Which metrics actually run a DTC store?
About two dozen, and they sort into four families: acquisition metrics like CAC and MER, margin metrics like AOV and contribution margin, retention metrics like LTV and repeat rate, and creative diagnostics like hook rate. Every one is defined below, formula included, because most metric arguments are just two people quoting numbers from different families.
TL;DR
- CAC and CPA are not the same number. CPA counts every attributed order; CAC counts only new customers. Bid with CPA, make scaling decisions with CAC.
- ROAS is one channel's claim; MER (total revenue divided by total ad spend) is the blended truth. When they disagree, believe MER.
- Contribution margin per order is the foundation metric. It is literally your breakeven CPA, and every target you set flows from it.
- Creative diagnostics (CPM, CTR, hook rate, hold rate) tell you where an ad fails, not just that it failed.
- Five numbers give us most of an account's story in the first hour: contribution margin, MER, new-customer CAC, CVR by traffic source, and hook rate spread.
Acquisition metrics: what a buyer costs
Every metric in this family answers the same question in a different unit: what did it cost to turn a stranger into an order? The units disagree with each other more often than founders expect. The disagreements are where the money hides.
CAC (customer acquisition cost). The formula is total acquisition spend / new customers acquired. Spend means all of it: ad budget, agency fees, creative production if you are honest with yourself. Returning buyers do not belong in the denominator. CAC matters because comparing it against first-order contribution margin tells you whether growth is funding itself or eating cash.
CPA (cost per acquisition). Ad spend / attributed orders, new and returning alike, computed inside each platform from its own attribution. It is looser than CAC by design. Use it to steer cost caps and in-platform bidding, and never present a platform CPA as your real acquisition cost, because repeat buyers flatter it.
ROAS (return on ad spend). Attributed revenue / ad spend, per channel, reported gross: VAT rides inside the revenue number. ROAS is useful for ranking campaigns inside one account and dangerous as a profitability claim, because it knows nothing about margin and overlaps with every other channel's attribution.
Breakeven ROAS. AOV / contribution margin per order. The ROAS at which an order earns exactly zero. We walked the full formula, including the VAT trap and refused COD parcels, in our breakeven ROAS deep dive. No ROAS figure means anything until you know the breakeven it has to clear.
MER (media efficiency ratio). Also called blended ROAS: total revenue / total ad spend, all channels, no attribution model anywhere in sight. Platforms cannot inflate it and finance cannot argue with it. When channel dashboards and MER tell different stories, MER is the one your bank statement backs.
POAS (profit on ad spend). Contribution margin generated / ad spend. ROAS with margin priced in. Two campaigns at an identical 3.0x can sit on opposite sides of profitable once product margins differ, and POAS is the metric that catches it. Mixed-margin catalogs should steer by POAS; single-product stores can usually stay on ROAS.
Attribution window. The period during which a platform claims credit for a conversion, such as the 7-day click, 1-day view default on Meta. Change the window and every CPA and ROAS in the account changes with it. Comparing numbers pulled under different windows is comparing two different realities.
Incrementality. The share of attributed orders that would not have happened without the ad. Holdout and geo tests measure it properly; watching how MER reacts when a channel's budget moves approximates it cheaply. It matters because retargeting and branded search always look heroic on attribution while often adding the least that is genuinely new.
Costs are only half of the ledger, though. The other half is what an order is actually worth.
Revenue and margin metrics: what an order is worth
Dashboards report revenue. Banks deposit margin. This family is the bridge between the two, and it is where most launch plans quietly break, so take it in the order the money leaves:
Net revenue = AOV / (1 + VAT rate)
Gross margin = Net revenue - COGS
Contribution margin = Gross margin - fulfillment - payment fees - other variable costs
AOV (average order value). Total revenue / total orders, gross, as the customer paid it. AOV is the ceiling every acquisition cost lives under. That is why a bundle that lifts AOV by a few euros often does more for the P&L than a month of bid tweaking.
COGS (cost of goods sold). The landed cost of everything inside one order: unit cost plus inbound freight, duties, and the box it ships in. Not the supplier invoice alone. Understate COGS and every metric downstream of it inherits the optimism.
Gross margin vs net margin. Gross margin is (net revenue - COGS) / net revenue, and it says whether the product can work. Net margin is what survives after variable costs, ad spend, and fixed costs, and it says whether the business does. Plenty of stores with a 70% gross margin lose money; the distance between those two numbers is where.
Contribution margin. What one delivered order contributes after VAT and every variable cost, before ad spend and fixed costs. At the order level this is the same number as your breakeven CPA; at the monthly P&L level, subtract ad spend as well and you get what the store actually banked. It is the number the rest of this glossary stands on, and you can compute yours in about two minutes with our free breakeven calculator.
Fixed vs variable costs. Variable costs arrive with each order: COGS, fulfillment, payment fees, the courier. Fixed costs arrive regardless: salaries, software, rent. Breakeven math only works when the split is honest, and the most common cheat we see is filing genuinely variable costs, like returns handling or COD collection fees, under overhead where the per-order math cannot see them.
Retention and LTV metrics: what a customer becomes
Acquisition math prices the first order. This family prices the relationship, and it decides how aggressive that acquisition math is allowed to be. Some mediocre-looking first-order economics hide a great business. Some great-looking ones hide a treadmill.
LTV (lifetime value). Cumulative contribution margin per customer over a defined window: 60, 90, or 180 days. Never the literal lifetime, because a projection nobody can measure is a number nobody should bid on. LTV is what earns you the right to pay above first-order breakeven, which also makes it the most abused number in pitch decks.
Repeat rate. Customers with two or more orders / all customers in a cohort. A consumable with a strong repeat rate can afford to buy customers at breakeven or even at a controlled first-order loss. A one-and-done product cannot. This is the metric that decides which CAC philosophy you are allowed to hold.
Churn. The share of subscribers or active customers lost per period, mostly a subscription metric and a compounding one. The gap between 5% and 8% monthly churn looks like rounding in a report and behaves like a different business model in a year of LTV.
Payback period. The time until a customer's cumulative contribution margin covers their CAC. Profit on paper is not cash in the account: ads are billed this month, inventory was paid for even earlier, and a nine-month payback can sink a launch that the spreadsheet swore was profitable.
Cohort. Customers grouped by their first-purchase month and tracked forward together. Cohort tables are the only honest way to read retention, because blended averages mix customers acquired years ago with the ones acquired last week, and that mix can hide decay for quarters at a time.
Creative diagnostics: where an ad fails
A dead ROAS tells you an ad failed. These six tell you where, and that difference decides whether you fix the hook, the offer, or the landing page. They read as a chain: the auction charges you (CPM), some people stop (hook rate), some stay (hold rate), some click (CTR, CPC), some buy (CVR). Pulling this chain apart is the first thing we do on any Meta account we take over.
CPM (cost per thousand impressions). Spend / impressions x 1,000. The price of showing up in your audience's feed, set mostly by the market, the audience, and the season. You influence CPM less than anything downstream of it, which is why we treat it like weather: worth knowing, rarely worth fighting.
CPC (cost per click). Spend / clicks. On Meta you buy impressions, not clicks, so CPC is a derived number: your CPM divided by the clicks a thousand impressions produce. Rising CPC under a flat CPM is not an auction problem. It is a creative problem wearing an auction costume.
CTR (click-through rate). Link clicks / impressions. Use link clicks; the all-clicks version counts profile taps and see-more expansions and flatters everyone. CTR is the market grading your angle and your offer, and it is the denominator quietly setting your CPC.
CVR (conversion rate). Orders / sessions on the site, or orders / clicks from a channel's point of view. We read it by traffic source rather than as one sitewide average, because paid social, branded search, and email land with completely different intent, and the blended number hides which entrance is leaking.
Hook rate. 3-second video plays / impressions: the share of people your first frame stopped mid-scroll. The hook is the cheapest part of an ad to iterate and the first place we look when spend refuses to scale, because a weak hook starves everything downstream of data.
Hold rate. The share of hooked viewers still watching at the 15-second mark or the end of the video: ThruPlays / 3-second plays. A strong hook over a weak hold means the opening wrote a promise the body never paid off. The fix lives in the script, not the thumbnail.
The five numbers we check first in any account
We have been operating ecom P&Ls since 2014, on client accounts and on our own brands, and every audit still starts the same way: five numbers, pulled in a fixed order, before anyone opens a single campaign. Each number only makes sense in the light of the one before it.
| Order | The number | The question it answers |
|---|---|---|
| First | Contribution margin per order | Can this store afford ads at all, and up to what CPA? |
| Second | MER, trailing 30 days | Is the blended engine profitable right now? |
| Third | New-customer CAC vs breakeven CPA | Is growth funding itself or quietly burning cash? |
| Fourth | CVR by traffic source | Which entrance to the funnel leaks the most? |
| Fifth | Hook rate spread across recent creative | Is creative the constraint on scale? |
Margin comes first because nothing else is interpretable without it: a 4.0 MER is excellent for one cost structure and a slow leak for another. MER comes second because it is the one efficiency number attribution cannot decorate. The CAC comparison is third because that gap, per order, is either the profit engine or the hole in the hull. Only then do we look at conversion and creative, in that order, because sending better ads into a leaking page wastes the better ads.
Notice what the five cover: unit economics, blended health, acquisition, conversion, creative. One number per part of the machine. If all five look sane, the account's problems are the interesting kind. If the first one is broken, nothing downstream matters yet.
Put the glossary to work
The glossary is the easy part. The habit is the hard part: the same honest numbers, pulled the same way, every week, until dashboard arguments turn into arithmetic. Start with contribution margin per order this week and the other four of our first-hour five will suddenly have context.
And if you would rather have all four families wired into one P&L and watched by a senior team, that is what our full-funnel growth retainer is built around: acquisition, conversion, retention, and the numbers that connect them, run as one system. Most founders we meet can quote their ROAS to two decimal places and cannot name their contribution margin per order. Be the other kind.
