In partnership with Ballpoint
Last week we covered knowing when you’re ready to scale and what to expect when you do.
If you missed it, go back and read Part 1 first. Don’t skip ahead.
This week we’re getting into the operational stuff, where I made most of my mistakes at Yonder, so hopefully you can learn from my incompetence rather than your own.
I’m writing this series with Josh Lachkovic, managing director at Ballpoint, a performance marketing agency that helped us scale at Yonder. He knows far more about this than I do, which is why he’s here.
Here’s what we’ll cover:
How to track and measure what matters
Building a creative production system
How targeting and experimentation works
Let’s get into it.
Goal: set up server-side tracking and get your metrics in order
At higher spend levels, small measurement issues become expensive. So make sure you get this bit right.
You’ll need engineering support for this, so copy and paste this to the person with the most screens in your office.
Meta’s Conversion API (CAPI) lets you send conversion data directly from your server to Meta, bypassing the browser. Every ad platform will have their own one but they do the same thing.
When someone converts on your site, you capture their details like email and phone number and send that to Meta. Meta then tries to match that person to a Meta account.
The goal is to get your Event Match Quality as high as possible. Meta shows this as a score out of 10. Aim for at least 6, ideally higher. Better match rates mean Meta gets clearer signal on who to show your ads to, which means better (cheaper) performance over time.
And try align your events to standard Meta events where possible (Purchase, Lead, AddToCart, CompleteRegistration) because these come with built-in optimisation learnings across millions of advertisers. You can set up custom ones, but unless you really know what you’re doing, which you don’t, because you’re reading this, then keep it simple.
Finally, sync your CRM with Meta so you can exclude existing customers from your advertising. It’s a small win but we gotta take these wins when we get them.
The metrics that matter most to the success of this part of your strategy are:
Blended CAC which is your total acquisition cost across all channels and
Payback, which determines how aggressive you can be with spend.
Some sort of proxy for LTV - like first-month revenue, repeat purchase rate etc.
These aren’t the same as platform data points like individual ad set or creative performance or funnel effectiveness but I’ll assume you’re following these and won’t get into them too much now.
Daily checks should take about five minutes to look at spend pacing and obvious anomalies. Basically, making sure nothing’s broken.
Weekly reviews of 30-60 minutes to check creative performance, test results, CAC trends, and writing new briefs based on learnings.
Monthly is when you zoom out for half a day to look at cohort analysis, channel mix, and bigger picture strategy. Are we acquiring quality customers? Should we shift budget between channels? Is CAC going up or down over time?
Goal: be able to produce 25+ pieces of creative a week
Do you have what it takes to make 25 good ads every week, indefinitely, while learning from what worked and what didn’t? You’ll need to.
Here’s how this worked at Yonder once we’d figured it out.
Our performance marketing lead (or Ballpoint, before we had one) would review what was working and what wasn’t. We used a tool called MagicBrief that let us search and save high-performing creative from other brands.
From there we’d write a brief which included the insight driving the test (why we think this will work), references from other brands, the funnel target (brand awareness vs conversion), the product pillar, and the format.
Our creative lead would then decide whether to produce in-house, send to an agency, or brief a UGC creator.
When creative came back, we’d review for quality and compliance.
Then ads go live, we monitor early performance signals, and the cycle starts again.
The whole loop - insight to live ad - needs to happen in days, not weeks. If your turnaround is two weeks, you’re not iterating fast enough.
UGC consistently performs well because it feels native, but finding and managing creators is its own challenge.
We found creators through Ballpoint’s network, UGC-specific agencies, and just reaching out directly to people whose content we liked. Just DM people. Most creators are happy to hear from brands.
We’d typically start them on minimum content deals - three pieces to test the relationship - which gives you enough material to see if their style works without over-committing.
Fair warning, they can be so bloody annoying to work with. Not following briefs, picky contract details and missed deadlines used to wind me up so much.
Give yourself the best chance with clear briefs, agreed usage rights (you need to be able to run this as paid media, not just organic), and a tight feedback/revision process. And if they’re really annoying just don’t work with them. You’ve got enough problems.
At Yonder, we used all three at different times and there’s no single right answer.
Ballpoint handled creative strategy and production early on when we were still learning. They got us to maybe 60% of what a full-time hire could do, but the real value was learning what good looked like. That year taught us enough to hire well later. Thanks Josh.
In-house made sense once we had enough volume to justify dedicated headcount and enough knowledge to know what we were looking for. We brought our creative lead in and she could move faster on simple executions than briefing an external team.
Freelancers and UGC creators filled the gaps for specific skills we needed occasionally, or creator content that wouldn’t make sense to produce ourselves.
“While Ballpoint is a full service agency, a lot of the clients we work with now just use us for creative production. They might manage spend in-house but need creative diversity or stronger strategic thinking going into creative. At which point, working with an agency is a great solution.”
— Josh
The mix will change as you scale. Start with more external support, bring things in-house as you learn and grow.
Goal: mix formats and test hooks and messages to find winning ads
This is the bit that took me longest to internalise. Traditional audience targeting through interests, lookalikes and detailed demographics matters far less than it used to.
Meta’s algorithm has gotten good enough that broad targeting usually outperforms narrow targeting, as long as your creative is strong.
This means that creative isn’t just what you say to your audience, it’s also how you reach different audiences (← Read again. Save. Pin. Pocket. It’s important)
We ran both brand campaigns and Advantage+ campaigns, and the key to making broad targeting work is creative diversity.
You need a mix of formats: static images, short-form video, longer video, carousels and a mix of production quality like high-polish studio content alongside iPhone-shot UGC.
You need different creator types: different faces, different demographics, different styles, because this naturally segments your reach without you having to manually target.
And you need different message angles: product features, social proof, founder story, comparison to alternatives and lifestyle aspiration.
Feed the algo with enough variety and it’ll do the rest.
The first few seconds of an ad, what we in the industry call a hook, is the most important element behind good creative. The internet is full of things trying to grab someone’s attention so work hard on making sure they stop to watch your ad.
Once you’ve got hooks that stop people you can test your message. Basically, whatever you want to say in your ad. Could be the core value proposition, whether that’s a product benefit, social proof, founder story, or comparison to alternatives. Play around with it.
Format testing (video vs static, length variations, aspect ratios) matters but usually less than hooks and angles and execution details like colour variations and font choices are your small optimisations once you’ve found a winning concept.
For timing, five days and 20 conversions is the threshold. Before that, you don’t have enough signal to make a confident call. After that, more time rarely changes the verdict.
“If a creative hasn’t hit your benchmark CPA having spent 8-10x your target CPA in budget, kill it. So if your target CPA is £50, give an ad £400-500 before making a call. Sounds brutal, but creative volume matters more than giving under-performers extra chances.”
— Josh
You’ve got your measurement in place and a creative system that produce loads of ads quickly. But things will still go wrong. This is marketing. They always do.
Next week in Part 3, we’ll cover what can go wrong when scaling, how to work with agencies effectively, and how to build your team as you grow.
See you then.
In partnership with Ballpoint
I’ve launched products at Monzo, Wise, and built the brand and marketing team at Yonder from scratch. I write about what actually works in startup marketing for marketers on the verge of breakdown.
If you want help understanding why your brilliant product isn’t selling itself, find me on LinkedIn.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.