top of page

New vs Returning Customers: Read Your Ad Results Separately

Writer: DIGITALTHROB
DIGITALTHROB
Oct 18
6 min read

Before increasing an acquisition budget, separate the people buying for the first time from those buying again. A strong total sales figure can contain both. It cannot tell you, on its own, whether finding new customers is affordable.


In the fictional ecommerce example below, $2,400 in media spend is linked to $7,500 in sales. Yet first purchases account for only $2,400 of that revenue. We work through the customer records, match the cost definitions and decide what the retailer should fund next.


Mint and orange editorial cover with two groups of sculptural customer figures and a circular repeat motif.

All figures are invented USD amounts, not client results or benchmarks. The method is practical: classify the buyers, reconcile their orders, label each cost ratio and write the next spending decision.


Define a customer by purchase history


A returning website visitor can still be a first-time buyer. Google Analytics describes new and returning users through website or app visits; its user-metric definitions do not establish whether someone has purchased before.


For this review, a first purchase means the person's earliest qualifying order in the retailer's reliably linked history. A pre-existing customer bought before the reporting period began. Someone acquired during the period remains in that acquisition cohort even if they place another order before the period ends.


Use the same business scope across online and connected offline records. Define a qualifying purchase consistently: this example excludes test orders, cancelled unpaid orders and fraudulent orders. A later refund changes the economics; it does not automatically erase a genuine past purchase and make the next order another acquisition.


Keep a customer ID, order ID, first qualifying purchase date and the order's purchase status. If records cannot establish earlier history or link an identity confidently, leave customer status unknown.


Reconcile customers and orders before reading the split


We select one completed period's orders linked to paid media under a documented reporting rule, counting each order ID once. That identifies attributed orders; it does not establish which sales advertising caused. Use the same selection rule throughout. Our guide to comparing attribution windows explains a separate source of reporting differences.


Revenue here is merchandise revenue after discounts and recorded refunds, excluding tax. Customers pay no delivery charge. Contribution is the money left after the matching product, fulfillment, payment, shipping and return costs, before advertising and management fees. All amounts are reconciled as of the review date.


  • First purchases: 30 confirmed new customers place 30 orders. Revenue is $2,400 and contribution is $960: $80 and $32 per first order respectively.

  • Further purchases by those same customers: six of the 30 people each buy once more. Those six orders add $540 revenue and $216 contribution.

  • Customers who bought before the period: 25 people place 40 orders, producing $4,000 revenue and $1,680 contribution.

  • Unknown customer status: eight orders produce $560 revenue and $168 contribution. Their number of distinct purchasers and earlier purchase history remain unresolved.


The totals reconcile to 84 orders, $7,500 revenue and $3,024 contribution. We have 55 reliably identified purchasers plus unresolved identities. The 30 newly acquired customers do not become 36 customers because six buy again.


For example, one of those 30 people places an $80 first order in week one and a $90 repeat order in week three. Both orders belong to the newly acquired customer cohort. Only the first order counts as a first purchase.


That distinction changes the revenue label: the acquisition cohort supplies $2,940 across its first and further purchases, while first-order revenue is $2,400. Either can be useful. Switching between them mid-report makes the acquisition result harder to interpret.


Give every cost ratio an honest denominator


The retailer spends $2,400 on media serving a mixture of acquisition and retention activity, plus $400 in management fees. Its report should label the resulting ratios explicitly:


  • Selected revenue divided by media spend: $7,500 ÷ $2,400 = 3.13, rounded.

  • Media cost per attributed order: $2,400 ÷ 84 = $28.57, rounded. This includes repeat and unknown orders.

  • Mixed media spend per confirmed new customer: $2,400 ÷ 30 = $80.

  • Total listed campaign cost per confirmed new customer: $2,800 ÷ 30 = $93.33, rounded.


The last two ratios describe account spending against confirmed acquisition volume. They are not isolated customer acquisition costs because the numerator also funds returning-customer activity. Dividing that same mixed spend by the 25 pre-existing buyers would not establish retention cost either.


First orders supply 32% of revenue, but that is not evidence that they consumed 32% of spend. Do not allocate the budget by revenue share to manufacture a customer-group cost.


Across all selected orders, $3,024 contribution minus $2,400 media and $400 management leaves $224 before fixed overheads and any unlisted costs. That is the account-level balance. It does not show which customer group deserves more funding.


Equally, comparing the $80 mixed-spend ratio directly with $32 first-order contribution would not prove acquisition is unprofitable. One number includes spending on existing customers; the other concerns first orders only. Match the scopes before drawing that conclusion.


Keep unknown buyers visible


Eight unresolved orders are eight order records to investigate, not eight confirmed new customers. They represent about 9.5% of the 84 orders; that is an order share, not a customer share.


The retailer checks guest checkouts, duplicate accounts and older sales records, retaining the evidence for each resolved match. Until a record is resolved, its revenue stays in the account total and outside both confirmed customer-status groups. The unknown orders must not disappear simply to make the split look complete.


Track that unknown order share alongside acquisition volume. A sudden change may reflect classification quality before it reflects advertising performance. If refund adjustments are still developing, keep the review date visible and update the economics consistently. Our guide to returns and refunds in ad results covers that reconciliation.


Use the split to approve a better next test


Our completed decision for this retailer is to hold the proposed acquisition budget increase. We can verify 30 new buyers, but cannot isolate their acquisition spending. The account also has only $224 left after its listed costs. Existing-customer sales are valuable; they do not resolve that acquisition question.


The owner instead considers a separately costed pilot: $600 media plus $100 servicing, with all $700 assigned to the acquisition objective upfront. The planning assumption is $32 contribution per first order. The owner wants $8 per confirmed new customer left after those pilot costs. These are fictional approval conditions, not recommended targets.


At 30 new buyers, $960 contribution minus $700 leaves $260, above the required $240 reserve. At 29, $928 minus $700 leaves $228, below the required $232. The proposal therefore needs at least 30 qualifying buyers at the assumed economics. This is a financial threshold, not a forecast that $600 will deliver them or evidence that 30 observations establish a reliable result.


The pilot brief records the customer-history rule, included costs, order-selection rule and review date before launch. Returning and unknown outcomes are shown separately and do not count toward the new-buyer condition. Costs stay assigned as agreed rather than being redistributed afterwards to make the result pass. Even this clearer costed objective does not establish how many buyers were incremental.


Judge first orders before borrowing future value


The six repeat purchases add $216 contribution to the acquisition cohort. That is observed value, not dependable lifetime value: other buyers have had less time to return, and later purchases may have different margins and costs.


If a first-order shortfall is acceptable, state how much the business can fund and when payback will be reviewed. Follow the same acquired customers over a defined period and deduct the costs of earning their later orders. All existing-customer revenue is not the future value of this month's new buyers.


Use our break-even ROAS guide for the fuller order-contribution calculation. Here, the immediate decision depends on keeping first purchases, repeat purchases and unresolved identities separate.


End the review with a decision and an owner


The retailer leaves this review with three assigned actions: operations investigates the eight unknown orders, finance confirms first-order contribution and affordable exposure, and the advertising team presents the pilot's cost and measurement brief. The acquisition increase stays on hold; meeting a pilot's financial condition would prompt another review, not automatic scaling.


Apply the same test to your own report: how many confirmed new customers did you acquire, which orders belong to them, and does the cost measure cover the same activity? If any answer is unclear, record the gap before approving more spend.


Our paid advertising management connects customer records with spending decisions. Bring your order history, reporting definitions and campaign costs to a strategy call so we can discuss what your next acquisition test needs to establish.


 
 
 

Comments


Discover clics solution for the efficient marketer
More clics

Never miss an update

bottom of page