Marketing Attribution in 2026: Why Last-Click Is Costing You Sales
Last-click gives 100% of the credit to the final touch and 0% to everything that created the demand. That single flaw quietly reshapes your budget toward the wrong channels.
Ask three ad platforms who drove yesterday's sale and all three will raise their hand. Meta counts it, Google counts it, TikTok counts it, and your Shopify dashboard counts it once. The numbers never reconcile, and the model underneath most of that confusion is last-click attribution: whichever touch happened immediately before the purchase takes all the credit.
For a brand spending real money across several channels, last-click is not a small reporting quirk. It is a budgeting instruction. Every week it tells you to pour money into the channels that close and starve the ones that create demand, and it does it so quietly that most teams never notice the leak.
What last-click actually measures
Last-click answers one narrow question: which ad did the customer touch last before converting? That sounds reasonable until you follow a single order through its real journey. Someone sees a TikTok video, forgets about it, gets retargeted on Instagram a week later, searches your brand name on Google three days after that, and finally converts through a branded search ad. Last-click credits branded search with the entire sale. The TikTok video that started everything gets nothing.
Scale that across thousands of orders and the pattern compounds. Bottom-of-funnel channels look like heroes because they are standing closest to the finish line. Upper-funnel channels look like waste because they rarely take the last step. You cut the TikTok budget, demand dries up over the next month, branded search volume quietly falls, and nobody connects the two because the report still looks fine.
Why the platforms can't fix it for you
Each ad platform optimises to the conversions it can see, and it can only see its own. Meta will happily claim a view-through conversion that Google also claims as a click. Neither is lying exactly; they are both reporting from inside their own walled garden. When you add their numbers together you get a total that is often 130 to 180 percent of the orders your bank actually processed.
The result is double-counted revenue, inflated platform ROAS, and a set of dashboards that disagree with each other and with reality. You cannot optimise spend on numbers that do not add up, and you certainly cannot forecast on them.
What better attribution looks like
The fix is not a single magic model. It is a stack that counts every sale once and gives upper-funnel channels fair credit for the demand they create.
- First-party source of truth: count orders from your own backend (Shopify, your CRM, your database), not from the platforms. One order, counted once, with its real value.
- Multi-touch attribution: distribute credit across the touches in a path instead of handing it all to the last one, so assisting channels are visible.
- Incrementality and geo-lift: for channels that rarely close last-click, measure the lift they create by comparing exposed and unexposed audiences or regions.
- Server-side conversions: send the de-duplicated, backend-confirmed conversions back to the platforms so their algorithms optimise to real outcomes, not inflated ones.
None of this requires throwing away platform data. It requires anchoring to a number you trust, then using the platforms for what they are good at: finding more of the people who actually convert.
Most teams do not have a spend problem. They have a counting problem, and the counting problem is writing their budget for them.
How to start this quarter
You do not need a data team to make progress. Start by pulling every order from your backend into one place and comparing that total to the sum of what the platforms claim. The gap is your double-counting, and seeing it is usually enough to change the conversation.
Then add assisted-conversion reporting so you can see which channels show up early in winning paths. Protect those budgets before you touch them. Finally, wire server-side conversions so the platforms learn from your real sales. Teams that make this shift typically find they were over-crediting branded search and under-funding the prospecting that fed it.
- Last-click credits the final touch and ignores every channel that created the demand.
- Adding platform-reported conversions double-counts revenue by 30-80 percent on average.
- Anchor to first-party orders counted once, then layer multi-touch and incrementality.
- Send de-duplicated server-side conversions back so platforms optimise to real outcomes.
Frequently asked questions
Is last-click attribution ever useful?
Yes, as one lens among several. It is fine for understanding which channel tends to close, but it should never be the only model deciding budget, because it systematically underfunds demand creation.
What is the difference between multi-touch attribution and MMM?
Multi-touch attribution follows individual user paths and splits credit across touches. Marketing mix modeling works at an aggregate level using statistics, so it can measure channels where you have no click data at all, like TV or broad awareness.
Do I need to replace Google Analytics?
No. GA4 stays useful, but treat it as one input, not the source of truth. Reconcile it against your backend orders so you know where and why the two diverge.