If your dealership runs paid social, the rules just quietly changed under your feet. Meta rolled out a major AI upgrade to how Instagram ads get ranked and delivered, and it’s already moving the numbers for advertisers who know how to feed it the right creative.
- Meta’s new ranking AI uses trillion-parameter, LLM-scale intelligence to match ads to user intent in real time.
- Early results show a 3% lift in conversions and a 5% lift in click-through rate for targeted Instagram users.
- For dealers, creative variety and broader Advantage+ targeting now matter more than tight manual audience controls.
What the Adaptive Ranking Model Actually Does
Meta just published details on what it calls the Adaptive Ranking Model, a new system that runs the final ad decision on Instagram (and, eventually, Facebook). It replaces the old approach with intelligent request routing, aligning model complexity with a richer read on user context and intent. In plain English, Meta stopped treating every ad auction the same way.
Before, Meta ran the same mid-range model for every single ad request, regardless of whether the user was a high-intent buyer or a casual scroller. Now the system decides how much brainpower a given impression needs. A shopper who just searched for a midsize SUV might trigger deeper processing than someone passively scrolling reels at lunch.
The tech behind it is wild. Meta is operating at what it calls “LLM-scale complexity,” roughly 10 billion floating-point operations per token, with trillion-parameter embeddings, while still holding sub-second response times. That’s comparable to the complexity of models like ChatGPT, but running fast enough to serve an ad before you finish swiping.
The Numbers Meta Is Reporting
Since launching on Instagram in Q4 2025, the model has driven a 3% increase in ad conversions and a 5% increase in click-through rate for targeted users. Those are platform-wide averages, not guaranteed lifts for any one advertiser, but they point in the right direction. The updated Adaptive Ranking Model will use less computing power to serve more relevant ads and improve return on ad spend.
For an auto dealer running Advantage+ Shopping or lead campaigns, that translates to cheaper test drives booked, cheaper form fills, and a higher share of impressions landing on people who actually care about the vehicle you’re promoting.
Why This Matters for Dealership Ad Strategy
There’s a catch. A smarter ranker only pays off if you give it useful material to rank. The model now processes each user’s full behavioral sequences across content and ads, not just recent clicks, but a layered history of how they engage with everything on the platform. To get value out of that level of signal, it needs genuinely diverse creative to work with. Slight variations of the same ad give the model low signal.
For dealers, that means running five lightly tweaked photos of the same inventory shot won’t cut it. A product video, an educational carousel, a UGC testimonial, an offer-led static, those represent different awareness stages, different emotional hooks, different reasons to buy. That’s high signal. That’s what the model can actually learn from and reward.
A few practical shifts worth making now:
- Mix creative formats. Walkaround reels, salesperson UGC, inventory carousels, payment-offer statics, and test drive invites all teach the system different things about your buyers.
- Lean into broad targeting. This model powers the final decisions inside your ASC and Advantage+ campaigns. If you’re consolidated on broader targeting and letting Meta’s automation do the heavy lifting, a stronger ranking model means better decisions being made on your behalf. But only if you’re giving it diverse inputs to work with.
- Protect the learning phase. Constant creative swaps and weekly campaign restructures starve the ranker of the data it needs to calibrate.
How the Adaptive Ranking Model Fits With Andromeda
If you remember the 2024 Andromeda update, this is the companion piece. Andromeda was a retrieval layer change. It changed how ads are matched to users at the top of the funnel, essentially widening the pool of candidate ads that get considered. The Adaptive Ranking Model is a ranking layer change. It changes how those candidate ads are scored and ordered after retrieval. Andromeda decided which ads enter the room. The Adaptive Ranking Model decides who walks out with the keys.
Together, they move Meta further toward the AI-run, creative-first ad engine the company has been telegraphing for two years. Google’s Performance Max and TikTok’s Smart+ both use large AI models to shape targeting and delivery at scale, so this is the direction every major platform is heading.
Smart Plays for Dealers Right Now
Stop fighting the algorithm. Feed it. Audit your current Instagram campaigns this month, count how many genuinely different creative concepts are running, and fill the gaps. Consolidate ad sets where you can, broaden your audiences, and let the ranking layer do what it’s now built to do. The dealers who win the next year of paid social won’t be the ones with the cleverest audience stacks. They’ll be the ones whose creative libraries look like a real marketing department, not a copy of a copy of last month’s best seller.
This post may contain affiliate links. Meaning a commission is given should you decide to make a purchase through these links, at no cost to you. All products shown are researched and tested to give an accurate review for you.