WTF is attribution!?
There's no holy grail in attribution, but there is a model that works best for you. Suzanna Chaplin, CEO of esbconnect breaks it down.
There's no holy grail in attribution, but there is a model that works best for you.
And, to figure that out, you need to understand WTF attribution actually is.
So, as someone whose middle name may as well be attribution, I’ve put together your very own cheat sheet.
What is attribution?
Attribution is the process of working out which touchpoints actually deserve credit for influencing someone’s decision to buy. These can include the likes of ads, emails and affiliate content.
For years,
last-click attribution has been the default measurement, in which the final touchpoint a customer interacts with before they buy gets all the credit. Last-click is a brilliant starting point, but it’s not an end point. It's easy and it's the best model for any business getting going. Lean on it for too long, however, and it can encourage the wrong behaviours.
That's where
multi-touch attribution comes in. Rather than crowning one single touchpoint, it spreads credit across multiple interactions along the path to purchase.
A popular multi-touch attribution tool is
linear attribution. This is a method that shares the credit for a conversion equally across every touchpoint in a customer's journey. For example, if a buyer interacts with four different channels before purchasing, each channel receives 25% of the credit.
For more on the different types of attribution models, this resource from Adobe is a great one.
There’s no perfect fit. But there’s one that’s right for you.
I’ll say this upfront: there’s no right or wrong answer when it comes to attribution, no perfect model. But what you can find is a model that gives you the best prediction of whether you’re spending in the right place and a tool that works best for your business.
If you’re waiting to find the “correct” way to attribute a sale, you’ll be waiting forever. The real question you should be asking yourself isn’t “is this right?”; it’s “does this give me the best evidence to make a good decision?"
Why does attribution matter?
Because businesses need to make money. And there’s only so much a business can afford to pay for a sale. Attribution is about working out the true value of that sale and figuring out who the right person is to pay for it.
Get it wrong and you’re unfairly allocating payment, or paying the wrong amount for the wrong thing. And that has real consequences for your partners and margins.
How to choose the right attribution model for you
My advice would be to pick one model and stick with it for a defined period. Test it and learn from it. But remember: as your goals change, your model may change.
The right model depends on what you’re trying to achieve. Ask yourself:
- Are we optimising for brand awareness or driving new customers?
- Do we want diversity across our partner mix or a heavy weighting towards one type of partner?
- Do we want balance across the whole funnel or are we happy leaning on one part of it?
- What cost are you willing to pay for the model you choose?
Every model has a trade-off, it’s important to figure out what yours is.
| Model | What it rewards | Best for | Main downside | Fraud/gaming risk | Discount & margin risk |
|---|---|---|---|---|---|
| First-click | Whoever introduced the customer to the brand | Proving the value of awareness and top-of-funnel content | Ignores everything that happened between discovery and purchase, overstates awareness channels | Low direct fraud, but can incentivise cheap, high-volume traffic (bots, incentivised clicks) purely to plant the first cookie | Low. No reason for a partner to discount just to be first in the journey |
| Last-click | Whoever was interacting with the customer right before they bought | Simple reporting, understanding what closes a sale | Rewards interception of a conversion that was already going to happen, ignores everything earlier in the journey | High. This is the model exploited in cookie stuffing, coupon extensions and cashback sites intercepting the final second, as seen in the Honey lawsuits | High. Publishers compete on discount depth to win the last click, eroding margin and training customers to wait for a code |
| Linear | Every touchpoint equally | A fair, easy to explain baseline across channels | Treats a passing impression the same as a real driver of intent, dilutes the signal | Moderate. Harder to game for cash directly, but rewards "touchpoint stuffing", low-value impressions flooding the path to pick up fractional credit | Low to moderate. No single partner captures enough credit to make heavy discounting worthwhile on its own |
| U-shaped (position-based) | First and last touch most (often 40/40), middle touches split the rest | Businesses that care about both discovery and closing | The 20% left for the middle can badly undervalue nurture channels like email that keep a prospect warm | Moderate. Two specific moments carry outsized reward, so there's an incentive to manufacture the "final touch", similar mechanics to last-click but for a smaller slice | Moderate. Same discount pressure as last-click on the closing side, softened because it's only worth 40% of the credit rather than 100% |
Attribution FAQ
Is AI changing attribution?
There’s no denying that AI is the next chapter. The challenge: we have to build systems that work for AI agents
and humans - and that makes identifying fraud a lot harder. Also, zero-click will be huge as AI summaries take over - we’re already seeing early signs of this in email. The opportunity: AI can process huge volumes of complex data points and give you a clear, fast view that would take a human team far longer to produce, so it could make attribution
less
complex, if used well.
What is media mix modelling (MMM), and how is it different from attribution?
MMM is a top-down statistical model that looks at aggregate spend and outcomes over time (weekly or monthly, across TV, PPC, email, affiliate, price, seasonality, even weather) and estimates how much each channel contributed to total sales. It doesn't touch individual user journeys or cookies at all.
How do attribution and incrementality differ?
This is the most commonly confused pair of terms in adtech. Attribution answers "who was present on the path that converted?" Incrementality answers "would this sale have happened anyway, without that touchpoint?" A channel can look great on attribution (lots of last-click credit) and be close to zero incremental (it was just intercepting people who were already buying). That's exactly the last-click and coupon-extension problem covered earlier.
Isn't Google Analytics the solution to all this?
No. Firstly, Google is a publisher so it will always have a bias and prioritise its referrals over other traffic. GA4 is a single-source, first-party analytics tool. It only sees what happens on your own site or app, and only for the sessions it can stitch together. It can't see offline conversions, other platforms' ad data natively, or cross-device journeys unless you've built that plumbing yourself. GA4's default model is data-driven attribution, which is a genuine improvement on last-click, but it's still attribution, not incrementality, and it's still bound by whatever consent and tracking signal actually reaches it.
Should attribution platforms running the "same" model give the same answer?
No, and this trips people up. Two platforms both running "data-driven attribution" or "linear" can produce different numbers because of:
- Different tracking windows (a seven-day click window vs 30-day)
- Different identity resolution (cookie-only vs logged-in ID vs probabilistic matching)
- Different data completeness (a platform can only attribute the touchpoints it can see, so a walled garden like Google or Meta will always look disproportionately strong in its own reporting)
- Different definitions of a "touchpoint" or "conversion" itself
This is why you'll often see total attributed revenue across all platforms add up to well over 100% of actual sales, everyone's counting some of the same conversions.
Regarding compliance, does tracking work regardless of whether someone accepts or rejects cookies?
No. If a user rejects cookies, or is on a browser that blocks third-party cookies by default, attribution tied to that identifier breaks. As of 2026 that's a meaningful chunk of traffic: Safari's Intelligent Tracking Protection and Firefox's Enhanced Tracking Protection have blocked third-party cookies by default for years, and Chrome, after repeatedly delaying then abandoning forced deprecation in 2024, now runs a user-choice model rather than blocking by default. So the honest picture is fragmented rather than uniform, tracking reliability now depends heavily on browser mix and consent rate, not just on which model you've chosen.
Under UK GDPR and PECR, non-essential tracking cookies legally require consent before they're set, so if someone declines, first-click, last-click, whichever model you've picked simply has no signal to work with for that user.
How does consent affect different publishers in attribution?
Cashback and loyalty publishers get a genuine structural advantage under consent rules. The ICO and CNIL treat their tracking cookies as strictly necessary, so affiliate platforms like Awin can set them even when a user rejects cookies, because the cookie is needed to deliver the cashback or loyalty service the user actually signed up for. Every other publisher type only tracks once the user has accepted.
Suzanna Chaplin
Founder & CEO at esbconnect
Suzanna is CEO of esbconnect, a role she assumed when she led a management buyout of the company in 2003, after almost nine years with the firm. Since then, she has transformed esbconnect from a legacy email marketing company into a strategic data and technology business that’s helping brands overcome today’s marketing challenges, while remaining centred on email.
Suzanna’s vision was to recognise that the humble email address is the passport to the internet – a critical piece of information can be used to identify real individuals online, irrespective of the device or platform they are on. In doing so, it can be used as a central identifier across digital and offline channels. Suzanna is steering the company’s evolving suite of products and services through the cookieless marketing ecosystem - helping brands to achieve KPIs across social and programmatic environments, thanks to first-party data strategies fit for the modern age.
Since taking on the role of CEO, Suzanna has also transformed how the company operates, investing vast amounts of time in helping people think, sell and approach the business differently, and instilling in the team the confidence to challenge clients and deliver real value. She also continues to work closely with the tech team on product and market innovation.
Prior to esbconnect, Suzanna spent just over five years at Freemax Media, working on lead generation and business development by building in-depth partnerships across Freemax Media’s in-house brands. Suzanna also managed and grew Freemax Media's expansion into overseas markets, including Denmark, Italy and Spain, and built and managed a diverse portfolio of accounts across many different verticals, including Land Rover, Halifax, House of Fraser and Achica.
