Attribution for shopping apps

Follow one shopper to see what needs measuring

Take a hypothetical buyer and the path she takes over a month.

1. She sees an ad for a jacket on her phone and taps it. She does not have the app, so she lands in the app store.
2. She installs, opens the app and sees the jacket's product page.
3. She adds it to her cart and leaves.
4. Three days later an email reminds her about the cart. She taps the link and the app opens on the cart.
5. She buys. Two weeks later a sale ad brings her back and she buys again.

Each step needs something different from your attribution setup. The sections below take them in turn.

Steps 1 and 2: the link has to land on the product

A shopping ad promises a specific item. If a new user installs and lands on the home screen, she has to search for the jacket again, and many people will not. A deferred deep link carries the product through the install so the first screen is the right one.

The same applies to people who start on your website. Shoppers often browse on the mobile web and are then invited into the app by a banner or a button. Measuring that handover is covered in how web-to-app attribution works. Airbridge Core includes deferred deep links, QR codes and web-to-app attribution in the plan.^[1]

Steps 3 and 5: track the funnel with values

A shopping app needs a short, fixed set of events:

• Product view, with the product ID.
• Add to cart.
• Checkout start.
• Purchase, with order value, currency and an order ID.

Each item is an in-app event. These are commonly offered as standard events. Check your tool's list. The order ID matters more than it looks: it lets you remove duplicates and match orders against your own sales records. Decide early whether the value you send includes tax, shipping and discounts, and keep it the same everywhere. Returns and cancelled orders will make your own books lower than the attributed total, so expect a gap and know its usual size.

With values in place you can read ROAS per campaign. Compare it with caution across campaign types, because a campaign aimed at past buyers will usually look better than one aimed at strangers.

Steps 4 and 5: credit the return visit

In an established shopping app, many orders come from people who installed long ago. Email, push and retargeting ads bring them back. If your tool only looks at installs, all of that work is invisible. Re-engagement attribution credits a purchase to the link or ad that brought an existing user back.

Two readings come out of this. First-order measurement shows what it costs to win a customer. Repeat-order measurement shows what a customer is worth over time, which is LTV. A campaign with an expensive first order can still be your best one if its buyers return.

What to do next

Check your current setup against the shopper's five steps. For each step, ask whether you could say today which ad or message caused it. Fix the earliest gap first: links that land on the wrong screen cost orders, while a missing report only costs insight. Then confirm that purchase values in the tool match a sample of real orders before you trust any revenue figure.