NCPA
What you pay in ads to win one first-time buyer.
What it means
NCPA is your Ad Spend for the period divided by the first-time buyers it won — the euro price of one new customer. Repeat orders are excluded from the denominator, so this prices growth rather than sales. It's a cost, so lower is better: when it climbs, you're paying more for the same new name.
Show the math
Formula and a worked example
Ad Spend is all your paid media for the period — MetaMetaWeigh against Meta's own revenue and ROAS., GoogleGoogleCompare its return with your other channels. and anything else you feed in, added together.
New Customers counts unique buyers with no earlier order. A second order from the same buyer doesn't count again.
Worked example. In April you spend €20,000 and win 400 first-time buyers. NCPA = 20,000 ÷ 400 = €50.
In May you push to €25,000 and get 440. NCPA reads €57 — but the 40 extra customers cost €5,000 between them, or €125 each. The average moved a little; the price of the next customer more than doubled. That marginal number decides whether you scale.
It answers the question
What does one more new customer cost right now? If the first order and what they buy later don't cover that figure, growth is losing you money.
Why it matters
NCPA is what you'd pay a promoter to walk one new face through your door. That's fine while the face is worth more than the fee, and it's the whole problem when it isn't.
It also moves before revenue does. A paid channel saturates quietly — the audience thins, costs creep, and Total Sales hold for another month on repeat orders while NCPA climbs. By the time revenue dips, the cause is months old.
What good looks like
There's no universal number: a €300 furniture store and a €25 candle store can't share a line. Judge NCPA against your own last three to six months at a similar spend level, and against what a first order leaves you after product, shipping and fees. Direction matters more than level — NCPA rising while Ad Spend is flat means the channel is getting harder, whatever the number says.
How to improve it
| Lever | What you do | Expect | How long | Watch out for |
|---|---|---|---|---|
| Fast Stop paying to reach people you already have | Exclude past buyers from prospecting audiences and search campaigns | NCPA down 5–15% | 1–2 weeks | Total orders usually dip — those ads were catching easy repeat sales that a flow or campaign now has to pick up. |
| Fast Cut the spend that wins nobody new | Turn off ad sets and search terms with no first-time buyers over two full weeks | Ad Spend falls, new customers roughly flat | 1 week | Some of that spend was warming people who convert later; expect a softer month before the saving shows clean. |
| Slow Fix the first-order offer | One clear first-time offer, on a landing page that matches the ad that sent them | More new customers on the same Ad Spend | 3–6 weeks | A first-order discount lowers what that customer earns you — NCPA improves while Contribution Margin falls. |
| Slow Grow the channels that don't charge per customer | Email and SMS capture, referrals, content people find on their own | Ad Spend falls while new customer numbers hold | 1–2 quarters | Nothing lands for weeks, and it eats agency or in-house time that never appears in Ad Spend. |
Every lever costs something somewhere. The last column is the one to read twice.
Read it with
NCPA alone tells you the price. Next to how many customers you actually got, it tells you whether the price was worth paying.
Growth got cheaper
More first-time buyers for less each. Creative, offer or targeting is working. This is the only quadrant where scaling is a straightforward decision.
Buying the growth
Volume came from budget, not efficiency. Sometimes that's the plan — a launch, a Q4 push — but the average hides how expensive the newest customers were.
Trimmed back
Efficiency improved by shrinking. That's a real result if the cut was surgical, and a slow bleed if you bought less of everything.
Paying more for less
Both directions wrong. This is rarely a bidding problem — look for tired creative, a site change, or a price rise that broke the maths shoppers were doing.
One is what a customer costs, the other is the share of each sale left after variable costs. Side by side they answer what neither gives alone: whether a first order pays for the customer who placed it. A €50 NCPA against a 35% Contribution Margin on a €100 order — €35 left — means every new customer starts you €15 down.
Frequency sets the ceiling NCPA is allowed to reach. Customers who buy three times a year justify a much higher acquisition price than one-and-done buyers, so the same €60 NCPA is comfortable in one store and reckless in another. Read them together before calling any NCPA too high.
Common misreads
New customers sit in the denominator. A month heavy on repeat orders — a big flow send, a restock, a loyalty push — can raise NCPA while every campaign performs exactly as it did before.
The cheapest NCPA usually comes from the smallest budget, spent on people who were going to buy anyway. Judge it at the volume the business actually needs, not the volume that flatters the number.
It charges all your Ad Spend against first-time buyers only. Repeat orders the ads helped along never show up here, which is why NCPA reads higher than a plain cost per order in any period with repeat business.
Also called
New Customer Acquisition Cost · CAC · cost per new customer · cost per acquisition
See yoursYour NCPA for the last 30 days, next to Ads Spend and MER.
Open Marketing →