Email capture and attribution: what a welcome flow adds

The flow report doesn't say, because it credits orders that follow an open or a click inside its window and keeps no group that went without the emails. By default Klaviyo credits an order to an email the buyer opened or clicked in the previous 5 days, and its documentation says a machine open from Apple Mail Privacy Protection counts unless you switch it off. Hold out a random share of new subscribers from the welcome flow and compare their orders in Shopify with everyone else's.

Why does the flow report show more revenue than the flow caused?

Three rules in the vendors' documentation decide what gets credited:

1. An open or a click, then a window. For new accounts, Klaviyo's default lookback is 5 days for email opens and 5 days for email clicks, on a last-touch model, and you can change it in settings (Klaviyo Help Center, 2026). An order inside the window goes to the flow whether or not the email changed anything.
2. Machines open email. Klaviyo's conversion-tracking page says conversions are tracked when Apple Mail Privacy Protection automatically opens the email, unless you exclude those opens. Apple's 2021 announcement says the feature stops senders from using invisible pixels to collect information about the user. Shopify says the same of its own pixel: email previews and security tools can load it when nobody opened the email.
3. The tools define credit differently. Shopify's email report uses the last non-direct click model by default, which needs a click. Klaviyo also credits opens. One flow, two revenue numbers.

Capture and flow are separate tests. A welcome flow starts when someone signs up, which can be while they are already weighing an order, so the popup changes who joins and the flow changes what they get next. The welcome popup post covers the popup and its discount. This one covers the emails.

Does a welcome flow add any orders?

Sometimes, and the published evidence is about triggered emails in general. In a peer-reviewed experiment at a multichannel retailer, about half of a sample of shoppers who had browsed and left were randomly chosen to receive automated emails and the other half got no message. The emails increased revenue in the online channel and in the triggered category (Goic, Rojas and Saavedra, Journal of Interactive Marketing, 2021). It wasn't about welcome flows, and the size of the effect is that retailer's, not yours.

Measuring it is harder than it sounds. Twenty-five large field experiments with major US retailers and brokerages, most reaching millions of customers, found that the median confidence interval on return on investment was over 100 percentage points wide (Lewis and Rao, Quarterly Journal of Economics, 2015). Those were advertising experiments, not email. The lesson carries: a small list gives a wide interval, so expect a direction and not a decimal.

How do you hold out a welcome flow?

Run it on your own list:

1. Add a split below the trigger and base it on a random sample. Klaviyo's flow documentation says the percentage you set goes down the first path, which it calls the control. Make that path the holdout and put the welcome emails on the other.
2. Tag both groups. Klaviyo's update profile property action writes a custom property to a profile, so each path can set a value such as holdout or flow and you can export the two groups.
3. Change nothing else. Campaigns and other flows go to both groups. Leave the popup and the offer alone for the whole test.
4. Choose the outcome first: orders and revenue per subscriber over a fixed window after signup, from Shopify orders matched on email, for both groups. Don't use Klaviyo's attributed revenue. The empty path sends no messages, so it shows zero by construction.
5. Size it before you start. The holdout test planner turns your daily orders, the smallest lift worth detecting and the share held out into a number of days. On a small list, plan a long test or accept a direction.

The sum is extra orders per subscriber times the subscribers in the flow group. For illustration: 0.11 orders per subscriber in the flow group against 0.09 in the holdout is 0.02 extra orders per subscriber, or 40 extra orders across 2,000 subscribers.

What the result means:

• Pass: orders per subscriber are higher in the flow group by more than chance allows at your list size, and the extra orders at full margin cover the discount and the sending.
• Fail: the two groups order at the same rate. The flow's attributed revenue was credit for orders that were coming anyway.
• Too small to read: the gap sits inside the noise. Run it longer, or add another signup period.

Sources, 30 September 2026: Understanding Klaviyo message attribution (Klaviyo Help Center, updated March 2026); Understanding message conversion tracking (Klaviyo Help Center, updated March 2026); How to A/B test flow branches (Klaviyo Help Center, updated February 2026); Flow Glossary (Klaviyo Help Center, 2026); Shopify Messaging email analytics (Shopify Help Center, 2026); Apple advances its privacy leadership with iOS 15 (Apple, 7 June 2021); The Effectiveness of Triggered Email Marketing in Addressing Browse Abandonments (Journal of Interactive Marketing, 2021); The Unfavorable Economics of Measuring the Returns to Advertising (Quarterly Journal of Economics, 2015). Vendor documentation describes settings, not effects; the two experiments are peer-reviewed.

Related answers

• Is your welcome popup discount incremental?
• Confirm a weak channel with a holdout test
• Five checks before you trust an attribution report
• Welcome email, defined
• Incrementality testing