Most media plans work by finding the best-performing audience and pouring the budget into it, which is fine until the audience has been covered and the campaign starts buying frequency instead of reach. Second-month performance dips, cost per result goes up, and the post-mortem concludes that the audience was fatigued when the real problem was that there was only ever one of them.
SodaStream’s Drink Better campaign had a better strategy than this, with two first-party audiences of deliberately different sizes running side by side on one budget, each curated to do a particular job. Across the full campaign they reached more than 860 000 people, and the average person saw the ad fewer than twice.
The structure behind it
The brief called for both national awareness and qualified landing page traffic for at-home sparkling water, so the campaign was built with an audience for each of these two objectives in mind.
The first was small and made up of people who already buy in the category, drawn from retail appliance and home beverage buyers. The second was much bigger and broader, and made up of online retail shoppers across lifestyle and homeware. That audience size gap was by design: the smaller audience was there to bring quality and the bigger one was there to bring volume, and between them the campaign never ran out of new people to reach. Six ads ran across both audiences, led by a sustainability message about replacing single-use plastic bottles.
KEY INSIGHT 💡
One audience, however well it performs at first, eventually just means showing the same people the same ad again. A second, much larger audience is what keeps a campaign finding people it has not reached yet.
High frequency is almost never a strategy. It’s a symptom of an audience that’s too small for the budget behind it.
What it delivered
Over the period, the campaign reached more than 860 000 people at an average frequency of 1.9, with no spike in the second month that usually signals an audience running dry. It drove 7 680 qualified landing page views from 16 648 link clicks and generated more than 22 000 interactions, including social shares and new Instagram followers.
Measured against a single-audience benchmark, the split delivered a 78% increase in landing page views and a 102% increase in link clicks.
Both audiences did roughly what buying data said they would do:
The smaller one brought better quality traffic, turning 49% of its clicks into real page visits and costing less for each one.
The bigger one accounted for 70% of the people reached, carrying most of the volume.
Someone who has bought an appliance is closer to buying another one than someone who merely browsed a homeware page, and that difference shows up in the results in a way an interest segment cannot promise in advance.
Creative did its share of the work. One ad, about replacing single-use plastic bottles, drove 87% of all the traffic, and each of those visits cost 30% less than the campaign average. Knowing what people buy tells you who to talk to. They still need a reason to care.
Why this depends on first-party data
Matching breadth to an objective is not new, and any planner can put a wide audience behind a reach goal and a tighter one behind conversion. What broad targeting cannot tell you is what an audience is actually made of before the money is spent, because interest-based segments are inferences about behaviour and you only discover what you bought once the report comes back.
First-party audiences are built from what people have actually purchased, so you can see how big an audience is, how close it sits to a real purchase and roughly what it is likely to cost you, all while the plan is still being written. Flow Platform’s insights reveal first-party data audiences frequently deliver as much as 50%+ higher conversion rates, and that lift matters, but the planning advantage is the quieter but just as valuable part: you know what you are buying before you buy it, so a budget split becomes a strategic decision rather than a hedge.
While the theory is easy to agree with, practically running audiences from two different retailers inside a single campaign is not that simple: two buying processes, two sets of reporting, and no shared view of what either audience actually did.
This is the power of Flow's Audience Marketplace, where you can combine and curate audiences for your campaign from hundreds of first-party data segments, from more than 35 data partners including Woolworths, Bash, Hirschs, Toys R Us and many more.
How to plan it
Begin with the campaign objectives rather than the audiences, because most briefs contain at least two of them, whether that is volume and efficiency or awareness and lower-funnel movement, and treating them separately is what makes the rest of the first-party data plan obvious.
From there, choose an audience suited to each job:
- A big, broad audience will carry volume and you should expect to pay a little more for each result.
- A smaller audience of people who already buy in the category will bring better quality traffic from a smaller pool of people.
- Determine the budget split against the priority of the objective. For example, if reaching unique people is the main objective, you can allocate 70% of the budget to the broader browsing audience and 30% to the smaller, highly qualified audience.
What to do next
Flow’s Audience Marketplace lists first-party audiences from retailers, marketplaces and platforms across South Africa, with clear detail of the underlying behaviours visible before you book. This means you can tailor your strategy, budget and creative with precision because you know exactly who you will be reaching.
Frequently asked questions
Is running two audiences always better than one?
Not necessarily, because a single audience is the right answer when the brief contains a single job. The case for two arises when the objective carries a tension inside it, such as national volume alongside cost efficiency, that no one audience can resolve on its own, so the question to start with is how many jobs the brief actually contains rather than how many audiences are available.
How do I tell the two kinds of audience apart?
Look at what the audience is built from rather than what it has been named, since one assembled from purchases inside or next to your category will behave quite differently from one built on general retail activity, even when both are first-party and both are large. Every audience on Flow carries a tier rating reflecting the strength of the signal behind it, which is the fastest way to read this before booking.
Doesn't the smaller audience always perform better per rand?
It often does on cost per result and it very rarely does on total volume, simply because there are only so many people in it. Pushing more budget through a small audience eventually raises frequency rather than reach, which is precisely the pattern this campaign was structured to avoid.
How should the budget be split between them?
There is no universal ratio, and it should be based on the priority of your objectives. Decide what each audience’s job is and let the proportions follow from that.
Can I run audiences from more than one retailer in the same campaign?
Yes, and this is where the approach becomes genuinely useful, because audiences from different partners are built from different behaviour and therefore widen the range of jobs a single campaign can cover. Buying them through one interface is also what makes comparing their performance meaningful rather than academic.
