The world of tracking has changed. So has the playbook.
Not long ago, Meta advertising worked like this: pick your audience, define their interests, layer on some demographics, and serve them your ad. You could get quite surgical. Had someone visited a competitor’s website? Target them. Were they a fan of a particular public figure? Target them. Household income, interests, browsing behaviour — all of it was fair game.
That era is over.
Privacy legislation, browser updates, and cookie restrictions have steadily stripped away Meta’s ability to track behaviour across the web in the way it once could. The data that used to flow freely back to Meta — who visited what website, what they read, what they did next — is now largely gone.
This isn’t a minor tweak. It fundamentally changes how Meta advertising works.
Meta’s answer: the creative is the targeting
Meta has moved to what they call creative-driven targeting. Their Andromeda algorithm — their AI — now does the heavy lifting of finding the right people. But here’s the catch: it needs something to work from.
If your creative (the ad image, video, copy, or audio) clearly speaks to a specific type of person in a specific situation, Meta’s AI can use that signal to find more people like them. It builds a behavioural model of each user based on what they scroll past, what they stop on, what they click, what they expand, what they pause. It does this fast — we’re talking minutes from the moment someone opens the app for the first time.
That said, tracking signals haven’t disappeared entirely — they’ve just become far patchier. Where browsers and users permit it, Meta still picks up behavioural data. Where the Meta Pixel and Conversions API are properly set up, Meta still knows who’s purchasing, who’s browsing, and how people are moving through a buying journey. When that data is available, Meta folds it in. It models across what it can see to build a more complete picture of audience behaviour — filling in the gaps where signals are missing.
The reality is it’s a blended picture. Some signals are rich, some are thin. The algorithm is doing its best with whatever it has. Which is exactly why the creative has become so important — it’s the one signal we can consistently control and sharpen.
You can still technically set broad interest or demographic targeting inside Meta. But the data behind those filters has thinned out so much that it rarely performs well. What actually works now is putting the right creative in front of Meta’s algorithm and letting it go to work.
The concept we build everything around: Locksmith Moments
Here’s where I take a slightly different approach from most agencies.
The goal isn’t just to make a good-looking ad. The goal is to speak directly to a moment — a specific situation in someone’s life, work, or day — when they are genuinely open to what you’re offering. Not idly scrolling past it. Actually susceptible to it.
These are called Locksmith Moments.
Think about how a locksmith gets found. You don’t plan to call a locksmith. You don’t browse their website at leisure. You lock yourself out of the house at 11 pm in the rain, and suddenly you need one right now. That specific, pressured moment is when the locksmith goes from irrelevant to essential.
Your product or service has its own version of these moments. A business owner who’s just lost their third member of staff in a month. A founder staring at a spreadsheet at midnight wondering why the numbers aren’t moving. A parent who’s just booked a holiday and realised they haven’t sorted X, Y, or Z.
These are the moments when people are genuinely open to a solution. So that’s where we focus.
How we turn Locksmith Moments into creative
We take the strategy work — what we know about your ideal customer, their world, their frustrations, their triggers — and we use it to generate a wide range of these moments. Not just one or two. We go deep. Different types of people. Different situations. Different emotional states.
For each moment, we build creative around it: copy that speaks to that exact situation, visuals or video that the person can place themselves inside, language that makes them think this ad is for me.
Because here’s what Meta is doing on the other side: it’s watching who stops scrolling on each piece of creative. Who clicks. Who watches to the end. And it’s quietly building a picture of what kind of person responds to what kind of content. The more varied the creative we seed, the more patterns Meta can find — and the more precisely it can start routing the right ad to the right person.
Alongside the creative signals, we also feed Meta your own first-party data wherever possible. This means setting things up so that Meta knows when someone visits your website, how far they scroll, which buttons they click, whether they submit a form, add something to a basket, or complete a purchase. Every one of those actions is a signal. The clearer the picture we can give Meta of what a high-value visitor looks like on your site, the better it gets at finding more of them — and at working out which creative is attracting the right kind of attention, not just any attention.
How we structure Meta Ads campaigns
In the old days, you might have built dozens of ad sets and campaigns, each tightly targeted at a specific slice of the audience. That complexity is now largely unnecessary. Meta’s algorithm doesn’t need you to carve up the audience by hand — it will do that itself if you give it the right variety of creative.
Here’s how we build it:
- One campaign per audience segment. A segment is defined by your strategy — it’s a meaningful grouping of people with a shared context, not just a demographic slice.
- Broad targeting only. We might apply regional or age parameters, but we don’t layer in behavioural or interest filters. We keep it wide and let the creative do the narrowing.
- One ad set. All the creative goes into a single ad set within that campaign.
- Many creatives. This is where the volume comes in. We load in a range of ads — different Locksmith Moments, different angles, different formats.
Meta will then run what is effectively a live test across all of those creatives simultaneously. It will start allocating more impressions to the ones that are working and pull back from the ones that aren’t. There’s no winning ad anymore in the traditional sense. Meta will find that different people respond to different things, and it will use that to dynamically match creative to audience.
If a particular creative shows consistently poor engagement over time, we’ll switch it off manually. But in most cases, the campaign itself is the test. There’s no need to run separate A/B experiments on top of it — that’s already happening in real time.
What this means for the work you’ll see from Unlock Growth
When I send you a document with Locksmith Moments and creative concepts, what you’re looking at is essentially a map of the different situations we’re going to speak to — and the different ways we’re going to speak to them.
From there, we build the actual creative assets, load them into the campaign structure above, and let Meta start learning.
The approach is iterative by design. We’re not placing a single bet on one winning ad. We’re giving Meta a range of angles and watching what it discovers.