POAS (Profit on Ad Spend)

POAS (Profit on Ad Spend) measures the contribution margin generated per unit of ad spend, where ROAS measures only revenue per unit of ad spend. It is the profit-truth sibling of ROAS — the metric the run-103→107 measurement cluster (Retail Media → Incrementality → Media Mix Modeling (MMM) → Multi-Touch Attribution (MTA) → ROAS) names as the correction for ROAS's profit-blindness. The term was coined and trademarked by the Danish vendor ProfitMetrics, so the concept is vendor-originated rather than a neutral industry standard.

Firewall: every claim below is what a source reports. See ../../CONTEXT.md Rule 1. This was a web-only, vendor-skewed run — every fetched source is a tool vendor or agency with a commercial interest (ProfitMetrics, JudeLuxe, Smarter Ecommerce/smec, Channable); the Reddit and YouTube practitioner streams were both down (MCPs not connected). No independent/academic source on POAS surfaced. Treat all benchmarks and case-study results as self-reported.

Definition & formula

Channable reports POAS measures actual profit after costs while ROAS measures only revenue from ads (Channable, 2026-04-20, vendor). JudeLuxe gives the full formula breaking out every variable cost:

POAS = (Revenue − COGS − Shipping − Returns − Payment fees − Discounts) ÷ Ad spend

— and stresses that the numerator is contribution margin, not gross profit (JudeLuxe, 2026-05-25, vendor).

ProfitMetrics frames POAS as a corrective to the fact that "ROAS" in practice became "Revenue on Ad Spend" rather than "Return on Ad Spend", because true return/profit was never available to Google or Facebook, so they substituted readily-available revenue (ProfitMetrics, undated, vendor + trademark owner).

[!unverified] "POAS" being a registered trademark of ProfitMetrics is stated by JudeLuxe (which notes it applies the metric but is unaffiliated). Relevant attribution caveat: the metric is vendor-originated, not a neutral industry standard (JudeLuxe, 2026-05-25).

Why break-even POAS is always 1.0

JudeLuxe reports the key structural advantage over ROAS: break-even POAS is always 1.0 — at 1.0, contribution margin exactly covers ad spend — which removes the ambiguity of break-even ROAS, which varies by product margin (a 50%-margin product breaks even at ROAS 2.0, a 20%-margin product at ROAS 5.0). A 2.0× POAS means £2 of contribution margin per £1 of ad spend (JudeLuxe, 2026-05-25, vendor). Compare ROAS, where break-even ROAS = 1 ÷ gross margin.

Worked examples — where high ROAS collapses to low POAS

• The 5× ROAS that earns nothing. JudeLuxe: a £10,000-spend → £50,000-revenue campaign (5× ROAS) on products with a 20% contribution margin produces only £10,000 of contribution margin — exactly equal to ad spend, i.e. POAS ≈ 1.0, literally zero net gain (JudeLuxe, 2026-05-25, vendor).
• The $95 product. smec illustrates a $100 product sold for $10 ad spend showing "incredible" ROAS even if the product costs $95 to make and ship — a net loss; the COGS blind spot POAS was built to fix (smec, 2026-03-20, vendor).
• Single high ROAS targets shut down profitable orders. ProfitMetrics' multi-order example: break-even ROAS varies wildly by product (a frying pan breaks even at ROAS 2.12; a mixer at ROAS 8.88). Applying one high ROAS target (e.g. 10) would wrongly shut down the three most profitable of four orders, whereas POAS>1 correctly flags all four as profitable (ProfitMetrics, undated, vendor).

[!unverified] ProfitMetrics' framing that "you can run a spectacular 800% ROAS and simultaneously make structural losses", and that a 4.0 ROAS on a 20%-gross-margin product is a loss-making campaign — exact "800%" phrasing came via search snippet (ProfitMetrics, undated, vendor).

The fashion case — returns as the hidden POAS killer

JudeLuxe's worked fashion (Google Shopping) example: £42,000 ad spend / £158,100 attributed revenue = 3.76× ROAS reported, but after COGS (45% = £71,145), shipping (£4.20/order = £3,570), payment fees (2.1% = £3,320), discounts (£6,200) and returns (32% of orders = £8,800), contribution margin is £65,065 → POAS = 1.55×. The gap between reported ROAS and commercial reality was ~£93,000 in costs ROAS never showed (JudeLuxe, 2026-05-25, vendor). This is the same case the ROAS page cites for "3.76× ROAS → 1.55× POAS".

JudeLuxe calls returns "the hidden POAS killer", with normal return rates of 30–40% in fashion, 10–15% in beauty, 8–12% in home (as-of 2026-06-27), and notes a returned order is "worse than zero" because it absorbs logistics, restocking labour and sometimes written-off inventory (JudeLuxe, 2026-05-25, vendor). See Returns Management and Bracketing (Fashion Returns).

Healthy POAS bands by category (as-of 2026-06-27)

[!unverified] Single vendor's 2026 client base, not a survey — directional only (JudeLuxe, 2026-05-25, vendor).

Category	Healthy POAS band
Luxury / premium	1.8× – 2.8×
Supplements	1.8× – 2.6×
Beauty	1.6× – 2.4×
Home & furniture	1.5× – 2.2×
Pet	1.5× – 2.1×
Food & bev	1.4× – 2.0×
Sports / fitness	1.3× – 1.9×
Fashion & apparel	1.2× – 1.8× (returns drag; full-price > markdown)

JudeLuxe adds that new-customer-acquisition campaigns may justifiably run at 0.8–1.2× POAS if backed by LTV, and liquidation campaigns at ~0.5× POAS on purpose — a single blended POAS target across an account is described as the wrong approach (JudeLuxe, 2026-05-25, vendor).

POAS vs ROAS vs MER vs iROAS

JudeLuxe's taxonomy: ROAS = revenue ÷ ad spend per channel (tactical, single-channel, consistent-margin use); MER = total revenue ÷ total marketing spend across all channels (blended business-level view); POAS = contribution margin ÷ ad spend (whether spend actually made money); iROAS = incremental revenue from a test group ÷ ad spend (causal lift vs background demand — see Incrementality). JudeLuxe's stance is "use all three" (ROAS/MER/POAS), each for a different job (JudeLuxe, 2026-05-25, vendor).

JudeLuxe flags a common error: using gross margin (COGS only) instead of contribution margin (all variable costs) — a 60%-gross-margin product with high returns and expensive shipping may have only ~25% contribution margin (JudeLuxe, 2026-05-25, vendor).

How it is implemented

• No native Google bid strategy. JudeLuxe reports Google Ads has no native POAS bidding; you implement POAS by pushing contribution margin (instead of revenue) into Google's conversion value via conversion-value rules, server-side conversion value, or the Conversion API — so Target ROAS effectively becomes Target POAS (JudeLuxe, 2026-05-25, vendor). See Server-Side Tracking.
• Three data sources, 90-day window. A POAS calc for a Shopify brand requires merging order/SKU-level data over ≥90 days (to capture the returns tail): Google Ads (spend/attributed revenue), Shopify (orders, variant-level COGS, discounts, shipping, payment fees), and a returns tool (Shopify Returns, Loop, ReturnsLogic) (JudeLuxe, 2026-05-25, vendor).
• Vendor pipelines. ProfitMetrics implements POAS via Server-Side Tracking and "Plug'n'Play" connections feeding profit data into Google Ads and Facebook so bidding optimises on profit (ProfitMetrics, undated, vendor). Channable imports cost data and auto-segments products into buckets (Stars, Potentials, Invisibles, Underperformers) for budget allocation (Channable, 2026-04-20, vendor).

[!unverified] Vendor/agency performance claims, self-reported: ProfitMetrics says customers saved "up to 60% on ad spend" while revenue rose; JudeLuxe's "BOI® (Bid On Intent)" SKU-classification claims Thermos 1.2×→2.3× POAS over 12 months and "UK Soccer Shop" £520k recovered spend in 6 months (ProfitMetrics; JudeLuxe, 2026-05-25).

Criticism / limitations

smec (which pitches its own "True Profit Optimization") argues POAS is "an efficiency ratio, not a volume driver" — structurally identical to ROAS (only the numerator changes), so in Google Smart Bidding a Target POAS acts as a rigid ceiling that restricts impression share and caps absolute cash profit; it claims static POAS bidding will be "obsolete for enterprise retailers" by 2026 (smec, 2026-03-20, vendor). smec adds that POAS is a per-item/per-transaction metric that ignores inventory depth, tied-up capital and dead stock — you can hit 500% POAS on one low-volume item while overall profitability stagnates (smec, 2026-03-20, vendor).

smec also notes POAS accuracy depends on attribution model and tracking quality — last-click oversimplifies, ad blockers make browser pixels unreliable, and delayed/multi-touch conversions misalign spend with reported profit (smec, 2026-03-20, vendor); see Multi-Touch Attribution (MTA) and Server-Side Tracking.

JudeLuxe lists situations where POAS is the wrong primary metric: new-customer acquisition (use LTV-adjusted CAC), brand-defence terms (raw POAS overstates value by 30–60%; measure Incrementality), liquidation (use cash recovered per pound), and top-of-funnel awareness (JudeLuxe, 2026-05-25, vendor).

Contradictions

Key terms

Term	Meaning
POAS	Contribution margin ÷ ad spend; break-even always 1.0
Contribution margin	Revenue − COGS − shipping − returns − payment fees − discounts (all variable costs)
Gross margin	Revenue − COGS only (narrower; using it for POAS is the common error per JudeLuxe)
Target POAS	Contribution margin pushed into Google's conversion value so Target ROAS = Target POAS
iROAS	Incremental revenue ÷ ad spend (causal — see Incrementality)

What practitioners report

Not gathered this run — the Reddit (reddit-research MCP not connected) and YouTube (Apify transcript actor not connected) streams were both down, so there is no practitioner counter-narrative on whether POAS is worth the data-engineering cost, or how real operators set targets. Candidate YouTube videos logged for re-fetch: "POAS vs ROAS … w/ Adriaan Dekker" (The Paid Media Lab, ~Jan 2025, zEyDyajiXss), DGWaGgcnXc0, ApT-nrx4_2s, dZZymHQ1FOQ.