI Readiness Score out of 100, separate from the Overall Score, and every finding says why it matters and how to fix it.

Site audit explains the checks and how to act on them.

Recommendations

Under What to do next: up to ten ranked fixes. Each card has a rank, category, target metric, title, the measured reason (WHY?), one to three steps (HOW) and a link to the evidence.

Acting on recommendations has a playbook for every category.

Execution History

Every run, newest first. Columns: RUN, AI VISIBILITY (the Overall Score), ANSWER ENGINES, STATUS and PDF.

Weekly runs of a fictional monitor. On Sep 17 Claude was unavailable, so that run is Partial and scores the three engines that answered.

Success: Every engine on your plan answered.
Partial: At least one engine was unavailable. Scores cover those that answered, and some sections may reuse data from a prior run.
Failed: No engine responded, so there's no visibility data.

The per-run report

Select a row to open that run: Overview (At-a-Glance Report Summary, a "What AI thinks about your brand" snapshot, discovery keywords and the questions), Visibility (each engine's key strength, growth area and trajectory), Competition, Sentiment, Buyer Journey, Topic Sentiment and Sources. Metrics maps its labels.

Free keeps the current report, Growth six months. Free covers ChatGPT and Gemini, leaderboard, sentiment, buyer journey and citations included; recommendations past your top two are locked. See Plans and limits.

Download a PDF

On Growth and Enterprise, select Download Report in the header or the PDF icon in Execution History. PDF report is being generated means try again in a minute; No PDF report for this run means there isn't one.

Compare runs fairly

Before you put two runs side by side in a deck:

Same questions. If you edit the question set on Growth, runs before and after the edit aren't comparable.
Same engines. Partial and full runs, or Free and Growth, score different engines.
Several runs, not one. Answers vary. Set the Trend window to Last 10 or more before you call a trend.
New audience, new monitor. Measure a different product, persona or region with another monitor.