Where Margin Quietly Leaks in E-commerce & DTC: A Unit Economics Teardown
Direct answer: In most e-commerce and DTC businesses, margin leaks are hidden inside blended averages — a healthy-looking gross margin masks unprofitable SKUs, rising CAC, discount stacking, return costs, and shipping subsidies. The fastest way to find the leak is to rebuild your economics at the unit level: profit per order, per SKU, and per customer cohort, fully loaded with variable costs. When you do that, the "quiet" leaks become loud.
Most DTC founders track revenue, blended gross margin, and blended CAC. Those three numbers can all move in the right direction while your actual per-order contribution is bleeding out. Unit Economics is the lens that catches it.
The problem with blended numbers
A blended gross margin of 62% tells you almost nothing about whether you're making money. It's an average that hides its own worst-performing pieces. The same is true for blended CAC and blended AOV.
Margin in e-commerce leaks through a handful of predictable channels, and every one of them is invisible at the aggregate level:
- SKU mix drift — your growth SKUs may carry lower margins than your legacy hero product, so revenue grows while contribution shrinks.
- Shipping and fulfillment subsidies — "free shipping" is a real cost you're absorbing, often unevenly across order sizes and geographies.
- Discount and promo stacking — welcome codes, cart discounts, loyalty points, and clearance can compound on the same order.
- Returns and refunds — especially in apparel, the return rate and reverse-logistics cost can quietly erase category profit.
- Rising CAC on paid channels — as you scale spend, marginal CAC climbs above blended CAC, and the next customer costs far more than the average one.
- Payment processing, packaging, and pick-pack fees — small per-order costs that only matter when you multiply by volume.
None of these show up until you rebuild the math per unit.
The Unit Economics walkthrough for DTC
Unit Economics forces you to answer one question: does a single transaction make money after every variable cost? Here's the sequence to run.
Step 1 — Define your unit. For DTC, run this at two levels: per order and per customer (cohort). Both matter. An order can be profitable while the customer relationship is not (or vice versa, once you factor repeat purchase).
Step 2 — Build fully-loaded contribution margin per order. Start from net revenue (gross revenue minus discounts and returns) and subtract every variable cost:
- Landed COGS for the actual SKUs in the order
- Payment processing fees
- Shipping cost (yours, not the customer's)
- Pick, pack, and packaging materials
- Return probability × return handling cost for that category
- Marketplace or platform fees if applicable
What's left is contribution margin per order. Ask: how many of my orders are actually contribution-positive? You will almost certainly find a tail of orders — small baskets with free shipping, or heavily discounted orders — that lose money.
Step 3 — Layer in CAC to get customer-level economics. For each acquisition cohort, compare:
- LTV = contribution margin per order × expected repeat purchases over your horizon
- CAC = fully-loaded acquisition cost (ad spend + creative + agency + platform fees)
The benchmark most DTC operators anchor to is an LTV:CAC of roughly 3:1, with a CAC payback period under 12 months (tighter if you're cash-constrained). But treat these as directional, not gospel — the right ratio depends on your margin structure and cash cycle.
Step 4 — Segment relentlessly. Cut contribution margin by SKU, by channel (paid vs. organic vs. email), by discount code, by geography, and by first-order vs. repeat. The leak lives in a segment, not in the average.
What "good" looks like: positive contribution margin on the large majority of orders, a clearly profitable first-order or fast payback, CAC that doesn't balloon faster than volume, and a return rate priced into every category's margin. If your best cohort subsidizes a large loss-making cohort, you don't have a margin problem — you have a mix problem, which is fixable.
Where Percision fits — and where it doesn't
Full disclosure: I work on content for Percision, so here's the honest version.
When a spreadsheet is enough: If you have clean order-line data and one analyst who can build a contribution-margin model, do that. A well-structured Google Sheet or a Looker/Metabase dashboard pulled from Shopify + your ad platforms + your 3PL will surface most leaks. You do not need software to divide revenue by cost.
When a fractional CFO or consultant is the right call: If the leak is operational — renegotiating 3PL rates, restructuring your promo calendar, or fixing a returns process — that's hands-on work software won't do for you.
Where Percision helps: When you want consulting-grade analysis and a board-ready narrative faster than an 8–12 week engagement. Percision runs your business context through structured reasoning steps across 27+ frameworks — Unit Economics among them — to produce contribution-margin scenarios, financial ratios, warning signs, and an Excel-exportable model with an audit trail, plus a board deck. It's positioned as a co-pilot, not an autopilot: your team supplies the data and stays in control of the decisions. It's most useful when you're preparing for a board meeting, a raise, or a planning cycle and need depth and speed together.
Independent research supports the general pattern that AI tools improve output quality and speed on structured knowledge tasks — for example, the 2023 BCG/Harvard field study on consultants found meaningful productivity and quality gains on tasks within the tool's capability, alongside a warning about tasks outside it. Treat AI as an accelerant on the analysis, not a replacement for operator judgment on the fixes.
If you want to pressure-test your unit economics and turn the findings into an execution plan, you can run your business context through Percision here.
What this looks like when the analysis is actually run
A unit teardown in DTC comes down to two numbers: contribution per order, and orders per customer. This run attacked the second.
The subject is Northaven Goods, a sample company profile we use for testing rather than a customer: a direct-to-consumer housewares brand, $72M net revenue, 95 staff.
Excerpt from a real Percision run · Cost Reduction & Efficiency (T7) · sample company profile
The unit being added. A replenishment subscription on high-wear components at a 65% gross margin, on a 12- or 24-month cycle, sold to 340K active customers who already own products carrying the lifetime guarantee.
The attach ladder. 12% by Month 12 — 40,800 subscribers, $2.7–3.1M of incremental revenue. 18% by Month 24 — $4.9–6.4M. 22% by Month 36 — $7.1–9.2M. Annual subscription churn capped at 35%; subscription gross margin at 62% or better.
The unit cost of building it. $600–820K total — $420–580K of parts inventory buffer plus $180–240K of platform and integration costs. Funded from the existing $9.2M cash position within the current 18-month runway.
The other lever on orders per customer. Repeat purchase rate from 31% to 35% and AOV on repeat orders from $86 to $94–98, lifting LTV/CAC from 2.4x to 3.1–3.4x on $400–600K.
The floor. Terminate if the attach rate remains below 8% after Month 12 or annual churn exceeds 45% for two consecutive quarters; inventory buffer liquidated at 40–50% recovery value.
| Assumption | Probability |
|---|---|
| ≥12% of 340K active DTC customers opt into subscription within 18 months | 0.75 |
| Subscription churn remains below 40% annually | 0.7 |
| Parts inventory buffer of $420K-$580K is sufficient to maintain 95% fulfillment SLA | 0.85 |
Both fixes work on orders per customer rather than contribution per order, which is the correct read for this business. Gross margin is already 58% and AOV is $86 — neither is the problem. The problem is that customers buy 1.56 times in a year, and both programmes are attempts to make that number larger.
The subscription is the more durable of the two because it removes the decision. A replenishment cycle produces an order without the customer choosing to place one, which is a structurally different kind of repeat rate from one produced by email persuasion.
Read a complete Percision report — every page, no email required.
FAQ
What's the fastest way to find a margin leak in DTC? Rebuild contribution margin per order with every variable cost loaded in, then segment by SKU, channel, and discount code. The average hides the leak; the segments reveal it.
What LTV:CAC ratio should DTC brands target? A common anchor is roughly 3:1 with CAC payback under 12 months, but the right target depends on your gross margin and how cash-constrained you are. Prioritize payback period if cash is tight.
Do I need software to run unit economics? No. If you have clean order-line data and analytical capacity, a spreadsheet works. Tools like Percision help when you need consulting-grade analysis and a board-ready output fast — not because the math itself requires software.