How should we account for discounts and returns in marketing metrics?
Our answer
Judge marketing on store net revenue after discounts and returns. Count each discount once, in CAC or first-order margin, book expected returns when you report, and restate campaigns once their returns land.
- Confidence
- Medium
- Strongest proof
- 3 official rules
- 4 practitioners, 7 opinions or claims
- Sources
- 13 sources
- 14 statements
- G
WE
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentAd platforms count revenue before returns come back. They credit it at purchase, while return windows run 30 to 60 days. In one worked example, a reported 4 times ROAS becomes 3.2 times (Webgility, ecommerce accounting software maker).
Return rates can reverse which campaign looks better. In one worked example, a 4.5 times campaign with 30% returns nets less than a 3.8 times campaign with 15% returns (JudeLuxe, UK paid search agency).
A discount can go in CAC or in first-order margin, as long as you count it exactly once. ConnectBooks (ecommerce bookkeeping service) puts first-purchase discounts in CAC, and Hologrow (ecommerce data platform) takes them from first-order margin.
Our judgment
- Use store net sales as the revenue line. Take them after discounts and returns for total sales divided by total ad spend (MER), new-customer revenue and campaign comparisons.
- Count each discount once. Put it in CAC or in first-order margin, and write the choice into your metric definitions.
- Book returns before they arrive. Report with an expected return rate and restate each campaign after its return window.
- Compare campaigns net of their own returns. Where return rates differ by channel or product, use ROAS net of each campaign's return rate.
Our judgment, built from Shopify, Flighted, Webgility, ConnectBooks, Hologrow, Fudge, Saras Analytics, Eightx, JudeLuxe and Admetrics: our reading of these sources, not a finding from any one of them.
What the sources say
14 statements from 13 sourcesGoogle lets advertisers retract conversions, such as returns, or restate their value for partial returns or lifetime value; retractions affect CPA and ROAS bidding and need a transaction ID.
“Retract conversions that should no longer be counted”
Shopify defines net sales as gross sales minus discounts and sales reversals, excluding taxes, shipping, duties and fees; total sales adds those back. It also tags each order as new or returning.
“Net sales = gross sales - discounts - sales reversals”
Google lets you retract or restate conversions by order ID or click ID within 54 days, but automated bidding reads changes only within 7 days; upload at least 24 hours after the conversion.
“Retracting removes a conversion completely from your reports.”
Beancount.io, which sells accounting software, relays that about 17% of holiday merchandise is returned, with no named source, and says to book discounts as contra-revenue and tag them by campaign.
“Tag discounts by campaign so January-you can answer which promotion moved profitable volume and which just trained customers to wait for codes.”
Saras Analytics, which sells a data platform, says finance has the reconciled numbers but weeks late, while marketing needs them sooner, so both need the same margin definitions.
“Marketing cannot wait for accounting cycles to make budget decisions.”
Fudge, a Shopify tool vendor, says after BFCM measure new-customer share, contribution per order by discount tier, the cohort's 30, 60 and 90 day repeat rate against earlier cohorts, and final returns.
“If your deepest discount tier contributed close to nothing per order, you bought revenue, not profit.”
Everboost, a retention agency, says Klaviyo's attributed revenue counts refunds and renewals and is not cash. It suggests tighter windows, holdouts, and judging email on repeat buying cohort by cohort.
“is a claim about influence, not a ledger of cash”
ConnectBooks, a bookkeeping firm, counts first-purchase discounts, creative, agency fees and staff time in CAC; its illustrative example gives $72.50 against a $50 ads-only figure, 45% higher.
“First-purchase discounts and acquisition coupons are real acquisition spend, not a revenue reduction to be ignored.”
Eightx, a CFO firm, uses an illustrative $100 order: about $20.60 of contribution margin after ads in a normal month but about $9.44 at Black Friday, so score BFCM on margin and the cohort's repeat buying.
“That second number is the one that pays rent.”
Hologrow, which sells a data platform, says to count a discount once, in CAC or in first-order margin, and to set CAC ceilings by customer quality, letting full-price buyers carry a higher CAC than discount-led ones.
“Either in CAC or in first order margin, but exactly once, and written into the definition doc.”
Admetrics, which sells measurement software, says finance should define contribution margin and growth should turn it into CAC ceilings and margin-based tROAS targets, rechecked after price or cost changes.
“Start with contribution margin per order, then subtract your required profit buffer. The remainder is your maximum CAC for that product or offer.”
Webgility, which sells accounting software, says ad platforms count revenue before returns arrive; in its example, $20,000 of returns turns a reported 4x ROAS into 3.2x. It rechecks ROAS 60 days after big campaigns.
“Ad platforms attribute revenue at the point of purchase, not after the return window closes.”
Flighted, a paid media agency, says to pick one revenue definition for MER (net means after returns and refunds) and keep it, and to judge promo periods separately because discounts compress MER.
“Evaluate promo-period MER separately from your baseline to avoid skewing your trend analysis.”
JudeLuxe, a Google Ads agency, says Google Ads ignores returns: in its worked example a 4.5x PMax campaign with 30% returns nets less than a 3.8x Shopping one with 15%. It imports return-adjusted values.
“A channel with 25% returns needs 33% higher ROAS target than a 10% return channel.”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer