New customer acquisition cost is the price of a buyer who has never bought from you. It is not the figure on your dashboard. What sits there is a blend. Dashboards mix first-time buyers, repeat buyers, and people who were returning anyway with an ad in front of them. Blended numbers are always kinder than the real thing. That gap decides whether your growth is real, or merely better billing of customers you already had. Google publishes figures on this problem, and they are far more modest than the claims built on top of them.
This is for ecommerce operators whose revenue is climbing while the customer count stays flat.
What new customer acquisition cost actually measures
One thing: acquisition spend divided by the count of buyers who had never purchased from you before.
Almost every account divides by all purchases instead. That denominator includes returning customers, returning customers are cheap to convert, and the average therefore comes out low and comfortable.
Picture two stores reporting an identical $40 cost per purchase. One of them bought 400 new customers with $16,000. Another spent the same money on 120 new customers and 280 repeat orders. Only one of those businesses got bigger.
Splitting the denominator takes an afternoon, not a project. You need order history keyed to a customer ID or email. Add a flag marking the first order, then join it to campaign spend. Most store exports already carry that flag, so the work is joining rather than building.
Why blended CAC flatters almost every account
Blended CAC is total spend over total customers. It drifts downwards for reasons unrelated to your ads.
As a store ages, the share of orders coming from people who already know it rises. Spend holds flat, orders rise, blended CAC falls. Nothing about your ability to find new buyers improved at any point in that sequence.
There is the trap. A brand can watch its blended figure improve for two years while new customer acquisition cost doubles underneath it. No report in the ad platform will mention it.
Retargeting deepens the problem. Any campaign shown to last week’s visitors will report the cheapest purchases in the account. A large share of those purchases were happening regardless. Cheap conversions and caused conversions are different things.
The Google setting built for new customer acquisition cost
Google built a bidding option for exactly this problem, and it publishes numbers alongside it.
The new customer acquisition goal sits on Performance Max and Shopping campaigns. One option, which Google calls new customer value mode, bids more for a first-time buyer than for a repeat one. Another restricts the campaign to new customers entirely.
Google reports that the value setting delivered roughly 9% higher return on ad spend and a 7% lower acquisition cost. Both come from Google Internal Data covering October 2022 to April 2023. For lead campaigns using the new customers only setting, Google reports a 13% higher new customer ratio over that same 2022 to 2023 window. Cost falls 19% alongside it.
Read the footnote before you trust any of it. That October 2022 to April 2023 window makes the figures close to four years old, self-reported by the vendor, and published with no sample size.
What the 13% figure does and does not prove
A 13% higher new customer ratio is a genuine claim, and it is smaller than the headline suggests. Google logged it between October 2022 and April 2023.
Ratio is a share, not a count. A campaign that was 40% new customers becomes about 45% new. It can reach that share without buying one extra person, if total volume fell at the same time.
So the fair reading is narrow. Google’s bidding can bias delivery towards first-time buyers, using signals about who has bought before. It cannot manufacture demand from people who were never going to order.
Compare that with what gets built on top of it. Pages claiming the setting transforms acquisition economics are extrapolating from a footnote whose only disclosure is a date range. That footnote is still better than most of this field manages, which is the depressing part.
Separating new customer acquisition cost from retargeting
Structure comes first, and reporting comes second.
Exclude existing customers from prospecting using an uploaded customer list. Keep retargeting in its own campaign with its own budget. Spend then lines up with the audience it actually bought, which is the point of the exercise.
For reporting, pull orders carrying a first-order flag and join them to the campaign credited with the sale. Then divide prospecting spend by first orders alone.
Nobody gets a clean answer here. Attribution is imperfect, customer lists decay between uploads, and some first-time buyers do arrive through retargeting because they browsed once and left. Approximation aimed at the right question still beats precision aimed at the wrong one.
Bidding also needs volume before any of this works. Google’s own documentation states that Search and Shopping campaigns need at least 15 conversions in the past 30 days for target return bidding.
What this looks like inside a live account
Numbers from an account I am running now, rather than an argument in the abstract.
A US flag and banner ecommerce client, Meta sales campaigns, June to August 2026. Six sales ad sets bought 64 purchases at $67.24 each, on a 1.58 return on ad spend. Drop one failing geographic test and the same period reads 61 purchases at $53.72, on a 1.99 return.
Inside those six ad sets my best one bought purchases at $29.37, and my worst active one paid $93.96. Meta does not tell me how many of those buyers were new.
So my honest position is uncomfortable. I know the cost per purchase on that account to the cent, and the new customer acquisition cost not at all, because nobody has joined the order history to it yet. That is the normal state of affairs across accounts I inherit, not an unusual failure.
New customer acquisition cost rises when you scale
Prospecting gets dearer as you push it, and two separate forces do that work.
Auction prices climb underneath everybody. Meta reported average price per ad up 12% year on year in both the first and the second quarter of 2026, in its own investor results. No account-level change offsets a structural price rise.
Audience depth runs out as well. People most likely to buy see your ads first, so the marginal buyer at $50,000 a month was already cheaper to reach at $10,000.
When I took a client from $10,000 to $50,000 a month, I did not raise the Meta budget and hope. I added campaign objectives, added awareness campaigns, built custom audiences and lookalikes, and brought in channels the client was not using, including TikTok and Klaviyo. More surface area, more chances to convert, and Klaviyo is email rather than an ad platform, so its next send costs almost nothing.
Lead generation has the same problem under another name
Lead campaigns hide it differently, and the damage is the same size.
A cheap lead can be somebody already sitting in your database, or somebody who can never buy what you sell. Both count as a conversion, and both pull the reported cost downwards.
I held 6,070 leads at £11.40 for a UK staffing and childcare client, on Google Search only. Performance Max was tested on that account and dropped, because it delivered people seeking UK visa sponsorship rather than candidates who could take the jobs on offer.
Volume looked healthy throughout. Suitability did not, and no cost metric in the interface could see the difference between the two.
Ecommerce gives you the cleaner version of this problem, because a purchase is money that has already moved. Even then, the denominator still needs checking before anybody sets a target from it.
Setting a new customer acquisition cost ceiling
Start from margin, then decide how long you can afford to wait for the money.
Across my ecommerce clients gross margin runs between 20% and 50%. On a $100 order that leaves $20 to $50 of contribution, and first-order break-even sits exactly there.
Shopify put average order value at $114.70 across its full merchant base for Black Friday Cyber Monday 2025, published on 2 December 2025. At 40% margin that order contributes $45.88, so $45.88 becomes the ceiling if the first purchase has to pay for itself.
Raise the ceiling only when your own store data shows repeat buying you would bet on. Repeat purchase rate from your own database is the input that permits a higher number. Category assumptions borrowed from a blog post are not, and paying a second order’s worth of acquisition on a hope is how stores run out of cash while hitting target.
What to change this week
Three things, in this order.
Export last quarter’s orders with a first-order flag and count how many buyers were genuinely new. Most operators are surprised by that share, usually downwards.
Then divide prospecting spend by that count and compare it with the blended figure your dashboard reports. The gap between them is the size of the problem you have been managing by accident.
Finally, split retargeting into its own campaign with its own budget line, so next quarter’s comparison is clean rather than reconstructed.
Why a cheap conversion is not always a good one covers the denominator problem on the lead generation side. Two dated figures for what an order actually costs gives my own volumes and date ranges. Where the rising cost story really comes from separates the sourced part from the recycled part, and the cost difference between a first order and a repeat one covers the split in more detail. Most of this arithmetic gets applied inside search and shopping campaigns, which is what my Google Ads work handles. To have your own denominator checked before you trust it, book a teardown.