Thomas Bienias, Agency and Brand Partnerships, Kayzen dives into mobile programmatic, and why we're still being asked to trust systems we can't interrogate...

Mobile programmatic spent a decade optimising for scale and speed. Visibility did not keep up, and in 2026, agencies and brands are still paying for it.

Here's the uncomfortable truth most platforms would rather you didn't dwell on: as supply chains expanded and optimisation became automated, transparency quietly disappeared. Every intermediary layer, every black-box model, every aggregated report added complexity to a channel that now eats one of the largest portions of digital investment.

Most agency teams can confirm what a campaign delivered. Far fewer can explain how. The question is no longer whether campaigns perform. It's whether agencies can understand, defend, and reproduce the results when conditions change. Right now, most can't, and they're being asked to trust a system they can't fully interrogate.

Cheapest CPM was never the point

Choosing a DSP used to come down to one number, the lowest CPM or CPA you could put in front of the client. That's not the case anymore. In nearly every mobile conversation I've had this year, the question isn't about the media itself, it's about the data behind it: what the team can see, what they can prove is working, and how it's working.

When performance shifts, sophisticated agencies want three answers: where did the budget go? which signals shaped execution? and when something changed, why did the system respond the way it did? Most DSPs are built to confirm outcomes, not explain them.

The legacy DSP problem nobody wants to say out loud

Here's where I'll be blunt. Real mobile brand performance cannot be achieved on a legacy DSP. These platforms were built for desktop, cookie-based and web-first environments, with mobile bolted on afterwards. You cannot retrofit a desktop architecture onto the most personal, most ID-rich channel we have and expect it to perform. Mobile runs on device IDs, often with an MMP utilising app-level signals and SDK-level data that legacy infrastructure was never designed to process effectively.

This isn't just an infrastructural problem, it's also a creative one. A DSP built for mobile renders creative through the SDK, meaning richer, faster, fully interactive formats that work in-app rather than clunky web units squeezed onto a phone. It also enables proper sequential messaging: serving the right creative in the right order across a user's journey, building a narrative from awareness to action rather than firing the same impression repeatedly and hoping for an outcome.

It matters for measurement too. Genuine user-level incrementality, the kind that proves your media drove a net-new user action rather than taking credit for one that would have happened anyway, has largely been the preserve of the social platforms. A mobile-native DSP brings that same rigour to the open app ecosystem: holdout testing and causal measurement outside the walled gardens, on inventory you can actually see. That's the conversation client-side teams are increasingly demanding, and legacy mobile platforms that don't own their own technology simply can't demonstrate. 

The fee question buyers stopped asking

There's one more layer of opacity, and it's the one closest to home: the fee structure itself. Too many mobile platforms still bundle their margin into a blended CPM, so neither the agency nor the client can see the gap between media spend and platform margin. If you can't separate the two, you can't defend your media costs to a procurement team, and you certainly can't prove value to a sceptical brand or your CFO.

Transparent, published pricing should be mandatory, not a concession won in negotiation. Flat fees, clearly stated, with no margin quietly skimmed off in the supply path or buried in an arbitrary tech layer. A buyer who can show a client exactly where every pound goes is in a fundamentally stronger position than one explaining away a number they don't fully understand themselves, particularly now budgets are tighter, and every pound has to work harder.

The non-negotiable questions to ask

"Trust the algo" was acceptable once. It isn't now. Before locking in budgets, here are four key aspects we consider at Kayzen: 

Before the campaign runs: Can the platform show how the brief is interpreted, covering delivery logic, inventory curation, supply path, and how brand KPIs (reach, frequency, attention, consideration) are actually pursued, not just assumed?

During the campaign: Can your team see what's being prioritised in real time and act on it, including whether brand safety and quality inventory are being favoured over cheap volume that flatters a delivery number? 

After the campaign: Does reporting explain the outcome, or does it just confirm it? Can you show a client that your media drove a genuine brand lift and real mobile performance outcomes, or are you being presented with a dashboard that proves spend happened?

Throughout: And finally, can you see exactly what you're paying in media versus platform and tech fees,  and therefore what actually reached a real person in a context that was worth being in? 

Mobile has been given more slack for opacity than it deserves. I think that's over. The agencies and brands winning won't be the ones spending the most or the lowest price point.  They'll be the ones who can still explain, defend, and repeat their results when the client asks how.

The best time to audit your mobile stack was before your last difficult client conversation. The second best time is now.