Resources · Attribution
Last-Touch, First-Touch or Linear? Choosing the Right Attribution Model for Your Shopify Brand
Every model has a blind spot. Here's how to pick one based on what you actually want to measure.
Every D2C founder eventually asks the same question: "Which attribution model should we use?" Last-touch? First-touch? Linear? Or should we just trust what Meta and Google are reporting?
The honest answer is that there is no single correct model. Each one is a lens, and every lens has a blind spot. The mistake most brands make is not picking the "wrong" model. It is using one model to answer every question.
In this article we break down how each model works, where it helps, where it quietly misleads you, and how to choose a setup based on what you are actually trying to measure.
First, a simple customer journey
Let's follow one customer, Priya, as she buys a ₹2,000 skincare kit from a Shopify brand.
1. She watches a YouTube awareness ad from the brand's top-of-funnel (TOF) campaign. She doesn't click.
2. Two days later, she scrolls past an Instagram Reel from the brand. She doesn't click.
3. A week later, she clicks a Meta carousel ad and browses the site.
4. She searches the brand name on Google and clicks a branded search ad.
5. She adds to cart, leaves, and finally buys after clicking an abandoned-cart email.
Five touchpoints. One order. Who gets the ₹2,000? That depends entirely on the model you use.
1. Last-Touch Attribution
How it works: 100% of the credit goes to the last touchpoint before the purchase. In Priya's case, the email gets ₹2,000. Everything else gets zero.
Pros
• Simple and easy to explain. Anyone on the team can understand it.
• Good for conversion optimisation. It tells you which channels are best at closing a sale.
• Easy to track. The last click is the touchpoint you are most likely to actually capture, with UTMs, click IDs and order data all lining up.
Cons and blind spots
• It over-credits bottom-funnel channels. Email, WhatsApp, retargeting, branded search, and coupon or affiliate sites look like heroes, because they are standing at the checkout counter when the customer arrives.
• It starves the top of the funnel. TOF prospecting campaigns, influencers and YouTube look like they drive nothing. Brands that cut them based on last-touch data often see branded search and email revenue decline a few weeks later, because nobody new is entering the funnel.
• It rewards harvesting, not creating demand. Think of a salesperson handing out his referral code to shoppers already standing in the checkout line. Last-touch is the model that pays him the most.
Last-touch tells you who closed the deal. It tells you nothing about who created it.
2. First-Touch Attribution
How it works: 100% of the credit goes to the first touchpoint in the journey. In Priya's case, the Meta carousel ad gets ₹2,000, because it was the first touchpoint actually tracked as a click.
Notice something already? The YouTube ad and the Instagram Reel she only viewed don't show up at all. We'll come back to that.
Pros
• It values discovery. It shows which channels are introducing new customers to your brand.
• Useful for new customer acquisition. If your goal is growing the top of the funnel, first-touch points you toward the channels doing that work.
• It counterbalances last-touch. Looking at both side by side quickly shows you which channels are openers and which are closers.
Cons and blind spots
• It ignores everything that happened after. The retargeting, email and search that nurtured Priya to purchase get zero credit.
• The "first touch" is often not the real first touch. Cookies expire, people switch devices, and iOS privacy features and ad blockers break tracking. For considered purchases, the true first touch may be weeks old and completely invisible.
• It over-credits cheap, broad reach. Low-intent traffic that happens to land first can look more valuable than it really is.
First-touch tells you who opened the door. It doesn't tell you who walked the customer to the counter.
3. Linear Attribution
How it works: credit is split equally across every tracked touchpoint. In Priya's case, with three tracked clicks (Meta, Google and email), each gets roughly ₹667.
Pros
• It acknowledges that journeys have multiple steps. Nobody gets 100% and nobody gets 0%.
• It's a fair, neutral starting point. Good for understanding which channels show up in converting journeys at all.
• It's less extreme than single-touch models, so budget decisions based on linear data are less likely to swing wildly.
Cons and blind spots
• Equal credit is an assumption, not an insight. A casual accidental click and a high-intent product page visit get the same weight.
• It rewards frequency. Channels that touch the customer often (retargeting, email, WhatsApp) accumulate credit simply by showing up more, even if they didn't change the outcome.
• It is only as good as your tracking. If a touchpoint is not tracked, it doesn't get a smaller share. It gets nothing.
Linear feels fair. But being fair and being right are not the same thing.
The blind spot all three share
Go back to Priya's journey. The YouTube awareness ad and the Instagram Reel she watched without clicking got zero credit in every single model.
That's because first-touch, last-touch and linear are all click-based, rules-based models. They can only distribute credit among touchpoints they can see. They cannot see:
• TOF awareness and reach campaigns optimised for impressions or video views
• Reels, Stories and YouTube views without a click
• Influencer and creator content, unless a code or link is used
• Podcasts, word of mouth and WhatsApp forwards
For a brand investing in TOF campaigns and brand marketing, that's a big problem. The rules-based models will consistently tell you your TOF spend is worthless, while quietly handing its credit to retargeting, branded search and direct traffic.
And none of these models answer the most important question: would this sale have happened anyway? They are all measures of correlation, not causation.
Should you rely on what the platforms report?
Short answer: use it, but don't trust it on its own. Meta, Google and every other ad platform report conversions using their own attribution windows (Meta's default, for example, is 7-day click and 1-day view). This creates three problems.
1. Everyone claims the same sale
In Priya's journey, Meta claims the order because she clicked a Meta ad within 7 days. Google claims it because she clicked a search ad. Your email tool claims it because she clicked the email. One ₹2,000 order becomes ₹6,000 of reported revenue. This is why the sum of platform-reported conversions is almost always higher than your actual Shopify orders.
2. Platforms grade their own homework
Each platform only sees its own touchpoints and has every commercial incentive to look effective. It is not lying, exactly. But it is not neutral either.
3. View-through credit cuts both ways
Platform view-through attribution is the one place views do get counted. That's useful for measuring awareness, but it is also easy to inflate. A user who scrolled past your ad for a second and then bought because of an influencer post or a friend's recommendation still counts as a Meta conversion.
Platform numbers are best used for in-platform optimisation (which ad set, creative or audience), not for deciding how to split budget across channels.
A balanced approach: start with what you want to measure
Instead of asking "which model is right?", ask "what decision am I trying to make?" Different questions need different lenses.
Measuring performance and conversion efficiency
Example question: which retargeting campaigns, email flows or search campaigns are closing sales most efficiently?
→ Last-touch (or last non-direct click) makes sense here. You are evaluating the closers, so a model that focuses on the closing touchpoint is appropriate. Just don't use it to judge prospecting.
Measuring new customer acquisition
Example question: which channels are bringing new people into our funnel?
→ Use first-touch, filtered to first-time buyers only. Compare it against last-touch to see which channels open journeys and which close them.
Understanding the full journey
Example question: which channels work together, and where should we avoid cutting budget?
→ Use linear (or a position-based model) as a neutral baseline, and look at how often channels co-occur in converting paths.
Measuring TOF campaigns and brand marketing
Example question: are our Meta prospecting, YouTube awareness or influencer campaigns actually creating new demand?
→ Click-based models will fail you here. TOF campaigns work by being seen, not clicked. Instead, measure:
• View-through conversions for video and Reels campaigns, with sensible, short windows.
• Lift in branded search and direct traffic during and after the campaign flight.
• Geo or time-based lift tests. Run the campaign in some regions and not others, then compare total Shopify orders.
• Post-purchase surveys. Simply asking "How did you first hear about us?" captures YouTube, Reels, influencers and word of mouth that no pixel ever will.
• Media mix modelling (MMM), once you have enough spend history across channels.
Knowing what is truly incremental
Example question: if we cut this channel tomorrow, how much revenue would we actually lose?
→ No rules-based model can answer this. Run incrementality experiments, and use the results to calibrate how much you trust each channel's attributed numbers.
Putting it together: a practical stack for Shopify brands
1. Make Shopify orders your source of truth. Reconcile every channel's claims against real orders, net of cancellations and RTO, not platform-reported conversions.
2. Fix your tracking first. Server-side tracking and Conversion APIs recover a large share of the signal lost to browsers, ad blockers and iOS. A model is only as good as the touchpoints it can see.
3. Look at multiple models side by side. When first-touch and last-touch disagree sharply on a channel, that's an insight, not a contradiction.
4. Add a post-purchase survey. It's the cheapest way to see the channels your pixels can't.
5. Validate with experiments. Run a lift test at least once a quarter on your biggest channels.
6. Use platform data for platform decisions. Trust Meta's numbers to choose between Meta creatives. Don't use them to decide how to split budget between TOF and retargeting.
How BooleanMaths approaches this
At BooleanMaths, we are cynical of any single number, including our own. That's why our attribution combines first-party, server-side tracking of every Shopify order with multi-model views and post-purchase survey calibration, so channels that pixels miss (TOF video views, influencers, word of mouth) still get their share of credit. The goal is not a "perfect" model. It is a set of numbers you can actually trust when you move budget.
Final thoughts
• Last-touch tells you who closes. It over-credits retargeting, email and branded search.
• First-touch tells you who opens. It ignores nurture and often misses the true first touch.
• Linear tells you who participated. It assumes every touch is equal and rewards frequency.
• Platform reporting is great for in-platform optimisation and unreliable for cross-channel budget decisions.
• TOF and brand campaigns need view-through, lift tests, surveys and MMM, not click models.
Pick the model that matches the decision. Attribution tells you what happened. Incrementality tells you what mattered.
