Most marketing teams are convinced they know where they lose deals.
The pricing page.
The pitch was off because of XYZ.
The competitor won because they have slicker demo.
Or a prospect who goes quiet after the second call because of course…
They just used Claude to build it.
Cope.
This is all just cope.
Because for some of these brands… When they open up Google, type their brand name, and add one word: “reviews.” Here’s what a lot of you are about to see:
But let’s pretend it’s not there.
Let’s pretend we don’t have a sentiment problem that is causing people to churn before they even reach out. Don’t get me wrong…
Most companies aren’t facing this level of Reddit Reputation problems.
Most are dealing with reputation problems like this:
These reviews are still not great.
But they’re still influencing buyers asking about your product.
The source this time isn’t Reddit though. The source is G2. A review site for SaaS companies where customers share the pros and cons around different software. ChatGPT (like many LLMs) use G2 as a trusted source to incorporate real customers feedback in the replies for user queries.
Some of these reviews are old.
Just like that Reddit thread…
Five years old and still ranking. Underneath it, Google’s People Also Ask box is asking the questions for you:
Why is [Brand] falling? Is it difficult to cancel? Is there a lawsuit against [Brand]? What are the disadvantages? Then the discussions pile on — “Does anybody actually like [Brand] products?” “Why is [Brand] the worst and most frustrating software on the market?”
Sit with that for a second.
This type of experience will do more to your pipeline than your entire content calendar.
The New Buyers Journey Starts Before Your Site
Buyers are being influenced the moment they Google you or ask an LLM about you.
It’s happening in the SERP.
It’s happening in the LLMs.
Half the Page Stopped Being Yours
I didn’t want this to be a vibes argument. So we pulled the data.
Across 309 SERP results, here’s how the real estate actually breaks down:
- Organic results: 49%
- AI Overview: 23%
- Video carousel: 14%
- Discussions & forums: 8%
- People Also Ask: 5%
- Local pack: 1%
Read that top number again. Organic is 49%. The thing every SEO team has optimized for over the last twenty years now controls less than half the page. The other 51% belongs to features you don’t write and can’t rank in the old-fashioned way.
And the biggest new tenant on the block? AI Overview at 23%. Almost a quarter of the page is an answer the machine wrote before your buyer scrolled to a single blue link.
So the obvious question is: where does that answer come from?
We looked at AI Overview citations across 19 brands and counted how many cited each source:
- Reddit: 14 of 19 brands
- G2: 8
- Trustpilot: 5
- YouTube: 5
- Gartner: 4
- Capterra: 4
- PCMag: 3
- SoftwareAdvice: 3
Reddit shows up in the AI answer for 14 of 19 brands. That’s 74%. It beats G2 nearly two to one. It beats Gartner, the source your category has trusted for thirty years, more than three to one.
Now put the two charts together. The fastest-growing block on the page is the AI Overview. The AI Overview is built on Reddit three times out of four. And Reddit is the same place where “[Brand] is a horrible company” has been sitting and ranking for five years.
If the AI answer is built on Reddit, your AI visibility strategy starts on Reddit. There’s no way around it.
You Can Flip the SERP: Case Study
Search results change.
You can outrank a bad thread eventually. Here’s the playbook our team ran recently with a $1B brand:
- Identify the negative sentiment in the SERP on Reddit
- Create a category subreddit & brand subreddit
- Create a persona account & brand account
- Create content in the Subreddits
- Build karma by adding value in relevant threads
- Create content targeting branded queries
- Reply to real customers in these threads with value
- Rinse & repeat until the SERP flipped
That’s it.
Build fresh content.
Earn comments & upvotes.
Change the narrative. We’ve done it time and time again.
BUT…
Training data doesn’t work like that.
When that Reddit thread gets pulled into the training data for the models people now use to research vendors, it stops being a ranking situation. It becomes the model’s memory of you.
Ask ChatGPT, Claude, Gemini, or Perplexity about your brand and they answer from what they absorbed. If what they absorbed is “horrible company” and “worst software on the market,” that’s the summary your buyer gets handed back in a calm, confident, authoritative voice.
You can’t push that down to page two.
There is no page two. There’s one answer, and the model already wrote it.
Don’t get me wrong…
You can dilute it. But that’s going to require intention & patience.
Rank Tracking & Citation Tracking Will Never Be Perfect
Most teams monitor Google rankings and call it a day. The SERP is changing every day. The response you see in AI Overviews today could be completely different in 30 minutes. The sites that are cited in one city are different from the sites cited in another. The memory that modifies the queries for one user are different than what gets modified for another. This is why in my honest opinion… Most marketers need to stop looking at citation data as the gospel and instead look at it as a directional guide.
Fix The Problems Creating The Training Data Issues
This might not sound like rocket science… But can you fix the thing the thread is about? If ninety people say canceling is a nightmare, that’s a product problem wearing a marketing costume. The smartest brands read negative content as the cheapest customer research they will ever get, and they go fix the root cause. I have always loved Reddit & Tripadvisor because it gives you real feedback. This is your chance to improve. Take it seriously and if your team doesn’t… There’s a lot of great companies hiring.
Let me wrap it up with this…
Almost nobody is doing this.
Most of your competitors are still tweaking meta descriptions while their brand gets defined by a five-year-old comment section and a model that read it. The brands that move now get to write the story the machines tell for years.. I’m telling you this is important.
Get ahead of it. Own the narrative before the model owns it for you.
The window’s open. Go.