Ecommerce Attribution: How to Understand Which Ads Drive Sales

When an online store makes a sale, more than one marketing channel usually had a hand in it. A shopper sees a Meta ad on their phone, searches for your brand on Google a few days later, then comes back through an email and buys. Each of those touches tells part of the story, and each platform would like to take the credit.

Ecommerce attribution is the work of making sense of that mess. This guide explains what it is, why the numbers in your dashboards never agree with your store, how the main models differ, and how to use attribution to make better budget decisions without pretending it is more precise than it is.

What is ecommerce attribution?

Ecommerce attribution is the process of giving credit for an order to the marketing touchpoints that came before it. An attribution model is simply the rule for how that credit is shared out.

A few plain examples:

• A shopper clicks a Google search ad and buys within the hour. Almost every model gives Google the credit.
• A shopper sees a TikTok video, ignores it, clicks a Meta ad two days later, and buys after a Google brand search. Now three channels have a claim, and the model you choose decides who gets the credit.

Attribution is a tool for comparing patterns and informing decisions. It is not a perfect record of what caused a purchase. The answer depends on the model, the tracking setup, the time window, and the data you can actually see.

Why your ad dashboards and your store never match

If you have ever added up the purchases reported by Meta, Google and TikTok and gotten a bigger number than your Shopify orders, you are not doing it wrong. This is normal, and it happens for a few reasons.

Every platform grades its own homework. Each ad platform reports from its own measurement system. If a customer interacted with two platforms before buying, both may report that order as theirs.

The time windows differ. Meta lets advertisers choose attribution settings such as a 1-day or 7-day click window and a 1-day view window, and these are set at the ad set level. Google Ads lets you set a conversion window for each conversion action. If two platforms look back different amounts of time, they will count different orders. You can read how each platform defines this in Meta's attribution settings help page and Google's conversion windows help page.

Tracking has gaps. Browser privacy settings, blocked cookies and people switching devices all leave holes in the data. Platforms fill some of those holes with modeled estimates.

Views count too. Some platforms count a purchase when someone merely saw an ad and later bought, without clicking. That is a real signal, but it is a different kind of claim from a click.

The result is that the sum of platform-reported sales often exceeds the orders your store actually recorded. Your store is the better source for how many orders and how much revenue really happened. What your store cannot tell you by itself is which marketing touchpoints influenced those orders.

The main attribution models, in plain language

Models available vary by tool. Google Ads, for example, currently supports data-driven and last click attribution; you can check the current list on Google's attribution models page. Google Analytics has its own attribution settings, explained in Google's guide to attribution in GA4. Shopify also reports marketing performance with its own attribution approach, described in Shopify's marketing performance help page.

There is no single correct model. A store with a short, impulse-driven buying cycle can get plenty out of last click. A store selling higher-priced items, where people research for weeks, will usually want a view that gives earlier touches some credit.

A worked example: one order, three stories

Say a customer buys a $120 jacket. Here is the journey:

1. Monday: sees a TikTok video, does not click.
2. Tuesday: clicks a Meta ad, browses, leaves.
3. Friday: searches your brand name on Google, clicks the ad, buys.

This is how each lens describes that single $120 order:

• Last click says Google earned $120, and Meta and TikTok earned nothing.
• First click says TikTok earned $120.
• Linear gives each channel $40.
• Meta's own report, depending on its window and settings, may claim the whole $120.
• Google's own report also claims the $120.

Your store recorded exactly one order and $120. The platforms together may report $240 or more. None of them is lying. They are answering different questions. This is exactly why you should anchor on what the store recorded.

A practical way to read attribution data

1. Pick the outcome you care about. Orders, paid revenue, new customers or something else. Use the same definition every time.
2. Use your store and payment records as the source of truth for transactions. That tells you what really happened.
3. Read each platform report on its own terms. Note its window and conversion definition before comparing channels.
4. Compare consistent time periods. One day is noise. A few weeks is closer to a pattern.
5. Look at the total. If you moved budget between channels, did overall revenue go up, stay flat or fall? Channel reports alone cannot answer that. A blended ROAS check against total revenue and total spend is a good sanity check.
6. Run a deliberate test when it matters. If you want to know whether a channel is truly adding sales, pause or reduce it for a set period in a controlled way and watch the total. This is the closest thing to evidence of cause that most small stores can get.

Attribution and budget decisions

Attribution can show how channels show up in customer journeys, but moving budget still involves uncertainty. A channel's attributed revenue is one signal. Others include total store revenue, total ad spend, margins, the share of new customers, and what actually happened after you last changed a budget.

Some habits that help:

• Do not take a big chunk of budget out of a channel after one bad week.
• Write down what you changed and when, so you can read the result later.
• Treat a channel that looks weak on last click but strong on first click as a possible discovery channel, not an automatic cut.
• Keep a small test budget for new creative and audiences.

If you are deciding between two big channels, our guide to Meta Ads vs Google Ads for ecommerce walks through a fair way to compare them.

Common mistakes to avoid

• Adding up platform-reported sales and calling it total sales. It double counts.
• Switching models to find the one that makes a favorite channel look best. Pick a method, then stick with it long enough to learn from it.
• Ignoring margin. A channel can look great on revenue and still lose money after product cost, shipping and fees.
• Treating view-through credit like click credit. They are different kinds of evidence.
• Overreacting to short windows. Recent days are often incomplete because late conversions have not arrived yet.

How Mindra approaches attribution

Mindra's approach is built around that gap between what ad platforms report and what actually reached the store. It connects ad spend from platforms such as Google, Meta and TikTok to revenue recorded in Shopify, Stripe and your attribution tool, then recommends where to move budget, and nothing changes until you approve it. You can see how it works on mindra.co.

No attribution setup will remove every bit of uncertainty from marketing measurement. A clear model, consistent definitions and a habit of checking against recorded business revenue will make your data more useful and your budget conversations a lot calmer.

Frequently asked questions

What is ecommerce attribution in simple terms? It is the method for deciding which marketing channels get credit for an order. Because one customer may see several ads before buying, attribution models are rules for sharing that credit.

Why do my ad platforms report more sales than my store? Each platform counts conversions in its own system, with its own windows and rules. When a customer touched more than one platform, more than one platform may claim the same order.

Which attribution model is best for ecommerce? There is no universal best. Short buying cycles often do fine with last click. Longer research cycles usually benefit from a model that also credits earlier touches. What matters most is using one approach consistently and checking it against store revenue.

Is last click attribution wrong? It is not wrong, it is limited. It is simple and easy to audit, but it tends to undervalue channels that introduce people to your brand.

Can attribution tell me which channel caused a sale? Not with certainty. It shows how credit is distributed under a model. To learn whether a channel adds sales, a controlled test against total revenue is stronger evidence.

How do I compare channels fairly? Use the same date range, the same outcome definition, include all spend, and check the results against your store's recorded revenue and your margins.