Last-click attribution is the analytics equivalent of judging a sales team only by who closes the deal, while ignoring everyone who identified the lead, built the relationship, and handled the objections.
And yet it remains the default measurement model for the majority of marketing programmes operating in the GCC today.
The mechanics of the problem
When a customer buys after clicking a Google Search ad, last-click gives 100% of the credit to that search ad. The display campaign that introduced the brand three weeks earlier gets nothing. The email that triggered the return visit gets nothing. The Instagram video that built category interest gets nothing.
The outcome is predictable: budgets shift towards channels that appear at the end of the customer journey — typically brand search and retargeting — while the channels that drive upper-funnel awareness are systematically defunded because they look unproductive in the attribution report.
This creates a vicious cycle. Brands cut awareness spending, then wonder why their branded search volume is declining. They increase retargeting spend, but without a full funnel feeding it, performance deteriorates. The attribution model rewards the behaviour that is killing the programme.
What the data actually shows
When we rebuild attribution models for new clients, the pattern is remarkably consistent. Under last-click, paid search typically receives between 60-80% of conversion credit. Under a data-driven or time-decay model, that typically falls to 30-45%, with the difference redistributed to display, video, and social channels that appear earlier in the journey.
This has real implications for budget allocation. In a hypothetical programme with AED 500K/month in media spend, the reallocation can be significant — often shifting AED 100-150K from late-funnel to mid-and-upper-funnel channels. That shift tends to produce better overall returns because the full funnel is being fed appropriately.
The compounding problem of cookie deprecation
Last-click models were already inaccurate before privacy changes. Now they are increasingly blind. Safari's ITP has been blocking third-party cookies for years. Firefox followed. Chrome's deprecation, while delayed, is proceeding. iOS 14 shattered mobile attribution.
In markets with high smartphone penetration like the UAE — where over 98% of the population is connected via mobile — the proportion of customer journeys that cross multiple sessions, devices, and browsers is enormous. Last-click attribution captures an increasingly small fraction of these.
The implication is that last-click programmes are not just crediting the wrong channels — they are crediting channels that happened to appear in the small fraction of journeys they can still see.
Building a better measurement approach
The solution is not to find a perfect attribution model — no such model exists. The solution is to combine multiple approaches to build a more complete picture:
Data-driven attribution within platform tools (Google's data-driven model, Meta's value-based bidding) provides a better within-platform view but cannot see across platforms.
Marketing mix modelling provides an econometric view of channel contribution over time, capturing offline effects and handling the signal loss that is increasingly problematic for digital-only models.
Incrementality testing — geo holdouts, ghost bidding, matched market tests — provides the most reliable measure of true causal impact, separate from correlation.
Used together, these approaches give marketers a significantly more accurate picture of where their budget is actually working. The investment in measurement infrastructure typically pays back within the first campaign optimisation cycle.
The organisational challenge
The deeper problem with last-click attribution is organisational: it is simple, it is built into every platform natively, and it is familiar. Changing it requires investment in measurement infrastructure and the willingness to accept that historical performance data may look different under a new model — which can be a difficult conversation with leadership.
The alternative is to continue measuring marketing the way it was measured in 2015, in a world that bears almost no resemblance to 2015. The cost of that is paid quietly, in budget allocated to channels that look productive in the report but are not actually driving growth.
