Community signals are AI’s largest third-party source

Throughout a pattern of 35,000 ChatGPT citations, user-generated content material (UGC) platforms maintain extra of the cited-domain share for SaaS-related prompts than assessment websites and publishers mixed. They maintain it on the high of the journey, on the backside, and at each level between. There isn’t any stage the place they’re skipped.

So how do you construct authority in a supply you don’t management?

Let’s speak model authority multiplication. Half 1 lined third-party citation signals. Components 2 and three lined proprietary data and what makes it citable. Each of these are levers you’ll be able to pull by yourself.

This one — UGC — will not be.

The uncomfortable half: The UGC sample is greater than anticipated. Drawback is, that is the authority layer you’ll be able to seed, immediate, and take part in, however by no means really personal. Whereas some platforms, like Reddit, are actually, actually difficult to affect.

Methodology and limitations

The info comes from an evaluation I ran for G2 in February 2026.

• Supply: ~35,000 quotation URLs captured in Profound.
• Protection: US solely, ChatGPT solely, month of December 2025.
• Every SaaS vendor-related immediate was categorized into one journey step: discovery (376 prompts), exploration (1,179), analysis (1,367), or centered analysis (255).
• Every quotation URL was lowered to a root area and categorized as review_platform, ugc_platform, writer, or vendor_or_other.
• Information had been de-duplicated to 1 per run, intent, and area, so a single reply citing 6 Reddit threads counts as soon as.

The analysis stage was the biggest journey step lined on this immediate set. 

And for reference, greater than half of the SaaS-related prompts on this research use business language. However just one.5% truly identify a vendor model. 

What this evaluation doesn’t do:

• The UGC set is large. It teams Reddit, Wikipedia, Quora, YouTube, and LinkedIn. These UGC platforms have totally different makes use of and audiences. We report on the UGC set in combination for the stage evaluation and get away composition later (in part 3).
• The evaluation measures share of distinctive cited domains, not share of quotation quantity. Part 1 measured quotation rows and put Video and Social at ~6.5%. Totally different denominator, totally different bucket, totally different dataset. 
• One engine, one market, one month. There are limitations with this information set. Given the consensus gap, the place 91% of citations seem in just one engine, don’t learn ChatGPT outcomes as general AI search outcomes.
• The prompts for this evaluation had been software program vendor-seeking by design, which is why vendor domains maintain 66.7% to 71.8% of citations at each stage. 

Strip out the seller domains included on this evaluation, and have a look at what’s left: UGC platforms maintain 17.1% of cited domains general, greater than 4x publishers at 4.0%.

Sit with the ratio for a second, as a result of most authority-building budgets are aimed on the 2 smaller buckets: Digital PR targets publishers. Evaluate campaigns goal assessment platforms.

However the largest outside-source class within the pattern, for this explicit set of prompts, receives the least deliberate funding. Largely as a result of most advertising execs aren’t positive what a plan for it even appears to be like like.

Much more importantly, excluding the seller, UGC is the most typical third-party supply to indicate up within the AI Reply (not simply the quotation set) on this SaaS immediate information. 

2. Evaluate websites transfer with buy intent. UGC doesn’t.

Learn the desk by column as a substitute of by row and a second sample exhibits up.

UGC dips barely within the Analysis journey step, passing its share over to publishers and assessment platforms. (Confirming that, sure, funding in Digital PR and assessment campaigns does matter.)

However assessment platforms swing. They sit at 7.4% in discovery and climb to 13.2% at analysis, roughly 1.8x, then fall again to eight.4% in centered analysis. A 5.8-point vary.

UGC barely strikes. 17.8%, 18.2%, 15.1%, 17.2%. A 3.1-point vary throughout the phases.

Evaluate platforms are a dependable bottom-of-funnel lever, whereas UGC is a ground.

That distinction may help resolve the place the cash goes and when. A assessment marketing campaign will be scheduled towards 1 / 4 as a result of its payoff concentrates at a identified stage. In different analyses, I discovered that (software program) evaluations increase a vendor’s AI visibility and shape AI solutions.

However neighborhood presence has no such stage, sadly. Very like traditional natural search visibility, it entails doing work to achieve visibility within the reply when somebody is studying what their wants are, and it nonetheless requires work to earn visibility when they’re selecting between 2 finalists.

The assessment platform shares peaks at analysis and falls to eight.4% in centered analysis. Evaluate platforms do their work whereas a purchaser is constructing a shortlist, much less as soon as they’re evaluating 2 finalists. UGC holds at 17.2% via that very same step.

Even on the narrowest hole between UGC and the assessment platforms (on the analysis step), the purpose the place peer proof is supposedly most targeting assessment websites, UGC nonetheless leads in quotation quantity.

3. Platform focus isn’t platform stability

A ground that holds at 17% doesn’t imply the platforms beneath it maintain.

Break the UGC bucket into its components: Wikipedia, Reddit, and LinkedIn account for 99% of UGC citations on this pattern. Every little thing else splits the remaining 1%.

Wikipedia alone runs 10.1 to 14.0 factors of that 17-point ground relying on the stage. It’s the only largest third-party supply within the dataset… bigger than all the assessment platform class at each journey step besides analysis. Meaning the largest slice of the largest third-party supply class is the one the place deliberate motion is least accessible to you.

Regardless of their maintain throughout the AI citations and solutions on this set, these platforms are unstable of their visibility. 

In GIB #20, Reddit declined on each dimensions for the primary time since I began monitoring AI mentions in February: visibility down 11.7% and AI mentions down 10.9% within the 28 days to June 8, 2026.

Three weeks later the bucket moved the opposite manner. Within the 28 days to June 29 (GIB #21):

• LinkedIn +43.3% search engine optimization visibility (289.4 to 414.7)
• X +38.6%
• Instagram +21.4%
• Fb +20.1%
• Reddit +18.1% (2,297.6 to 2,765.7)
• YouTube on the backside of the vary: 11%

All UGC. The two home windows overlap by every week, so deal with these as adjoining reads on a transferring bucket relatively than a earlier than and after.

Each a kind of platforms misplaced AI Overview citations in the identical 28 days it gained natural visibility. Google rewarded them in search and cited them much less in AI solutions, on the identical properties, in the identical window. A platform can focus share in a single floor whereas dropping it within the different.

So deal with UGC as a portfolio and provides every platform a job.

The mixture is the quantity to defend. The composition is what you rebalance. And the platforms doing the work right here aren’t interchangeable:

• Wikipedia is the biggest and least actionable. You don’t marketing campaign right here. You make sure that the sources Wikipedia editors are required to depend on (issues like commerce press, authentic analysis, major documentation) exist and are correct. 
• Reddit is probably the most unstable, however presence right here is price it.
• LinkedIn is the place a named individual outperforms a model account. 
• YouTube presents sturdiness as a Google-owned property the place you will have some management, and it’s your hedge.

Deal with the platforms the place your viewers truly lives. An viewers analysis device like Sparktoro will let you know the place. What you don’t do is wager the entire neighborhood effort on one platform after which test the field that you simply’re “doing UGC.”

None of that is an argument for getting evaluations or astroturfing threads. That’s a short-term tactic that may harm your model within the long-term. However there’s an argument right here for exhibiting up the place the solutions are already being assembled.

Discover which UGC platforms your class feeds

Run your highest-intent prompts and file which UGC domains seem, principally throughout citations. In some classes, it will be Reddit and YouTube. In others, it’s a Discord, a Stack Trade, or a single commerce discussion board. The bucket is 17% on common and the composition is native to your subject. Half 1’s discovering applies right here: The supply set AI trusts is rebuilt per subject.

Spotlight folks within the locations that matter to your viewers, not simply the model account

In Topics matter for third-party signals, we reported {that a} named writer with a byline seems to outperform the identical content material underneath a model, and LinkedIn’s personal testing pointed the identical manner. A neighborhood treats an individual as a participant… and a emblem as an advertiser. This may be one thing so simple as your greatest gross sales rep making a YouTube shorts collection about suggestions and tips together with your product.

Reply the questions your assist queue already sees on repeat

Ticket logs, gross sales name transcripts, and in-app search queries maintain the precise phrasing patrons use. Why proprietary data is your most defensible asset made this case for owned content material. It applies even more durable in neighborhood, the place a query is sitting there already requested.

Make your prospects’ phrases retrievable by yourself pages

Pull actual third-party assessment quotes, with hyperlinks to the supply, into the highest of your highest-cited pages. The science of how AI pays attention discovered 44.2% of citations come from the primary 30% of a web page. 

Whether or not the presence of optimistic third-party proof in that high third band modifications how AI describes your model is untested (outdoors of first-hand expertise), so run it as a check with a said speculation and a management set relatively than as a tactic.

Measure the combination, rebalance the combination

Monitor UGC-sourced citations as 1 quantity, and platform composition as a second. The primary tells you whether or not the ground is holding, and the second tells you the place to maneuver subsequent quarter.

Mine communities for questions and terminology

Extract questions your viewers asks on Reddit & Co. and monitor what phrases they use to explain your class, options, and types. Then, use this data to tell your prompts and replica.

Right the file that’s already sitting there

A 2023 thread claiming you lack a characteristic you shipped final 12 months stays retrievable and retains feeding solutions. Reply in-thread with a dated correction and a hyperlink to the change. Removing requests not often land, and deleting a thread deletes the context AI is studying from. 

Set alerts on model plus class phrases (F5Bot for Reddit and HN, Syften should you want Discord and Slack protection), so that you catch it in weeks as a substitute of quarters.

Give prospects and energy customers one thing price posting on their very own: early entry to your information, a free device, a quantity they’ll quote in an argument. That’s as near possession as this bucket will get.

If a discovering from this week modifications a call in your model, the Growth Memo Premium Resource Library has 45+ (and rising) checklists, gentle instruments, AI workflows, and stakeholder slide decks to place it to work.

5. Google began measuring social, and omitted the platforms that matter most right here

On July 7, Google introduced platform properties in Search Console, a property kind that studies clicks, impressions, CTR, common place, and queries for Instagram, TikTok, X, and YouTube. You have to hyperlink owned accounts to view, so this isn’t the identical as model point out monitoring throughout the net.

However Reddit isn’t on the listing. Neither is LinkedIn, the 2 platforms doing probably the most work within the UGC bucket above are the 2 you continue to can’t measure natively. 

Plus, this new characteristic studies visibility in Google Search and Uncover, not AI solutions. A staff that connects these properties and calls the output AI visibility measurement can be measuring the improper floor with actual numbers, which is worse than measuring nothing.

Use it for what it’s: 

• The primary native learn on how your off-site content material performs in traditional Google. (The neighborhood indicators feeding AI solutions keep unmeasured for now.)
• An essential sign of how essential multichannel model visibility is for AI search.

Out of the components within the model multiplication collection (we examined subjects, proprietary information, and authentic first-party analysis), neighborhood proof is the one one which retains working when you sleep. 

Third-party authority takes 1 / 4 to maneuver. Proprietary information takes a analysis cycle. Neighborhood presence takes a 12 months, and as soon as it takes off organically, then holds a ground that doesn’t care which stage of the journey your purchaser is in.

Fund it as a standing line relatively than a single marketing campaign, as a result of there isn’t any stage the place turning it off is protected once you wish to preserve or develop natural AI search visibility.

This submit first appeared on the writer’s web site and is republished right here with permission.