This is the exact report we email you, as a PDF. Every section carries a short margin note explaining what it is and what to do with it. The company and competitors here are fictional, but the numbers are modeled on real assessments so the report reads true to life.
What it is. Three independent AI models scored the same channel without seeing each other's numbers. The big word is your tier. "Invisible to AI" means most videos ship almost no readable text, so an engine cannot tell what they are about.
What to do. Read the tier first, then check the three scores land close together. When they agree, as they do here in the high 20s to mid 30s, the number is trustworthy.
The same video metadata was scored blind by two other AI models, ChatGPT and Gemini, with no knowledge of the audit's numbers. Running three independent models helps rule out the bias or hallucination of any single one. The table breaks your catalog into content tiers and shows each model's readiness score, 0 to 100.
| Content tier | Claude | ChatGPT | Gemini |
|---|---|---|---|
| Keynotes & product films 210 videos · 8 min +, usually best prepared | 42 | 46 | 44 |
| Explainers & how-tos 430 videos · real questions, thin descriptions | 34 | 40 | 36 |
| Brand, event & promo clips 600 videos · one-line or empty, no transcript | 15 | 20 | 16 |
| Channel-wide (video text layer) | 26 | 31 | 28 |
Readiness measures whether the text layer can be read and cited, not whether a video wins a given answer.
What it is. Your catalog broken into the kinds of video you actually make, each scored on its own, so the weakness is not hidden in one channel average.
What to do. Aim effort at the tier with the most videos and the lowest score. Here that is the 600 promo clips at 15 to 20, so fixing that tier moves the channel number more than polishing the keynotes already ahead.
Here is how your channel stacks up against your competitors on the video text layer, the part AI actually reads. Every score below comes from YouTube metadata alone, so it is a true apples to apples comparison.
| # | Channel | Video text layer | Question titles | With chapters |
|---|---|---|---|---|
| 1 | Wayne Enterprises @wayneenterprises · 820 videos | 48 | 44% | 39% |
| 2 | Oscorp @oscorp · 610 videos | 41 | 37% | 30% |
| 3 | Stark Industries you @starkindustries · 1,240 videos | 26 | 19% | 12% |
| 4 | Hammer Industries @hammerindustries · 340 videos | 19 | 14% | 8% |
Your channel is scored across its full catalog; competitor channels are scored from a representative sample of recent uploads. Question titles and chapters are shown as the share of scored videos.
What it is. Your three competitors scored on the exact same YouTube-only signal you were, so the ranking is fair. Video text layer is the 60-point lever, so this is the comparison that matters most.
What to do. Note the gap to the leader, not just your own number. Here Wayne and Oscorp both clear 40 while you sit at 26, so closing that gap is a competitive move, not just housekeeping.
The 0 to 100 is built from three weighted layers. The video text layer on YouTube (title, description, captions, tags, chapters) is 60 of those points, the biggest lever by far.
| Layer | Score | Read |
|---|---|---|
| Channel signals 15% weight | 72 | The brand channel is set up well: handle, links, and custom thumbnails across the catalog. This layer is not the problem. |
| Video text layer 60% weight | 26 | Weak. Most videos have thin or empty descriptions, no chapters, and no usable transcript. This is where the score is lost. |
| On-site + off-platform 25% weight | 30 | A few product pages carry schema, but most declare no video and no transcript is published as text anywhere. |
Catalog counts: 620 question-style titles, avg 46 words per description, 190 with chapters, 410 with no tags, 180 linking back to your site.
What it is. The three ingredients behind the single number, each weighted by how much AI engines rely on it. A strong channel layer cannot rescue a weak video layer, because the video layer is worth four times as much.
What to do. Spend where the weight is. The channel layer is already 72, so there is nothing to win there. The 26 on the video layer, worth 60 points, is the whole opportunity.
Ranked by scoring impact. Most need no new production, just text added to video you already have.
What it is. Your to-do list, ordered by how much each item moves the score, not by how easy it is. The counts in parentheses are how many videos the fix touches.
What to do. Work top down. The first three cover the transcript and description gaps that all three models flagged, and they touch the most videos, so they move the number fastest.
Sequenced so the fastest score movement comes first. Re-run this assessment after each wave to show the climb.
All three waves are work ISSIMO does for you. This assessment is the plan; the retrofit is the service. We start with Wave 1 on your highest-potential videos, publish the changes to your channel, and re-run this exact score so you can see what moved.
Scoring is directional. It measures how prepared existing video is for AI citation (GEO), not video quality. The three model scores measure readiness, not whether a video wins a given citation.
What it is. The order of operations, not just a wish list. Wave 1 is the cheapest, highest-impact work; Waves 2 and 3 compound it.
What to do. Run Wave 1, re-score, and let the jump decide the budget for Waves 2 and 3. You do not have to commit to all three up front.