LONG RDDT: Narrative 180 -- What If AI Mode Actually Helps Reddit?

RDDT +6.8% today as they called out better demand trends in a Linkedin Post:

The data from Ahrefs, which tracks clicks a website is estimated to receive from unpaid (organic) results on Google’s SERP, points to a big increase in June traffic, while every other website is declining. This data jives with SensorTower data that is showing an uptick in June for the first time since late February - the data has steadily been declining since late February from 10%+ y/y growth to close to 0% y/y at the end of May and has now bounced back to 2.5% y/y in early June in just two weeks on tougher comps. (The comps get easier as we head through earnings in late July, so its likely we continue to see an improvement in the data).

What is going on here? Could it be another Google algo change helping RDDT? Or is something bigger and more permanent going on, something big enough to flip the narrative on RDDT completely and cause a DEFCON 1 for shorts: namely that AI Overviews/AI Mode is actually helping RDDT. Let’s dive in:

Google’s new AI Overviews routinely cites Reddit—Semrush ranks it the second-most-referenced site behind Wikipedia—so even when an overview answers the question in-line, the outbound link most likely to win a click is often a Reddit thread. That prominent placement creates incremental brand-impression loops that other sites, now shown lower on the page or folded into the overview itself, simply don’t capture.

The improvement of the data coincides perfectly with AI mode/AI overview expansion in late May.

Said differently, moving AI Mode/Overviews from an opt-in experiment to a quasi-default experience in late May created an immediate step-function in how often Reddit gets surfaced. Put even more simply: the most likely catalyst for the early-June inflection is Google widening the funnel of queries that show an AI Overview—and those Overviews cite Reddit far more than a classic blue-link SERP ever did.

This is a complete 180 from the narrative last month when we shorted post-earnings.

Remember this:

I’ll miss you dearly…

This is just me piecing together the pieces of the mosaic: the timing + improving data and coming to what is the most likely conclusion.

I went down a quick Q&A hole with OpenAI’s o3 to confirm. It’s worth a read in full. You can find it here. o3 came to the conclusion:

If a query does get an AI Overview, Google is ~10× more likely to pull Reddit into the answer than it ever was to put a Reddit URL into the old ten-blue-links stack…Across the whole search universe you’re now ~2–3 times more likely to be shown Reddit in an AI context (≈ 0.7 %–6 .5 % of all searches, depending on whether you count implicit use) than you were to click a Reddit blue link before AI Overviews expanded (≈ 0.3 %)

Short interest hasn’t really declined at all (MS Prime gives it a 5 out of 5 short crowding score) and we think that better sell-side checks heading into earnings + better US DAU growth + this new narrative which will be extremely hard to disprove on improving data will put bears on the back foot and cause not only weaker hands to cover, but even believers in the structural short begin to question why they are involved. How do I know this? Because I was one of those believers! (How many times did I post that annoying meme?)

The company will be at Cannes tomorrow at 2pm France time, and will likely sound good.

The set up is also completely different from a month ago when we shorted post-earnings. The overall ad market is now improving vs. decelerating. Expectations are lower: some on the sell side like Wells Fargo have even 0’d logged out revenue by ~‘27. At the same time, ramping LLM growth, RDDT’s Anthropic suit and META’s purchase of Scale AI have supported the idea that RDDT’s data is significantly more valuable than the ~$60M deals they signed last year with OAI and Google and more than enough to outweigh the only 15% of revs that currently come from logged out users

Remember ETSY rallying 30% in just a couple of weeks? Well, the set up here is better: highly shorted stock + improving data (and easier comps in near-term). Except with RDDT we have the potential for a very powerful narrative shift.

The main downside: we don’t have as a good a meme for our long as we did with our short last month.

Feedback welcome as always…