E-Commerce Analytics: A DTC Operator's Guide to the Numbers That Matter

Most DTC brands don’t have a data problem. They have a reconciliation problem. The numbers exist — they’re just scattered across Shopify, GA4, Meta, Klaviyo, Triple Whale, and a warehouse export, and none of them agree. Your CAC reads one way in the ad platform, another in your attribution tool, and a third on the bank statement. So you buy another dashboard. Now you have eleven dashboards and still can’t answer the only question that matters: where should we put the next dollar?

This guide is the map I wish more founders had before they hired their third analyst. It covers what e-commerce analytics actually is, the five layers of the metric stack that run a brand, why your reports never line up, and how to build something your team will actually use on a Monday morning.

What e-commerce analytics actually is (and isn’t)

E-commerce analytics is not dashboards. Dashboards are the output. Analytics is the work of turning scattered, contradictory data into a small number of decisions you can defend.

A useful analytics function answers three questions, in order:

What happened? (descriptive — most teams stop here)
Why did it happen? (diagnostic — where the money actually is)
What should we do about it? (prescriptive — the only part that pays)

If your reporting tells you revenue was down 8% last week but can’t tell you which channel, cohort, or product drove it — and what to do — you don’t have analytics. You have a very expensive rear-view mirror.

The five layers of the DTC metric stack

Every e-commerce brand runs on the same five layers. Most founders obsess over the first and ignore the two that decide whether the business actually works.

1. Acquisition — what it costs to get a customer

This is CAC (customer acquisition cost) and the channels feeding it. The trap here isn’t measuring CAC — it’s attributing it. Last-click says one thing, the ad platform’s in-app number says another, and both are wrong. Acquisition analytics is really an attribution problem, which is why it deserves its own deep dive on why your channels never add up.

To compare channels honestly on incremental contribution rather than platform-reported ROAS, run them through the Channel Scorecard.

2. Conversion — what happens on-site

Sessions, add-to-carts, checkout starts, completed orders. This is your funnel, and it’s where most brands have the fastest wins because the traffic is already paid for. A one-point lift in checkout completion drops straight to the bottom line. The discipline here is experimentation — sizing tests properly and reading them without fooling yourself. If your conversion rate is the constraint, score the opportunities first with the CRO Opportunity Scorer.

3. Retention — what happens after the first order

This is where most DTC brands quietly die. If you acquire customers who never come back, every channel looks unprofitable, because the whole model assumes a second purchase. The core metric is your repeat / retention rate, and the right way to read it is by cohort, not in aggregate — a blended number hides the trend. Start with how to calculate customer retention rate (CRR) on Shopify data, then graduate to cohort analysis to see when customers leak. Benchmark your repeat rate against your category with the Repeat Purchase Scorer.

4. Unit economics — whether any of it makes money

The two layers above collide here. LTV:CAC is the ratio that tells you whether the business is fundamentally sound or just buying revenue at a loss. Get this number wrong — and most brands do, usually by inflating LTV with revenue instead of margin — and you’ll scale yourself into insolvency. This is important enough that I wrote a full guide on how to calculate the LTV:CAC ratio and what “good” looks like. Model a customer’s lifetime value with the LTV Calculator, and pressure-test the basket size you’re planning around with the AOV Calculator.

5. Inventory — whether you can fulfill it profitably

The layer founders forget until cash is frozen in dead stock or they stock out of the hero SKU mid-campaign. Inventory analytics is about ordering to real demand and understanding which products subsidize the catalog and which drag it down. See how to calculate Economic Order Quantity in Python for the ordering math, and use the Product Performance tool to separate the stars from the dogs.

Why your dashboards never agree

Here is the single most common reason DTC reporting is a mess: every platform is incentivized to take credit.

Meta counts a sale if the customer saw an ad in the last 7 days. So does TikTok. So does Google. So does your email tool. Add up the platform-reported revenue and you’ll often “explain” 140% of your actual sales. The platforms aren’t lying exactly — they’re each answering “did I touch this order?” with a yes, and you’re reading it as “did I cause this order?”

The fix is not another dashboard. It’s:

One source of truth for revenue (almost always Shopify / your order system, reconciled to the bank).
A single attribution model you’ve decided to trust — ideally validated against holdout tests, not last-click theatre.
Incrementality thinking — judging channels on the orders that wouldn’t have happened otherwise.

This reconciliation work is unglamorous and it is the entire game. I walk through a real example in the DTC attribution rebuild case study, where six platforms telling six different stories about CAC turned into $1.4M of revenue that had been written off as “channel overlap.”

The metrics that actually drive decisions

You do not need 200 KPIs. A brand making good weekly decisions usually watches a handful:

Contribution margin after ad spend — the real profit number, not gross revenue.
Blended CAC vs. new-customer contribution margin — are you acquiring profitably?
LTV:CAC by cohort — is it improving or decaying?
Repeat purchase rate / time-to-second-order — is the retention engine working?
Incremental ROAS by channel — where the next dollar should go.

Everything else is diagnostic detail you pull when one of these moves. The art is choosing the five numbers your team looks at every Monday — which is exactly what a metric tree helps you map from a single North Star down to the levers you can actually pull.

How to build an analytics system that gets used

A report nobody opens is worth zero, no matter how correct it is. The systems that survive contact with a busy founder share four traits:

One screen, five numbers. If the executive view doesn’t fit on a screen, it won’t get read.
A decision cadence. Numbers attached to a recurring Monday meeting with an owner, not a dashboard that exists in a vacuum.
Reconciled inputs. The screen agrees with the bank. Trust is the whole product; one wrong number and the team goes back to gut feel.
Documented and owned. Your team can run it without the consultant who built it.

That last point matters more than the modeling. The goal isn’t a clever model — it’s a brand that can make a confident revenue decision every week without you in the room.

Where to go deeper

This guide is the map; here are the deep dives:

When you want the numbers to finally agree

If you’re staring at eleven dashboards that don’t reconcile and a team making revenue calls on numbers they don’t trust, that’s the exact problem I solve. I stitch Shopify, GA4, Triple Whale, Klaviyo, ad platforms, and the warehouse into one screen — then tell you where the money is and what to do about it.

See how the e-commerce analytics engagement works, or book a call and we’ll find the leak.