ThumbAd Whitepaper: Why Affiliate Tracking and Click Attribution Fall Short in Programmatic Advertising

Updated June 2026 | Originally published April 2023

Attribution is arguably the most consequential — and most misunderstood — variable in digital advertising. Get it wrong and you don’t just misread performance; you systematically defund the channels that are actually working and over-invest in the ones that aren’t.

For programmatic advertising specifically, the traditional methods operators tend to reach for — affiliate tracking and click attribution — don’t just fall short. They actively distort how growth is understood and where budget flows.

This article explains why, what has changed since 2023, and what a more honest attribution architecture actually looks like.

The Problem With Affiliate Tracking in a Programmatic Environment

Affiliate tracking was designed for a specific purpose: tracking and crediting conversions to specific affiliate partners in a last-touch or first-touch model. Within affiliate marketing, that logic is coherent. Applied to programmatic advertising, it breaks.

Programmatic operates across multiple touchpoints — display, CTV, mobile, DOOH, audio, native — where a user may encounter your brand several times before converting. Affiliate tracking has no mechanism for this. It assigns credit to whichever vendor introduced the user, meaning that every subsequent programmatic touchpoint — the CTV ad that built intent, the mobile display unit that reinforced it, the retargeting impression that preceded the final conversion — goes unrecognised.

The practical consequence is severe: programmatic channels are systematically undervalued in attribution models that were never designed to account for them. DSPs and programmatic vendors appear to generate fewer conversions than they actually influence, retargeting looks expensive relative to volume, and operators redirect budget toward affiliate channels that show inflated credit — often for conversions that programmatic infrastructure drove.

The CEO of Awin publicly called for industry-wide reform in 2024, highlighting critical flaws in affiliate tracking systems and arguing that inaccuracies and outdated practices are compromising trust between brands and affiliates.  The issue isn’t new — but it is now being acknowledged at scale even within the affiliate industry itself.

For iGaming and other performance-sensitive operators, the shift toward Privacy Sandbox and cookie-based attribution changes means affiliate attribution accuracy is expected to decline further — with some estimates suggesting an 8–15% reduction in deterministic attribution accuracy as third-party cookie infrastructure degrades.  Operators still relying on browser-based affiliate pixel tracking face a structural reckoning.

Click here Attribution: A Narrow Lens on a Wide Journey

Click attribution — crediting the last vendor to serve a click before a conversion — has an intuitive appeal. Someone clicked, then bought. The click caused the purchase.

The problem is that clicks represent only a small fraction of the actual customer journey in programmatic environments. The majority of programmatic conversions originate from viewable impressions, not clicks. Last-click attribution systematically overvalues channels like branded search while undervaluing the upper-funnel awareness generated by programmatic display campaigns — leading marketers to cut funding for programmatic activity that is effectively building intent, simply because it wasn’t the final touchpoint before a sale. 

Research on programmatic incrementality has produced striking findings: the last-touch attribution framework — still the industry standard in many setups — undervalues programmatic display and native channels by as much as 87% when compared against incremental conversion data derived from controlled experiments.  That isn’t a marginal measurement error. That is a fundamental misread of which channels are doing the work.

The nature of online advertising, where most conversions occur due to multiple channels of influence, means the majority of programmatic conversions are classified as view-through — where a user saw an ad, then later converted through a different method than clicking in the same session.  Click attribution captures none of this.

As of 2025, only one in three advertisers can confidently tie ROI to at least half of their programmatic campaigns, with structural challenges in linking exposures to outcomes across devices, media types, and platforms.  This isn’t a data problem. It’s an attribution architecture problem.

Google Analytics Still Doesn’t Solve This

The original version of this article noted that Google Analytics was not built to capture view-through conversions — and that this remained a significant blind spot for operators relying on it as their primary attribution layer.

That observation remains valid, and has become more complex.

The shift to GA4 provided a more event-based measurement model and improved some cross-channel visibility. However, GA4 remains fundamentally limited in its ability to attribute programmatic view-through conversions accurately. It operates on a session and click-based logic that doesn’t align with how programmatic exposure influences behaviour across days or across devices.

The persistent gap between what platforms like Meta report in their native dashboards and what third-party analytics tools show has been a well-documented frustration — eroding confidence, muddying budget decisions, and making cross-platform performance comparisons difficult.  Meta moved to address some of this in early 2025 with attribution model changes, but the underlying structural issue — that no single analytics tool accurately captures the full programmatic conversion picture — has not been resolved.

For operators running programmatic at scale, relying on GA4 as the authoritative attribution layer means accepting a systematic undercount of programmatic-driven outcomes.

The DSP Remains the Most Reliable Programmatic Attribution Source

In programmatic environments, the DSP serving the creative assets has the most direct visibility into impression delivery, view-through windows, cross-device exposure, and conversion events.

This is not because DSPs are infallible — they aren’t, and honest attribution requires acknowledging what can be measured directly versus what must be modelled and inferred. But the DSP sits closest to the actual ad delivery layer, which gives it signal access that post-hoc analytics tools simply don’t have.

CTV return on investment, for example, is increasingly tracked through attribution models that link household-level ad exposures to web visits, app installs, or downstream conversions — with incrementality testing and cross-device retargeting used to confirm causal impact.  This type of attribution is only possible from within the programmatic infrastructure layer, not from an external analytics tool observing events after the fact.

The practical implication: operators should treat their DSP as the primary source of truth for programmatic performance data, supplemented by — not replaced by — external analytics platforms.

The Growing Role of CDPs and First-Party Data Infrastructure

The CDP (Customer Data Platform) argument made in the original version of this article has become considerably more urgent.

Despite Google’s 2024 decision to reverse the forced phase-out of third-party cookies and keep them enabled by default, US programmatic and digital ad spending reached approximately $309 billion in 2024 — a 15% year-on-year increase — reflecting continued growth even amid attribution complexity.  But the fundamental infrastructure risk hasn’t gone away. In late 2025, the French data protection authority levied nearly half a billion euros in combined fines against major platforms for deploying cookies without clear user consent, while the European Data Protection Board launched coordinated enforcement for 2026 focused on GDPR transparency compliance.  The regulatory direction is clear even if the technical timeline has shifted.

For iGaming operators specifically, building attribution infrastructure around third-party signals is a structural liability. The alternative is a CDP-anchored first-party data strategy — collecting, unifying, and activating owned signals across CRM, app events, web behaviour, and payment data, then feeding those signals into programmatic systems for targeting, modelling, and attribution.

CDPs enable advertisers to turn first-party customer data into a single source of truth regardless of its origin — supporting both attribution and cookieless activation while maintaining privacy compliance and data governance. 

When first-party data is integrated with a DSP’s attribution infrastructure, operators gain something neither system provides alone: deterministic matching on owned users combined with probabilistic modelling for new and non-logged-in audiences. This is the measurement architecture that reflects how programmatic actually works.

Incrementality: What Attribution Should Actually Answer

The most sophisticated shift in attribution thinking over the past two years is the move toward incrementality measurement — answering not “which channel got credit?” but “would this conversion have happened without this ad?”

The cost of not measuring incrementality increasingly exceeds the cost of testing. As platform attribution degrades due to privacy restrictions and competition intensifies, distinguishing truly incremental investments from attributional mirages has become essential for operators making serious budget decisions. 

Incrementality testing via holdout groups — running controlled experiments where a portion of the audience is not exposed to ads — gives operators ground truth on whether programmatic is actually driving new behaviour, or simply claiming credit for conversions that would have occurred organically. This is especially important in iGaming, where brand-familiar users are likely to deposit regardless of ad exposure, and where attribution inflation can quietly consume significant media budget.

The operators who will have the clearest view of what is actually working over the next two to three years are those building attribution systems that combine DSP-level signal, first-party data integration, and incrementality testing — rather than relying on a single attribution model to resolve a genuinely multi-layered question.

What Honest Attribution Looks Like

Programmatic attribution is not a single number. It is a set of signals — some measured directly, some modelled, some inferred — that together give operators a working picture of how media investment is shaping customer behaviour.

Directly measurable: click-through conversions, view-through windows, deterministic cross-device matches, first-party signal integration.

Modelled and inferred: CTV influence on downstream mobile conversion, DOOH awareness impact, cross-publisher frequency effects, incrementality versus organic baseline.

The goal is not perfect attribution. The goal is attribution honest enough to make better decisions than the alternative — which, in most fragmented vendor setups, is making decisions based on whichever channel reports the most conversions in its own dashboard.

Affiliate tracking and click attribution will continue to have value in their native contexts. But in programmatic environments, they answer the wrong question — and in doing so, they systematically misdirect the budget decisions that determine whether programmatic works at scale.

ThumbAd designs attribution architectures that reflect real signal behavior, not last-click fiction. If attribution clarity is a live problem in your setup, it’s worth a conversation. 

Notes on the update: Key developments since April 2023 include: Google’s reversal on third-party cookie deprecation (July 2024); Meta’s attribution model restructuring (early 2025); growing regulatory enforcement on cookie consent in EU markets; industry-wide acknowledgment of affiliate tracking reform; and the acceleration of incrementality testing as the gold standard for programmatic measurement.