The Ecomm Analyst

Growing stores, one honest take at a time.

Is Northbeam’s attribution accurate?

Northbeam’s attribution is accurate in the sense that matters most. It reconciles to your real orders. Its docs say it uses your orders as the ground truth for revenue, so it is not inventing sales the way ad platforms can when Meta and Google both claim the same purchase. Where accuracy gets slippery is the credit split. Northbeam offers several models and two accounting modes, and the same month can show noticeably different channel numbers depending on which combination you pick. The tool is consistent. The settings decide the answer.

What the default model does

Northbeam’s default is a model it calls Clicks-Only. According to its documentation, credit is divided equally among the touchpoints in a journey, excluding bottom-funnel ones like direct, branded search, email and SMS. That exclusion is a deliberate opinion. It assumes those channels harvest demand that something else created, so they get no credit in this view.

For a brand where email drives a lot of repeat revenue, that opinion will make email look weak. It is not a tracking error. It is a model choice, and you need to know it is there before you cut a Klaviyo budget based on it.

Modeled views are an estimate

Northbeam also offers Clicks + Modeled Views, which its docs describe as a proprietary machine learning model that uses impression and engagement data. It exists because social and video ads drive sales people never click through to. That is a real effect, and Clicks-Only undercounts it. The modeled number is still a model. Northbeam’s own docs frame it as directional, useful for ranking ads within a channel, and suggest using it alongside Clicks-Only rather than instead of it.

A third option, Clicks + Deterministic Views, uses platform-verified views instead of modeled ones. It is less of an estimate, but it depends on what each ad platform is willing to share.

Accounting mode changes the dates

Cash Snapshot credits revenue on the day of the order. Accrual Performance spreads credit back to the days the contributing touchpoints happened. Northbeam’s knowledge base notes that accrual revenue for a period can come out smaller than cash revenue for the same period, because some credit lands in earlier months. If your CFO compares a Northbeam accrual report to Shopify revenue, they will see a gap and assume the tool is wrong. It is reporting a different thing.

Where the inputs can go wrong

Any attribution tool is only as accurate as its install. On Shopify, Northbeam has a native app and a purchase script on the order status page. On other platforms, orders come in through its Server API and must carry customer and order IDs that match what the pixel sends. If those IDs do not match, orders arrive without journeys. Northbeam also asks you to use its own UTM parameters on major ad platforms. Inconsistent UTMs are the most common reason I see channel credit land in the wrong bucket, on Northbeam or anywhere else.

How to check it yourself

Northbeam suggests comparing its tracked visits against platform-reported clicks, a ratio it calls Visit-to-Click. If Northbeam and GA4 roughly agree on visits but both sit well below a platform’s click count, the platform is likely overcounting. I add two more checks. Compare total attributed orders in Cash Snapshot against your store’s order count for the same week. Then run one holdout, such as pausing a single channel in one region for two weeks, and see whether the drop in sales matches what the model predicted. A model that matches a holdout has earned trust. One that does not needs a closer look.

More on this in how accurate are e-commerce attribution tools and what I check before I trust a new tool’s numbers.

If you want simpler numbers

Northbeam’s flexibility is the point for teams with an analyst. For smaller teams it can mean three people reading three different ROAS numbers. ThoughtMetric is an e-commerce attribution tool for Shopify, Magento, WooCommerce, BigCommerce and custom stores, with click-based multi-touch models, server-side tracking and post-purchase surveys as a second signal on every plan. (Disclosure: ThoughtMetric sponsors this site.) It does not model view-through conversions, so it will not credit unclicked video the way Northbeam’s modeled views do. Pricing starts at $99 a month based on pageviews. For Northbeam’s costs, see how much Northbeam costs.

How to read Northbeam well

  • Reconciling to your books? Use Cash Snapshot and compare against store orders.
  • Judging email or branded search? Remember Clicks-Only excludes them by design.
  • Judging video or social? Look at Clicks + Modeled Views next to Clicks-Only, not alone.
  • Numbers look off? Check UTMs and order ID matching before blaming the model.
  • Want fewer settings and a second signal from customers? Look at ThoughtMetric (our sponsor).

One response to “Is Northbeam’s attribution accurate?”

  1. Northbeam vs Polar Analytics – The Ecomm Analyst Avatar

    […] Northbeam defaults to its Clicks-Only model, which its docs say divides credit equally across touchpoints while excluding bottom-funnel ones like direct, branded search, email and SMS. It also offers Clicks + Modeled Views, a machine learning model that credits impressions, and Clicks + Deterministic Views, which uses platform-verified views. Two accounting modes, Cash Snapshot and Accrual Performance, change which day revenue lands on. The flexibility is real, and so is the learning curve. I covered how those settings change the numbers in is Northbeam’s attribution accurate. […]

    Like

Leave a Reply

Navigation

About

Six years in e-commerce. Three Shopify stores across different niches, one scaled past seven figures. I’ve tested hundreds of ad creatives, obsessed over email flows, and learned more from my failures than my wins.

Now I focus on conversion optimization, retention marketing, and the analytics behind it all. This blog is where I share what actually works, backed by real numbers. No fluff, no guru energy.