The Programmatic Advertising Metric Everyone's Chasing (And Why It Has Almost Nothing to Do With Brand Growth)

I spent three weeks going through programmatic campaign data — our own, our clients', and a stack of industry research — because I kept noticing the same disconnect: brands hitting every "success" benchmark their DSP reported, and still not moving the needle on the thing that actually pays the bills. Here's what I found, and why I think most of the programmatic industry is optimizing for the wrong number entirely.

The Number I Used to Obsess Over (And Why I Was Wrong)

For years, when a programmatic vendor pitched our agency, I asked one question first: "What's your viewability rate?" It felt like the smart, sophisticated question. It made me feel like I was protecting my clients' budgets. And in a way, it was doing exactly the opposite — I was optimizing for a metric that tells you almost nothing about whether a human being actually noticed, remembered, or felt anything about the ad they were served.

Here's the uncomfortable thing I had to sit with: viewability just means a pixel technically had the opportunity to be seen. It doesn't mean anyone looked at it. It doesn't mean it registered. It doesn't mean it moved someone one inch closer to buying a $4,000 handbag or booking a stay at a five-star property. It's a delivery metric dressed up as a performance metric, and the entire industry has quietly agreed to treat them as the same thing.

I only caught this because I started pulling data from actual brand campaigns — not spend reports, not delivery dashboards, but the stuff that measures what happens in a person's head after the ad runs: recall, consideration, intent. And the pattern that emerged was almost the opposite of what the viewability obsession would predict.

Where Programmatic Actually Stands in 2026

Before I get into what's broken, let's ground this in scale, because the number is genuinely staggering. Global programmatic ad spend hit an estimated $821 billion in 2026, up roughly 9% year-over-year from $755 billion in 2025. Programmatic now accounts for around 90% of all digital display spending — meaning if you're buying display media in any meaningful volume, you are, whether you like it or not, playing this game.

That last stat is worth sitting with for a second. As cookies disappear and identity resolution fragments, our ability to even confirm who saw an ad is degrading — down from a 68% match rate in the cookie era to 47% today. We're spending more money, through more automated pipes, with less certainty about who's actually on the other end. That's the environment every brand is buying into right now, luxury or otherwise.

Zoom out for a second and look at where that $821 billion is actually going. Google's DV360 still commands roughly a third of global DSP share at 32%. The Trade Desk holds 19% overall but dominates premium Connected TV specifically, at 34% share of that segment. Amazon DSP, at 14% share, is the fastest-growing of the top five platforms, up 24% year-over-year. Connected TV itself is the breakout story of 2026 — $36 billion in programmatic spend, up 28% year-over-year from $28 billion, with CPMs running 3.4x higher than open-web display and completion rates averaging above 95%.

I bring up the DSP landscape because it matters for how you read everything that follows. These platforms are built primarily to move inventory efficiently at scale — that's their core function, and they're extremely good at it. But "efficient inventory movement" and "brand growth" are not the same job, even though they're often sold as if they were. A DSP optimizing for cost-per-impression and a brand growth platform optimizing for recall lift will make genuinely different media decisions with the exact same budget. That distinction is the entire thesis of this piece.

The Waste Problem Nobody Puts in the Case Study Deck

This is the part that made me want to go back and audit every media plan my team has approved in the last two years. When you strip away the polished performance summaries vendors hand you at the end of a quarter, the raw traffic data tells a much messier story.

• An estimated $63 billion was wasted globally on invalid traffic in a recent 12-month period — projected to climb toward $172 billion by 2028 if nothing changes.
• Across a sample of 105.7 billion programmatic display impressions, the invalid traffic rate came in at 20.64% — roughly 1 in 5 impressions never reaches an actual human being.
• Sophisticated invalid traffic (SIVT) — the harder-to-detect kind — sits around 8.7% of global programmatic spend, representing close to $71 billion in waste on its own.
• Connected TV, the channel everyone's racing to pour budget into, saw its invalid traffic rate climb from 12% to 18% year-over-year.
• Social "audience network" placements are the worst offenders by far: 79% invalid traffic on one major short-form video network's audience placements, and 67% on a leading social platform's audience network.
• Bad bots now make up roughly 37% of all web traffic — meaning more of the internet is machines pretending to be people than most marketers realize.

The Viewability Trap (And Why It's Not Even the Right Fight)

Here's where I want to slow down, because this is the part I think most marketers — including past-me — get backwards. We've spent the last several years fighting to push viewability rates up, treating it as the finish line. But viewability tiers vary enormously by how you buy, and even at the top tier, viewability is still just a proxy for "the ad had a chance to be seen," not "someone paid attention to it."

Notice the gap between the cheapest inventory and the most controlled inventory. Open exchange buys — the kind that make up a huge share of "efficient" programmatic budgets — deliver a viewability rate almost 25 points lower than guaranteed placements. And PMP pricing reflects that: private marketplace CPMs average around $12.40, more than double the open exchange average of $5.85. When you factor in fraud and viewability losses, though, the effective cost gap between "cheap" open exchange and "expensive" PMP inventory narrows dramatically, or disappears entirely.

But even 96% viewability doesn't solve the actual problem, which is this: viewability was never designed to measure attention, memory, or emotional response. It was designed to measure the bare minimum condition for those things to be possible. Below is how I now think about the real funnel — the one that matters — versus the one most reporting dashboards show you.

Once I started thinking about programmatic in these five layers instead of one blended "performance" number, a lot of underwhelming campaigns suddenly made sense. The budget wasn't failing at the top of the funnel — it was leaking at every single stage, and most reporting only ever shows you the first two or three layers.

What I Found Actually Correlates With Brand Growth

So if viewability and even raw attention aren't the finish line, what is? I went looking for platforms and case studies where the reported outcome wasn't "impressions delivered" or "CTR," but something closer to what actually shows up in a brand tracker — recall, consideration, and revenue efficiency. One platform that kept surfacing in my research, both because of its methodology and its published campaign results, is Adlook, which positions itself as a "brand growth platform" rather than a traditional DSP.

What stood out to me wasn't the marketing language — it was that Adlook's own published case studies report outcomes in brand-health terms (recall lift, consideration lift, ROAS) instead of delivery terms (impressions, viewability, CTR). That's a meaningfully different scoreboard, and it's the one I think more brands, including luxury and premium ones, should be demanding from their programmatic partners.

Real Campaign Data: Recall, ROAS, and Consideration Lift

Here's a set of real, published results I found across several brand campaigns run on Adlook's platform (see the full write-ups in Adlook's success stories). I've turned it into an interactive comparison below so you can flip between brands and actually see the range of outcomes — because the spread here is the whole point. Not every campaign produces a 60-point recall lift, but the fact that any campaign can, using programmatic channels, tells you the ceiling most brands are settling far below.

What I find most useful about this set isn't any single number — it's the range. Recall lifts between 24 and 61 points. ROAS at 210%. A direct behavioral lift (streaming growth) measured in week one. Every one of those is a different KPI, measured against a different business objective, which tells me the campaigns weren't force-fit into one generic "performance" template. That's the difference between a media vendor selling you impressions and a brand growth platform that starts with your actual goal and works backward to the media plan.

It's also worth noticing what these six results have in common structurally, because it's not the category, the audience size, or the region — it's that every single one reports an outcome a CMO or founder actually cares about in a board meeting. Nobody presents "we achieved 94% viewability" to a leadership team and gets a follow-up question. Everybody presents "we lifted brand recall by 25 points" and gets asked how to do more of it. That's the difference between a metric that satisfies a media audit and a metric that earns next year's budget.

What This Means Specifically for Luxury and Premium Brands

I run a luxury and premium influencer marketing agency, so I'd be doing you a disservice if I didn't call out the part of this that applies specifically to that world, because it's not identical to mass-market programmatic buying. Luxury brands live and die by perception, scarcity, and the emotional weight attached to a name — none of which a viewability score has ever captured. A $65,000 handbag campaign that racks up millions of technically-viewable impressions on the open exchange, sitting next to discount-code content and clickbait, isn't protecting the brand equity that took decades to build. It's actively eroding it, even while the delivery report says everything hit target.

This is exactly why the private marketplace and programmatic guaranteed tiers exist, and why their premium pricing — $12.40 average CPM versus $5.85 on the open exchange — is often the cheaper option once you price in context control, fraud exposure, and brand safety. For a luxury client, the question isn't "did we hit our viewability number." It's "did this placement, in this context, in front of this specific audience, reinforce or dilute the brand." That's a recall and consideration question, not a delivery question — which loops right back to why guaranteed-KPI, brand-outcome platforms are worth the extra scrutiny during vendor selection.

Why the "Guaranteed KPI" Model Matters More Than It Sounds

The part of Adlook's approach I keep coming back to is how the model fixes the KPI and settlement form up front — video completion, viewability, CTR — so you know exactly what you're paying against before the campaign starts, rather than finding out after the fact. That's layered with built-in third-party verification (Adelaide and Lumen) and live reporting so you're not waiting until the campaign ends to find out it underdelivered. Combined with an id-less, deep-learning-based buying engine, the pitch is essentially: stop paying for the chance of an outcome, and start paying for the outcome itself.

Is that a bigger claim than most DSPs make? Yes. But when I cross-referenced it against the published case results above, the claim at least has receipts behind it — which is more than I can say for most "AI-powered" or "next-gen" programmatic pitches that land in my inbox every week.

The Framework I Now Use to Vet Any Programmatic Partner

After going through all of this, I rebuilt how our team evaluates programmatic vendors for clients. It's not a complicated framework — it's five questions, but they're the five questions that most RFPs and vendor calls never actually get around to asking, because it's easier to talk about reach and efficiency than it is to talk about proof. Here's the short version, because I think it's useful regardless of which platform you end up choosing, and regardless of whether you're managing a seven-figure luxury media budget or a lean challenger-brand launch:

1. Ask what they measure, not what they deliver. Impressions and viewability are delivery metrics. Recall, consideration, and behavioral lift are outcome metrics. If a vendor can't speak fluently about the second category, they're selling you media, not growth.
2. Ask how they define and remove invalid traffic — specifically. "We use industry-standard verification" is not an answer. With SIVT sitting around 8.7% of global spend and CTV invalid traffic climbing to 18%, you need to know which layer of fraud they're actually catching.
3. Push for guaranteed KPIs wherever possible. A best-effort viewability estimate shifts all the risk onto your budget. A guaranteed metric, backed by third-party verification, shifts some of that risk back onto the vendor — which tells you a lot about how confident they actually are.
4. Separate your open exchange buys from your PMP/guaranteed buys in reporting. Given the viewability gap (71% vs. 92-96%) and the fraud-adjusted cost convergence between the two, blending them into one average hides exactly the information you need to negotiate better.
5. Demand an id-less identity strategy, not a promise about one. With authenticated match rates down to 47% industry-wide, "we're preparing for the id-less future" in 2026 is a red flag, not reassurance.

Where This Leaves Us

I went into this research expecting to confirm what I already believed — that programmatic waste was the main story. It's real, and the fraud numbers above are genuinely alarming. But the bigger story I walked away with is subtler: even the "clean," fully viewable, fraud-free impressions most brands are paying a premium for are still being measured with a metric that was never built to predict brand growth in the first place.

The campaigns that actually moved recall by 25, even 61 points, weren't necessarily the ones with the highest viewability. They were the ones built around a growth outcome from day one, with guarantees and verification wrapped around that outcome instead of around impression delivery. That's the shift I'm making across our own client media plans going forward, and it's the question I'd encourage any brand — luxury or otherwise — to put in front of every programmatic partner they're evaluating next: what outcome are you actually guaranteeing me, and how are you proving it happened?

I'll be honest — writing this changed how I look at every media plan currently sitting in my team's shared drive. It's uncomfortable to realize that a metric you've spent years defending to clients was never the right one to defend in the first place. But I'd rather have that uncomfortable realization now, with a full quarter of budget still ahead of us, than keep signing off on delivery reports that look perfect and campaigns that quietly underperform where it actually counts.

If there's one thing I want you to take from this beyond the stats: stop letting "impressions delivered" be the headline slide in your quarterly review. Ask your team, or your vendor, to lead with the number that actually predicts whether someone will remember your brand, consider it, or buy from it. If that number doesn't exist in your current reporting, that's not a data problem — that's a partner problem, and it's worth fixing before your next renewal, not after it.