One Shopify order. Four ad touchpoints. Depending on which attribution model your analytics tool uses, Google Shopping gets 100% of the credit — or just 25%. That difference determines whether a campaign looks like a 4x ROAS winner or a break-even disappointment. Attribution models are not a reporting technicality. They are the lens through which your entire marketing budget gets evaluated.
Why attribution models matter more than most brands realise
Here is a real-world scenario. A customer sees a Meta prospecting ad on Monday, clicks nothing. On Wednesday they see a YouTube pre-roll. On Friday they click a Google Shopping ad. On Saturday they search your brand name, click a branded search ad, and buy.
That single order will be attributed differently depending on your model:
- Last-click: Branded search gets 100% of the revenue. Meta, YouTube, and Google Shopping get zero.
- Linear: Each of the four touchpoints gets 25%.
- Time decay: Branded search gets roughly 40%, Google Shopping gets 30%, YouTube gets 20%, Meta prospecting gets 10%.
If you are running last-click attribution, your prospecting campaigns will always look unprofitable — because by definition they touch customers early in the funnel, before the converting click. You will cut them, dry up your top-of-funnel, and wonder six months later why your branded search volume is declining.
Choosing an attribution model is not about finding the most flattering number. It is about building a model of how your customers actually make purchase decisions, so your budget allocation reflects reality.
Last-click attribution
Last-click gives 100% of the conversion credit to the final touchpoint before purchase. It is the oldest and still the most widely used model — partly because it is simple, and partly because it is the default in many platforms including Google Ads (historically) and most basic e-commerce reporting tools.
When it makes sense: Last-click is reasonable if your purchase cycle is very short (under 24 hours) and your customers genuinely do discover, evaluate, and buy in a single session. Impulse-purchase categories — consumables, low-ticket items, time-sensitive promotions — can fit this pattern.
What it over-credits: Bottom-funnel channels: branded search, Google Shopping, retargeting. These channels intercept customers who have already been warmed up by earlier touchpoints. Under last-click, they look like the reason customers buy. In reality, they are often the last step in a journey that started elsewhere.
What it under-credits: Prospecting campaigns on Meta, TikTok, YouTube, and display. Any channel that introduces customers to your brand will be systematically undervalued. If you set budgets using last-click ROAS, you will perennially under-invest in top-of-funnel and over-invest in bottom-funnel — until your bottom-funnel starts drying up.
Brands that use last-click attribution often conclude their Meta prospecting "doesn't work" and cut it. Three months later, branded search volume drops and they cannot figure out why. Top-of-funnel spend creates the audience that bottom-funnel converts. Last-click makes this invisible.
Linear attribution
Linear attribution divides conversion credit equally across all touchpoints in the path. If there were four touchpoints, each gets 25%.
What it does well: It forces you to value every part of the customer journey. Prospecting campaigns get meaningful credit. No single channel monopolises the revenue.
Its limitation: Not all touchpoints are equal. A display impression that a customer barely registered is not equivalent to a product page visit driven by a targeted search ad. Linear treats them identically, which can overvalue passive touchpoints (view-through conversions, broad display) and distort budget decisions in the other direction.
Who it suits: Brands with longer purchase cycles (7–30 days), multiple active channels, and limited data volume for more sophisticated models. It is a significant improvement over last-click for most Shopify brands spending across Meta, Google, and email.
Time decay attribution
Time decay gives more credit to touchpoints that occurred closer to the conversion, with credit decaying exponentially the further back in time a touchpoint occurred. The half-life is typically set at 7 days, meaning a touchpoint 7 days before purchase gets half the credit of a touchpoint the day before.
The logic: The most recent interaction had the most influence on the purchase decision. A customer who clicked a Google Shopping ad five minutes before buying was probably more influenced by that ad than by the Meta prospecting ad they saw three weeks earlier.
When it works: Time decay is well-suited to shorter purchase cycles (3–14 days) where recency genuinely does correlate with influence. Fashion, beauty, and home goods brands often fit here. It is also useful when you run a lot of promotional campaigns — the ad that coincides with a sale arguably deserves more credit than a brand awareness campaign that ran a month before.
Where it falls short: For high-consideration purchases (furniture, B2B, expensive electronics), customers may research for 30–90 days. A time decay model heavily weights the final few touchpoints and writes off the early research-phase content that was actually critical to the decision.
Data-driven and position-based attribution
Data-driven attribution uses machine learning to assign credit based on the actual observed paths of converting vs non-converting customers. It answers the question: which touchpoints, when present in a path, actually increase the probability of conversion? Google Analytics 4 offers this as its default model when you have sufficient data volume.
The catch is volume. Data-driven models need thousands of conversions per month to produce statistically reliable weights. If you are converting 50 orders a month, the model does not have enough signal and the outputs are not meaningfully better than a simple rule-based model.
Position-based (also called U-shaped) attribution splits 40% of credit to the first touchpoint, 40% to the last, and distributes the remaining 20% across middle touchpoints. It reflects a common intuition: the first touch (brand discovery) and the last touch (conversion decision) both matter most.
"The best attribution model is not the most sophisticated one — it is the one you apply consistently across every channel, so your blended ROAS actually means something."
Comparison: what each model credits and penalises
| Model | Over-credits | Under-credits | Best for |
|---|---|---|---|
| Last-click | Brand search, retargeting, Shopping | Prospecting, upper funnel | Very short purchase cycles (<24h), single-session buyers |
| Linear | Passive impressions, view-throughs | Nothing systematically — but lacks nuance | Multi-channel brands, 7–30 day cycles, limited data volume |
| Time decay | Late-funnel touchpoints near purchase | Early research phase, awareness content | Short–medium cycles (3–14 days), promotional-driven stores |
| Position-based | First and last touch only | Middle-funnel nurture | Stores where discovery and closing moments are clearly defined |
| Data-driven | Nothing, when done well | Nothing, when done well | High-volume stores (1,000+ conversions/month) |
How to choose: a simple decision framework
Start with two questions:
- How long is your purchase cycle? Look at your Shopify data: what is the median time between a customer's first site visit (from any channel) and their first purchase? If it is under 24 hours, last-click is defensible. If it is 3–14 days, time decay is a reasonable default. If it is 14 days or longer, linear or position-based will give you a more honest picture.
- How many channels are you running? If you are only running Google Shopping and nothing else, attribution is almost moot — there is only one touchpoint. But if you are spending across Meta, Google, TikTok, email, and influencers simultaneously, the model you choose will dramatically change which channels look profitable. In multi-channel setups, last-click is especially dangerous.
A practical starting point for most Shopify brands spending $5K–$50K/month across 2–4 channels: use linear or time decay (with a 7-day half-life) as your primary model, and treat it as the lens for budget decisions. Use last-click data as a secondary signal to understand closing behaviour, not to set budgets.
The critical caveat: cross-channel consistency
Here is where most brands go wrong even after they have thought carefully about attribution models. Meta reports ROAS using its own attribution window (default: 7-day click, 1-day view). Google Ads uses its own model. GA4 uses its own. Shopify uses last-click by default. If you are looking at ROAS by channel in each platform's own dashboard, you are looking at four different attribution models simultaneously — and when you add them up, the total credited revenue will exceed your actual Shopify revenue by 2–4x.
This is not fraud. It is just that every platform counts the same order and gives itself full or partial credit. The result is a blended ROAS that is meaningless because it has no consistent denominator.
Cross-channel consistency means: pick one attribution model, apply it to all channels using the same data source, and evaluate every channel against the same standard. That data source should be your Shopify orders — the only record of revenue that is unambiguous.
Your blended ROAS is only meaningful if it is calculated from a single source (Shopify) using a single attribution model applied consistently across all channels. Platform-reported ROAS figures cannot be combined — they use different models, different windows, and different conversion definitions.
How DataMaster handles attribution consistently
DataMaster pulls your Shopify order data as the single source of revenue truth, then matches that revenue back to ad spend across Meta, Google, and other channels using a consistent attribution model that you configure once. Every channel's ROAS is calculated against the same denominator — actual Shopify revenue — using the same model and the same attribution windows.
This means when you compare your Meta ROAS to your Google ROAS in DataMaster, you are actually comparing like with like. You are not comparing Meta's self-reported 7-day-click ROAS against Google's last-click conversion value. You are seeing what each channel actually contributed to your Shopify revenue, under a consistent framework.
You can switch between attribution models in DataMaster's settings and immediately see how your channel mix ROAS changes — which is the fastest way to understand how dependent your current "winning" channels are on the model you happen to be using.
See your real ROAS across all channels with one consistent model
DataMaster connects your Shopify store, Meta, and Google Ads to give you a single attribution-consistent view of where your ad spend is actually going. No more comparing platform dashboards that can't be added together.
Start free trial — no credit card required