For years, almost everyone in marketing measured the same thing in the same way: the last click before the sale took all the credit. That logic is beginning to break down, and the AI-powered marketing measurement It is the approach that is gaining the most traction amongst brands that need to demonstrate results. At Vandelay, we see this every week with clients who invest wisely and sell well, yet still cannot explain which part of their budget is actually delivering results.
This article isn’t about trends. It’s about a fundamental shift in how value is attributed to every euro invested, why the old model has fallen short, and what you can do about it without needing a team of data scientists.
Why last-click attribution no longer tells you the truth
Last-click attribution makes a convenient assumption: that the channel someone clicked on just before making a purchase is responsible for the sale. In practice, a customer’s actual journey is far more convoluted. They see an advert on Instagram, search for your brand on Google a few days later, read a review, ask an AI assistant and, finally, click through from an email. The last click gets the credit, but it rarely did the heavy lifting. This is where AI-powered marketing measurement starts to make a difference, because it’s based on the idea that no channel works in isolation.
The problem has been exacerbated for two reasons. The first is signal loss: privacy restrictions, the phasing out of third-party cookies and cross-device tracking limitations are leaving huge gaps in the data. The second is that consumers are spending more and more time on sites that traditional attribution barely captures, ranging from creator-generated content to conversational search engines.
What the data tells us: traditional measurement is falling short
It is not a matter of perception; it is documented. According to the report State of Data 2026 produced by IAB and BWG Global, and reported by the specialist publication MarTech, three out of every four marketing professionals They recognise that their current measurement methods — attribution, incrementality and media mix models — do not provide them with the speed, accuracy or confidence they need.
The same report points out that models are looking in a direction where attention no longer lies. 77% of professionals admit that the gaming is under-represented in its media mix models; the e-commerce It appears to be underestimated for the 50%, and the creator economy for the 48%. When the model fails to take into account the channels where the audience actually is, the result is predictable: poorly allocated budgets and decisions made on the basis of incomplete information.
This is where technology comes in. The report estimates that artificial intelligence could unlock some $26,300 million in advertising investment value by making measurement faster and more adaptable, and a further $6,200 million in productivity gains by automating data cleaning and classification tasks. It’s not magic: it’s about stopping wasting time balancing spreadsheets so you can focus on understanding what really drives sales.
Marketing measurement using AI: from the click to the unified model
The central idea is simple to explain but difficult to put into practice: rather than relying on a single metric, the AI-powered marketing measurement It combines various sources and reconciles them using models that learn continuously. This approach is underpinned by three key elements.
Self-updating media mix models
Media mix models (MMM) have been around for decades, but they used to be updated once or twice a year and required highly technical teams. AI is changing that frequency. A specific example: in February 2026, Google launched a no-code scenario planner built on its Meridian model, designed to enable a marketing team to translate an MMM into budget decisions without any coding, as reported by MarTech. What used to be a quarterly process is now becoming monthly, weekly or almost real-time.
Incrementality always active
Incrementality answers the question that really matters: would this sale have taken place anyway without the advert? Traditionally, this was measured through one-off tests a couple of times a year. The new approach turns it into an ongoing practice, with continuous experiments that adjust investment based on the actual – rather than apparent – contribution of each channel.
Reconciled allocation, not eliminated
Attribution does not disappear; it simply takes on a different role. Rather than being the sole source of truth, it intersects with incrementality and the MMM. When the three models diverge, that divergence ceases to be a nuisance and becomes a signal: it indicates where to look more closely. It is AI that makes it possible to reconcile these sources at a speed that is impossible to match manually.
The blind spot of AI: what attribution fails to see
There is an interesting paradox. The very same artificial intelligence that is improving measurement is also creating a new blind spot. More and more people are discovering brands and products by asking conversational assistants and generative search engines – an influence that traditional attribution does not capture because it almost never results in a trackable click. If you’re interested in this aspect of the phenomenon, we cover it in depth in our article on the Signs of the future of marketing with artificial intelligence, as seen at DES 2026.
The practical implication is clear: measuring only what is trackable amounts to underestimating the true impact of much of your branding work. That is why unified models are gaining ground, as they seek to estimate the total contribution rather than just the part that leaves a technical trail. Ultimately, that is the aim of AI-powered marketing measurement: to get as close as possible to the full picture, not just what is easy to quantify.
How we apply AI-powered marketing analytics to your business
The theory sounds good, but it’s the execution that makes the difference. This is what we at Vandelay recommend to SMEs or growing brands that want to modernise their measurement without taking unnecessary risks.
Start by validating your data before modelling anything: a model fed with dirty data will only produce convincing but erroneous conclusions. Next, stop treating attribution, incrementality and MMM as silos; the real value lies where they intersect. Include the channels you usually ignore, from connected TV to retail media, because that is where misallocated investment often lurks. And always ensure that budget recommendations suggested by AI are reviewed by a human: the technology makes the proposals, but the responsibility still lies with people.
This way of working ties in directly with how we understand the rest of the strategy. Robust measurement is what lends credibility to the AI-powered hyper-personalisation and what gives meaning to the agentic marketing: Without proper measurement, increasing automation simply means making mistakes more quickly. If you want to delve deeper into the analytical foundations, our guide on advanced data analysis applied to advertising.
It is worth adding a note of caution. The IAB report itself warns that half of professionals anticipate legal, privacy or accuracy challenges over the next two years, and that the main concern is the “black box” problem: models whose conclusions cannot be explained or traced. That is why transparency and data governance are not mere details; they are an integral part of the work.
If you’ve been investing in marketing for some time and the feeling of not knowing what works sounds familiar, it’s not your fault or your team’s: it’s simply that the rules of measurement have genuinely changed. The sensible approach isn’t to chase every new tool, but to build a system you can trust. At Vandelay, we help you set up exactly that – with a clear-eyed approach and no smoke and mirrors – by applying AI-powered marketing measurement to your specific situation. If you’d like to review how you’re currently measuring your results and what could be improved, Let's have a chat and we put figures to the conversation.
Source of the data cited: report State of Data 2026 from IAB and BWG Global, reported by MarTech.
English (UK) 
Español