Omnisend With Causality Engine: Check the Attributed Sales

The short answer: when Omnisend's attributed sales and your ad platforms' claims add up to more than the store took, the reports are doing what their rules say. Omnisend, the email, SMS and push platform built for online stores (entry), credits every order placed within 7 days of an email open or click, and 24 hours for SMS and push, per How Omnisend Attributes Sales. If the store also records purchases in GA4, the other side of the comparison is already there, because Omnisend tags its links without being asked. Export the Attribution paths report and put it on the start page of Causality Engine, which estimates what each channel caused from that file (entry). The first finding costs nothing.

Disclosure: I founded Causality Engine and own it, and this page recommends it. Every claim about it links to Causality Engine's own pages.

Before you believe the sales column:

• Do the calendar sum. For example, one campaign a week with a 7-day window means 52 x 7 = 364 days a year inside a window for any contact who opens each send. For that contact, nearly every order is an Omnisend order.
• Use the tags you already send. Every Omnisend link carries utm_source=omnisend and a medium, so GA4 has been filing those visits all along (UTM tags article).
• Read three settings. Under Store settings, Sales attribution: the window length, when the window starts, and whether Exclude Apple MPP opens is on (attribution article).

The facts in one table

What Omnisend writes on every link

Nobody switches tagging on. The UTM tags article says tags go on all links in Email Campaigns, SMS Campaigns, Push Notifications and Automation Workflows. They can be customised, not removed. The defaults:

• utm_source is always omnisend.
• utm_medium is email, sms or PushNotification.
• utm_campaign is "campaign:" plus the campaign name and ID, or "automation:" plus the workflow name and ID.
• utm_content is the message ID, on workflow messages only.

Those mediums land where they should. Google's default channel group files medium email under Email, medium sms under SMS, and any medium containing "notification" under Mobile Push Notifications, which is where Omnisend's web push ends up. The definitions are not case sensitive, and a value that matches no rule is filed as Unassigned.

So the risk sits in the edit screen. Edit UTM tags lives at the Settings step of a campaign and on each workflow message. Change the medium to "newsletter", say, and the Email rule no longer matches. Omnisend's own advice: lowercase only, and short, clear names.

What counts as an attributed sale

The attribution article is exact. Receiving a message earns nothing: the customer must open or click an email or push, or click an SMS. Then:

• Every order in the window counts. Two orders after one click are both credited.
• The last engaged message wins among Omnisend's own messages.
• The window starts at the send or at the engagement, your choice. The second gives each message longer.
• Revenue is filed on the day the message was sent, not the day of the order.
• Refunded and cancelled orders stay in.
• All synced orders qualify, POS and physical store sales included, with no built-in exclusion.
• Apple MPP opens count by default. The switch is Exclude Apple MPP opens, under Tracking.

The window is changed under Store settings, Sales attribution, where the store settings article gives 3, 7, 14 and 30 days as examples. Export your reports first: a change recalculates every historical report.

Omnisend says of its gap with Shopify that both numbers are accurate and measure differently. Fair. Trouble starts when one is read as money the emails made.

The claim gap: a worked example

For example, with figures chosen for illustration and no store behind them: October net sales of €212,400. Omnisend attributes €71,300 to its emails. Meta's dashboard claims €88,950 and Google Ads €79,600. GA4's last click gives Email €23,850.

71,300 + 88,950 + 79,600 = €239,850 claimed, against €212,400 booked. That is 239,850 - 212,400 = €27,450 more than the store took, with organic search and direct still unheard. The email gap is 71,300 - 23,850 = €47,450: orders that followed an open, or a click on another device, and reached the store by a different route.

A causal read does not referee those two email figures. Per its how-it-works page, it sets each channel's caused share next to last-click from the same export, grades it with a data-health score from 0 to 100, and attaches a next step with a one-region test. If email's caused share sits far from Omnisend's claim, test on a held-out group before any budget moves.

What the read cannot do for an Omnisend store

Nothing links the two products: no app, no API key, no sync. Causality Engine's integrations page says there is no connector to wait for, because it reads the channels your GA4 export records.

That has consequences. A sale at the till does not pass through the GA4 tag on your site, so the POS orders Omnisend counts are missing from the export unless you send them to GA4 yourself. The read sees email as a channel, not as separate workflows; which message to rewrite stays an Omnisend question. It needs a purchase key event and 12 months or more of history (start page). And the homepage says that where spend is too small for the model to find a signal, it says so and declines.

Two of Omnisend's defaults, MPP opens counted and refunds left in, raise the figure that sits beside its invoice, and a larger attributed total is the strongest argument a subscription has at renewal. Switching the first off falls to you, and the article lists no setting for the second.

A split test first, the export second

The free check uses a method Omnisend documents for another purpose. Create A/B Test Segments for Campaigns splits contacts at random with an A/B Test Split block that tags them variant_a or variant_b, on all plans, and recommends 2,000 contacts or more. The article sends a different version to each group. Send to one group only and you have a holdout, by hand.

For example, 6,000 subscribed contacts split evenly: 3,000 receive the campaign, 3,000 do not. Count orders per group in the store's export for 7 days. If 174 come from the first group and 150 from the second, the campaign caused about 174 - 150 = 24 orders. That answer stands on its own.

For every channel at once, Causality Engine, which I founded, takes the GA4 export on the start page. The preview is free and worked out in your browser. The full read is €99 once, with a refund inside 30 days (pricing, checked on 11 November 2026).

See also

• Omnisend: what it does, costs and gets criticised for
• How to use Omnisend: cart recovery on Shopify, nine steps
• Omnisend on the marketplace: Omnisend's affiliate program terms
• Causality Engine: the product, its prices and limits
• How to use Causality Engine: from export to a regional test
• Causality Engine on the marketplace: the affiliate terms, stated as the owner's
• Omnisend vs Klaviyo: the lower bill or the deeper data
• Email marketing ROI calculator: return on a send, your inputs

FAQ

Is there an Omnisend app or sync for Causality Engine?

No. Causality Engine works from a file downloaded from GA4, where Omnisend appears as the Email, SMS and push visits its UTM tags produced (integrations page).

Why is Omnisend's revenue higher than the Email row in GA4?

Omnisend counts an order after an open and across devices, inside a 7-day default window. GA4 needs a tagged click it can tie to the purchase. The UTM tags article covers this in its FAQ.

Should I exclude Apple MPP opens?

My recommendation: compare both ways. Export your reports, turn on Exclude Apple MPP opens under Store settings, Sales attribution, Tracking, and note the change; Omnisend recalculates attributed sales from April 3, 2021 (attribution article).