THE SETUP

We went looking for buyer signals. We found a landfill.

Our reviews are built by counting what real buyers say in public — so last week we pulled every YouTube video from the past year that reviews three AI-search-visibility tools, and read the top comments on all of them. The plan was to harvest signals.

The harvest: out of 164 comments, fewer than 10 came from someone who had actually bought or seriously evaluated a tool. The rest was marketing wearing a costume. None of it is illegal — most of it is ordinary growth tactics — but if you can’t tell the costume from the customer, every “honest review” you watch is quietly selling you something.

Here are the four patterns, in the order you’re likely to meet them.

PATTERN 01

The alternatives matrix

One channel alone produced 10 of our 48 search results. Its catalog is an assembly line: “Top 7 X Alternatives,” “5 Best Y Alternatives,” “6 Better Alternatives to Z” — a video for every tool name people search. And under every one of them, the same pinned comment: a referral link to the same competing tool, complete with ?ref= tracking code.

That’s the tell: the “alternatives” format exists to intercept your search for tool A and route you to tool B, which pays commission. The comparison was never the product — you were.

It gets better: under one “I Tested Them All” video, a viewer asked the question we all should: “feels a bit AI generated — have you really tested it or only Claude ;-)”. No answer from the channel.

PATTERN 02

The outreach program

Several glowing walkthrough videos of one tool share a curious detail: the “try it free” links under them carry the exact same tracking tag — utm_campaign=awareness_outreachprogram_2026_global. The vendor’s own team shows up in the comments to say thanks for the feature.

To be fair: influencer outreach is standard marketing, and at least the utm string is honest about it. But it means these videos are distribution, not evaluation. The reviewer’s incentive is the next collaboration, not your purchase decision. Treat them as product tours — useful for seeing the UI, worthless as a verdict.

PATTERN 03

The commenter who always recommends the same tool

Across unrelated videos, we kept meeting the same comment, structurally: “Nice list! I’ve been using [tool you’ve never heard of] lately and it’s been a game changer — anyone else tried it?” Different accounts, same tool, same faux-casual question mark at the end.

The smoking gun: under one comparison video, two different accounts posted the same recommendation, word for word, down to the emoji. Copy-paste astroturf, caught in the act.

A subtler variant name-drops a proprietary-sounding “index” or “framework” in academic tones across many videos — same phrase, many accounts. If a comment reads like a whitepaper and mentions a brand you can’t find a website for, it’s a plant.

PATTERN 04

The applause machine

Two interview videos in our set each had 80+ comments — jackpot, we thought. Then we read them: “Very informative video.” “Nice sharing.” “Thank you for sharing this wonderful vedio.” Dozens of near-identical one-liners, posted in tight clusters, praising everything and saying nothing. The interview subject? Founder of the tool being praised.

Engagement pods inflate the numbers that make a video look trustworthy before you’ve watched a second of it. High comment count is not social proof — specific comments are.

THE DEFENSE

Five checks, ninety seconds

Before you trust any tool review on YouTube:

01
Read the pinned comment first A ?ref=, discount code, or utm-tagged signup link tells you the business model before the video tells you anything else.
02
Open the channel’s video list Ten “alternatives” listicles in two months is not a reviewer, it’s a funnel. One deep dive per month usually is.
03
Copy a suspicious comment into search Word-for-word duplicates across accounts or videos = astroturf. Takes ten seconds, works disturbingly often.
04
Check who’s being thanked Vendor employees saying “thanks for featuring us!” in the comments means outreach happened — read the video as a product tour, not a verdict.
05
Hunt for the dissenting comment In our whole dataset, the single most useful comment was a methodology critique buried under a promo video. The one voice arguing with the room is usually the one that used the product.
THE SIGNAL IN THE NOISE

The nine real voices were worth the dig

Buried under the costume party, the genuine buyers were saying consistent things: praise for what these tools show, frustration at what they cost, and one sharp skeptic questioning whether prompt-sampling can represent real user behavior at all. Those nine comments — verified, source-linked — are going into our upcoming tool reviews, counted the way our methodology counts everything.

That’s the whole thesis of this site, demonstrated in one dataset: the signal exists, but nobody should have to read 164 comments to find 9. We’ll keep doing the reading.

DATASET & DISCLOSURE · 48 videos surfaced by YouTube search for three GEO-tool names (last 12 months), top-25 relevance-ranked comments per video via the official YouTube Data API, collected 2026-07-05 and read in full by a human + AI triage. We review tools in this category, which is why we were digging. This article contains zero affiliate links; outbound video links are nofollow. Spotted an error? Email us — corrections are made in public. · HOW WE COUNT →