Why interest targeting is no longer a reliable edge (and what replaced it)
The constraint has shifted
For the better part of a decade, paid social success was dictated by how clever you were at platform optimisation.
Audience targeting. Bid strategies. Campaign structures. Lookalike audiences. Interest stacking. The manual levers that separated good media buyers from great ones.
That era is ending.
The platforms have automated most of that optimisation. What they haven’t automated is what you feed them. Performance is no longer dictated by how clever you are at Meta optimisation. It’s dictated by creative supply.
The shift from targeting to creative supply
Meta’s algorithm can now evaluate tens of millions of ads per auction, matching individual creative to individual users in real time. That’s a fundamental shift from even a few years ago, when the system chose from thousands of ads and your targeting decisions were the main lever.
With that expanded capacity, the algorithm handles audience matching better than any human media buyer can. It considers signals you can’t access: device, time of day, recent behaviour, content preferences, thousands of other factors, all processed in fractions of a second.
The question is no longer “who should I target?” but “what should I give the algorithm to work with?”
Three Eras of Advertising
To understand why creative became the primary lever, consider the evolution:
Era 1: Broadcast (Pre-Internet)
One-to-many advertising. You made one ad, ran it on TV or in print, and hoped it resonated with enough of the mass audience to drive results.
The constraint was reach. Could you afford enough media to be seen?
Era 2: Internet Advertising
One-to-some. The platforms gave us audience targeting: behavioural data, demographics, interests, lookalikes. You could show different ads to different audience pools.
The constraint was targeting skill. Could you find the right audiences before your competitors?
Era 3: AI-Powered Advertising (Now)
One-to-one. The platforms can now serve personalised creative to individual users based on their predicted preferences.
The constraint is creative supply. The algorithm can only optimise from the options you provide.
Why targeting matters less
In the targeting era, media buyers added value through manual optimisation:
- Finding high-performing audience segments
- Adjusting bids based on performance data
- Structuring campaigns to control delivery
- Building custom audiences and lookalikes
These skills still have some value. But the platforms have automated most of this work, and the AI handles it better than humans at scale.
When you set up broad targeting and let the algorithm optimise, it processes signals you can’t access. It considers device type, time of day, recent behaviour, content preferences, and thousands of other factors in real-time.
You can’t out-target the algorithm. But you can out-supply it.
Creative variety functions as targeting
Two users who look identical demographically might respond to completely different creative. One prefers UGC-style content with authentic testimonials. The other prefers polished product shots with text overlays. One responds to problem-led messaging. The other responds to aspiration-led messaging.
If you only have one style of creative, you’re reaching one type of person effectively and losing everyone else.
If you have multiple creative styles at scale, the algorithm serves each user what they’re most likely to engage with. Your creative diversity becomes your targeting strategy.
This reframes how to think about creative volume. It’s not just about having more ads. It’s about having different ads that appeal to different people.
What this means for production
Old model (open loop)
- Brief creative team on campaign concept
- Produce 3-5 hero assets
- Test to find winner
- Scale winning creative
- Refresh quarterly when performance drops
This is an open loop. Performance data lives with the media team. Creative briefs come from brand. The two rarely connect in real time. By the time insights reach production, they’re weeks old. By the time new creative launches, the insights are stale.
New model (closed loop)
- Performance data flows directly into creative briefs
- Production structured by intent, informed by what’s working now
- Diverse assets across formats, styles, and messages
- Algorithm finds complementary combinations
- Results feed the next production cycle immediately
This is a closed loop. Performance insight informs production. Production generates new data. Data sharpens the next brief. Each cycle builds on the last.
The old model assumed you were looking for the one ad that works best. The new model recognises that different ads work best for different people, and you need enough variety to capture the full opportunity.
Volume trajectories
The benchmarks have shifted.
Four or five years ago: Brands could succeed with 5 hero creatives. Find what works, scale it, refresh quarterly.
As Josh describes from his time at Huel:
“We ran the same ad for about 18 months. We must have made millions from new customer acquisition revenue from that single ad. But that doesn’t really fly anymore.”
Today: Competitive brands are running 30-50 assets per month minimum. Leaders are running hundreds.
Tomorrow: Dramatically more. As Will Ashton, CEO at Nest, puts it: “You need to be doing 50, 100, 500.”
This reflects what the platforms can now do. The brands providing more options get more opportunities for optimal matching.
The variety requirement
Volume alone isn’t enough. Creative variety must be genuine.
This is NOT meaningful variation:
- Same image, different headline
- Same concept, different background colour
- Same hook, different thumbnail
This IS meaningful variation:
- UGC-style content vs polished brand content
- Product-focused vs lifestyle-focused
- Problem-led messaging vs aspiration-led messaging
- Video vs static vs carousel
- Hi-fi production vs lo-fi authenticity
The test: Would these ads appeal to different people, or are they all optimised for the same viewer?
If you’re creating 100 ads that all appeal to the same customer archetype, you’ve created 100 attempts to win the same person. If you’re creating 100 ads across genuinely different approaches, you’ve created pathways to win different people.
Testing in the new model
The testing philosophy shifts alongside everything else.
Old model: Find the winning ad. Kill the losers. Scale the winner.
New model: Find the complementary mix. Multiple “winners” serve different purposes for different segments.
At Nest, we’ve seen testing shift direction. It’s no longer about finding a single winning ad. It’s about finding a complementary mix that collectively captures more value across your addressable audience.
A UGC testimonial ad might have lower average performance than a polished product ad. But it might convert customers the product ad would never reach. Both belong in the mix.
Speed as competitive advantage
Creative-as-targeting only works if you can execute fast enough.
The algorithm aggressively serves what’s working, which means your best ads burn out faster than ever. Creative fatigue accelerates when the algorithm aggressively favours what’s working.
The old creative cycle (brief, produce, test, analyse, repeat every 1-2 months) is too slow for this environment. Competitive brands are moving to weekly cycles. Leaders are approaching continuous production.
As Josh puts it:
“The future of paid social is to close the loop between creative production, campaign management, and data analytics, and speed up the frequency. Instead of once a month, you want to be doing this on a weekly basis, giving you 52 optimisation points per year. If you’re getting 1-2% marginal gains per cycle, the effect compounds over time.”
If you can optimise 52 times a year versus 12, that compounds. The brands with faster creative loops pull ahead of those stuck in quarterly refresh cycles.
What this means for buyers
The role of media buying hasn’t disappeared. But it has changed.
Less time on:
- Manual audience building
- Bid adjustments
- Campaign structure optimisation
- Interest targeting
More time on:
- Creative strategy and briefing
- Performance analysis across creative variables
- Identifying what’s working and why
- Coordinating production pipelines
The best media buyers are now creative strategists who happen to operate platforms. They understand that the lever they’re actually pulling is creative supply, not platform settings.
The bottom line
Meta built an algorithm that can match ads to individuals at a scale humans can’t replicate. The response isn’t to fight that with manual targeting. It’s to supply the algorithm with enough creative variety that it can find your customers wherever they are.
The brands that understand this and build systems to execute it will outperform those still optimising for the targeting era.
The algorithm can only work with what you give it. What are you giving it?
How Hummingbird helps
Hummingbird is marketing infrastructure for the AI era, built to close the loop between what the algorithm tells you and what you feed it back.
For media buyers who’ve mastered the platforms: You know targeting skill matters less. You know creative volume is the lever. What you need is infrastructure that turns that understanding into execution at scale.
Closing the loop:
- Performance data reveals which creative approaches work for which audience segments
- Insights translate directly into production briefs structured by intent
- AI-assisted workflows produce genuine variation (not template swaps) at the speed the algorithm demands
- Every cycle feeds the next: analysis informs production, production generates data, data sharpens analysis
Human creative direction stays in control. AI handles scale.
When creative is the targeting, you need infrastructure that treats creative production as a core marketing function, not a support service. You need systems that can produce diverse creative as fast as algorithms can consume it.
That’s what Hummingbird provides: the operational layer that makes creative-as-targeting work. See how it works.
Ready to see what we can do for your brand?
Book a call with our team to get know more about it.
