POAS vs ROAS vs MER: which measure answers your question?

ROAS compares attributed revenue with advertising spend. MER compares total revenue with the marketing spend included in your calculation. POAS compares a defined profit or contribution figure with advertising spend. They answer different questions: channel attribution, business-wide efficiency and product economics. State the period, attribution basis and included costs before comparing them.

In this article, POAS uses contribution before ad spend as its numerator: net revenue less the direct variable costs of the orders kept. It is not net profit; overheads, salaries and finance costs sit below it. None of the three ratios, on its own, proves that advertising caused the revenue or contribution it is compared with.

To run the comparison on your own figures, use the POAS calculator. Our broader approach is set out in how we audit Google Ads accounts.

ROAS, MER and POAS compared

Use all three, each for its own question. POAS for whether attributed orders left contribution behind after their variable costs. MER to sanity-check the blended marketing investment. ROAS for tactical bidding within a channel that has consistent margin. None of them, alone, shows that advertising drove incremental revenue; that needs a controlled test.

The mistake is using ROAS as the only metric. The other mistake is using MER as the only metric - it tells you the marketing function works but not which channels carry it.

ROAS, MER and POAS on one set of numbers: a hypothetical worked example

Most ecommerce POAS calculations break at the data integration step. You need three sources merged:

• Google Ads - spend and attributed orders and revenue
• Your commerce platform - order detail, COGS per variant, discounts, shipping cost, payment fees
• Returns data - refunds, resaleable stock recovered, and the cost of handling each return

Merge at order or SKU level, over a window long enough for the returns tail in your category to land. Fashion needs a longer window than consumables; take the window from your own observed return and refund lag rather than a fixed rule.

The figures below are a hypothetical D2C fashion brand, not client data. They exist to show how the three metrics behave on one month of trading. Note the two different denominators: ROAS and POAS use Google Ads spend and Google Ads attributed revenue; MER uses whole-business revenue and total marketing spend across all channels. Dividing whole-business revenue by one channel's spend is the single most common MER error.

Hypothetical D2C fashion brand - one month

CHANNEL VIEW (Google Ads only)
Ad spend:                                  £40,000
Attributed revenue ex VAT, before discounts:
                                          £180,000
  less discounts:                         -£12,000
  less refunded revenue (returns):        -£18,000
Attributed NET revenue:                   £150,000
ROAS (net revenue / ad spend):               3.75x

  less COGS on goods kept (45% of net):   -£67,500
  less shipping and fulfilment:           -£13,500
  less payment fees (2.2% of net):         -£3,300
  less return handling and write-off:      -£6,000
Contribution BEFORE ad spend:              £59,700
POAS (59,700 / 40,000):                      1.49x
After-ad contribution (59,700 - 40,000):   £19,700
  (contribution, not net profit - overheads sit below)

BUSINESS VIEW (all channels)
Whole-business net revenue:               £400,000
Total marketing spend (all channels):      £80,000
MER (400,000 / 80,000):                      5.00x

Read across the three: 3.75x ROAS looks strong; MER at 5.00x describes the ratio of whole-business revenue to total marketing spend, which on its own says nothing about efficiency until margin, non-marketing costs and attribution are known; and POAS at 1.49x (59,700 ÷ 40,000 = 1.4925, rounded) says each pound of Google Ads spend returned about £1.49 of contribution before ad spend. £19,700 of contribution is left after paying for the ads, and overheads still come out of that. Refunded revenue is removed once, on the revenue line, and the COGS of resaleable returned stock is excluded from the cost line - deducting both the refund and the full COGS again would understate contribution. None of these ratios shows that the advertising caused the revenue or contribution; that needs an incrementality test.

The single most common mistake at this step is using gross margin instead of contribution margin. Gross margin only subtracts COGS. Contribution subtracts all direct variable costs. A product with a high gross margin, a high return rate and expensive shipping can contribute far less than the gross figure suggests.

Why ROAS lies

ROAS - Return on Ad Spend - is the most commonly reported metric in Google Ads. It's also the most commonly misunderstood.

The formula is simple: ROAS = attributed revenue / ad spend.

So a £10,000 monthly ad spend with £50,000 of attributed revenue gives you a 5× ROAS. Sounds great. Often is great. Sometimes it's a disaster.

Here's the disaster version. Say your products have a 20% contribution margin - that's revenue minus cost of goods, minus shipping, minus payment processing, minus discounts, minus returns. So that £50,000 in revenue from your 5× ROAS campaign actually generates £10,000 in contribution margin. Same number as your ad spend. You've made literally nothing.

Now scale the campaign. 6× ROAS. £60,000 revenue from £10,000 ad spend. Sounds even better. The contribution margin is £12,000 - still positive but barely. Spend more, push for growth, and you'll happily run a 4.8× ROAS account into the ground without anyone noticing until cash gets tight.

Five specific things ROAS hides:

1. Margin mix shift across SKUs. Your blended ROAS averages winners and losers. The losers can be eating the winners' profit and you wouldn't see it.
2. Returns. Conversion value is normally recorded at order confirmation, while refunds land later - the lag depends on your category and returns policy, so take it from your own data. Unless refunds are fed back through conversion value adjustments or an offline import, reported ROAS will not reflect them.
3. Cost of customer acquisition vs lifetime value. The same 3× ROAS can be a reasonable trade or a loss depending on measured repeat behaviour and margin. Whether it is one or the other is a measurement question, not something the ratio can tell you.
4. Cash cycle. Cash settles some time after the order, depending on payment provider, terms and returns window. Ad spend is paid on the platform's own billing cycle. Strong ROAS with a slow cash cycle can still create a funding problem.
5. Working capital tied up in inventory. That £50,000 revenue assumed you had inventory to fulfil. If your stock costs £15,000 to produce, that £15,000 was capital you can't redeploy until the next cycle.

The agency reporting a 5× ROAS and celebrating in your monthly review is optimising for the number that makes them look competent. They're not optimising for whether your business is making money.

What POAS actually measures

POAS (Profit on Ad Spend) measures contribution before ad spend, per pound of advertising spend. Unlike ROAS, it reflects the direct variable costs an ecommerce P&L actually pays. It is not net profit: overheads, salaries, software and finance costs sit below the line POAS stops at.

The full formula, with every variable cost broken out:

POAS = Contribution before ad spend / Ad spend
Contribution before ad spend = Net revenue − COGS on goods kept − Shipping and fulfilment − Payment fees − Return handling
After-ad contribution = Contribution before ad spend − Ad spend

A 1× POAS means every pound spent on ads returned exactly one pound of contribution before ad spend: break-even on variable costs, before any overhead is recovered. Anything above 1× is contribution towards overhead. There is no universal target above that; the right level depends on your own margin structure and the overhead you need advertising to help carry.

Each variable in that formula matters, and in our experience one or more of them is usually missing from the data an account can actually report on.

Net revenue. Revenue excluding VAT, after discounts and after refunds. Getting this base right prevents the most common error in the calculation: deducting the same cost twice.

COGS (cost of goods sold). What it costs you to acquire or produce the product before any other expenses. Most Shopify accounts have this in the product object, but variant-level COGS is often missing. If you sell t-shirts in 5 sizes and 4 colours, that's 20 variants - and each may have different supplier pricing. Generic COGS at product level overstates your margin.

Shipping. Outbound delivery costs, customs duties for cross-border, the unsexy operational tax. If you offer free shipping at £50+, the cost still exists - it's just been hidden in your AOV math.

Returns. Return rates vary widely by category, and fashion is usually at the high end. Treat returns explicitly: take the refunded revenue out, take the COGS of resaleable returned stock back out of costs, and deduct the real cost of handling the return - inbound carriage, inspection, restocking labour and any written-off units. A returned order is not a clean reversal, but it is not a total loss either.

Payment fees. Rates differ by provider and by method, and your blended rate depends on the payment mix. Take the blended rate from your own settlement reports rather than a headline card rate.

Discounts. Both code-driven (NEWSALE15) and the harder-to-see ones (loyalty programme cashback, BNPL platform discounts taken from the merchant side). Subtract before calculating margin.

The reason agencies don't lead with POAS is most don't have access to half this data. Or, if they have access, they don't have the data integration discipline to keep it accurate week-to-week.

How to set your own POAS target

There is no universal "good" POAS, and published category averages are not a substitute for your own numbers. We do not publish benchmark bands, because a target built from someone else's margin structure will misprice your catalogue. Derive it instead, in four steps.

1. Start at break-even. A POAS of 1.0× means the spend returned exactly its own value in contribution margin. Below that, the campaign consumes margin. Your break-even ROAS is the same constraint expressed in revenue terms - the break-even ROAS calculator converts between the two.
2. Add the overhead you need advertising to carry. Contribution margin pays for overhead before anything reaches net profit. Decide what share of fixed cost this channel is expected to recover, and the required POAS follows arithmetically.
3. Adjust for repeat behaviour, with evidence. A lower first-order POAS is defensible only where repeat purchase is measured, not assumed. If you cannot evidence the second order, price the first order as if it is the only one.
4. Split the target by SKU job. Review contribution by product group before deciding whether separate campaign targets are justified. Each job carries its own target - see the BOI® framework.

Different commercial jobs justify different targets on purpose. An acquisition campaign backed by measured lifetime value can run below the blended target. A clearance campaign can run below break-even deliberately, because the objective is cash recovery inside a contribution floor and a time box, not margin. Brand defence is an incrementality question before it is a POAS question.

A single campaign target can be appropriate when values reflect contribution and there is enough shared signal. Separate targets require a documented commercial reason and sufficient data.

SKU-level POAS - why blended targets kill margin

The biggest leverage move on POAS isn't the calculation itself. It's segmentation.

A single blended POAS is a spend-weighted average. Where margin, return rates and price points differ across the catalogue, that average can sit comfortably above 1× while some product groups are below break-even and others are funded below what their contribution would justify. You cannot tell which from the blended figure; you have to look at contribution by product group and by spend.

At JudeLuxe we run a framework called BOI® - Bid On Intent - that classifies every SKU into one of five commercial jobs, then bids against that job. The five jobs:

• Scale. Demand and margin both support growth, so the constraint is how much volume can be bought while contribution holds.
• Profit. Margin is healthy but demand is capped, so the job is protecting price and contribution rather than chasing volume.
• Protect. Brand or category-defining products where the question is share of the auction, and where incrementality testing matters more than the reported ratio.
• Recovery. Ageing or overstock inventory, worked inside a defined contribution floor and time box, measured on cash recovered rather than margin.
• Gateway. Products that recruit new customers. Any allowance made on the first order should be held against measured repeat behaviour, not an assumed lifetime value.

Jobs are reviewed as inventory, cash position and demand change; the review cadence follows how fast those inputs move. Product data can help separate products where commercial differences justify it. Margin itself reaches bidding through the conversion value you send, not through conversion value rules, which adjust a reported value by conditions such as audience, device or location.

The whole framework lives at judeluxe.com/boi. The point for this article: where contribution varies materially across the catalogue, review whether the values and campaign structure reflect those differences.

Using POAS in bid automation

Google Ads doesn't have a native POAS bidding option. It has Target ROAS and Maximise Conversion Value. To bid against POAS you have to push margin data into those mechanisms.

The buyer-facing questions are whether contribution is measured accurately, how corrections such as refunds are handled, and whether the account structure reflects meaningful commercial differences without fragmenting the data.

• Send a measured contribution value with the conversion. Pass contribution, not revenue, as the conversion value - through the tag, an offline conversion import or the Conversions API. Target ROAS then operates against margin rather than revenue. This underpins our Performance Max and Google Shopping management.
• Conversion value rules are a different tool. Value rules adjust a reported conversion value by conditions such as audience, device or location. They are useful for known value differences between segments; they do not carry per-order margin, and they should not be presented as a POAS implementation on their own. Conversion value adjustments are a third mechanism again, used to restate a value after the fact - for example when an order is refunded.
• Structure follows material commercial differences. Products should only be separated where the business question and available evidence justify it. Excessive fragmentation can weaken the data available to each campaign.
• Per-order contribution needs reliable source data. Where the business can maintain accurate product costs and returns, a measured contribution value can give bidding a more relevant commercial signal.

For a full account-level diagnostic of where your POAS is leaking, see how we audit Google Ads accounts. If you want these decisions handled as part of ongoing ecommerce PPC management, the commercial page explains the scope.

Sending a measured contribution value changes what bidding optimises towards. Value rules and value adjustments handle segment differences and corrections, not per-order margin. Product structure matters when the catalogue has material commercial differences, while any per-order contribution signal depends on accurate source data.

What absolutely doesn't work: running default ROAS bidding, hoping the algorithm figures out margin on its own. It can't. It optimises for what you measure.

When NOT to use POAS

POAS is the right primary metric for most ecommerce Google Ads decisions. It's not the right metric for every decision.

New customer acquisition campaigns. A first-order POAS below 1.0× can still be the right trade if repeat purchase is measured, not assumed. Lifetime value is an estimate with a confidence range, and cohorts move; treat first-order POAS as a constraint and hold the acquisition target against observed repeat behaviour rather than a modelled promise.

Brand defence campaigns. Bidding on your own brand terms protects share-of-voice from competitors, but a large share of those customers would have found you anyway. Raw POAS therefore flatters brand spend by an amount you cannot know without testing. The right measurement is incrementality, not the reported ratio.

Liquidation / clearance campaigns. You are moving inventory that would otherwise age or be written down. POAS on that activity can sit below 1×, and after-ad contribution can be negative, and it may still be the right decision. Work it inside a defined contribution floor and a time box, and judge it on cash recovered per pound spent.

Top-of-funnel awareness. YouTube, Demand Gen, display - these often have direct response components but the primary value is brand-aided downstream conversion. POAS as a sole metric here is misleading.

The discipline isn't "POAS for everything." It's "the right metric for the commercial job, and POAS for everything that's directly aimed at profitable revenue."

How JudeLuxe applies POAS for clients

POAS isn't a metric we report on. It's the metric we bid on.

Onboarding follows the same sequence in every account. How long it takes depends on how accessible the margin and returns data is:

1. Data audit. What's the variant-level COGS situation? Returns integration? Payment fee accuracy? We surface every gap before reporting numbers we don't trust.
2. POAS baseline. What is the contribution-based POAS, by product group, today? Because ROAS carries none of the variable costs, it is normally higher than POAS on the same period; how much higher is specific to your margin structure.
3. SKU job classification. Every SKU gets one of the five BOI® jobs assigned. Custom labels go into the feed.
4. Bid restructure. Campaign structure, targets and listing groups are redesigned around job-level targets rather than one blended number, with contribution carried in the conversion value.
5. Governance loop. Regular re-classification of SKUs as commercial reality shifts (stock-out risk, margin drift, promo, lifecycle change).

For Thermos, the case study reports a +94% increase in contribution margin at account level, without an ad spend reduction. That is a published before-and-after across a period in which several changes were made at once; it is not an isolated causal measurement of any single change.

For UK Soccer Shop, £520k of gross cash was released from ageing stock during a 45-day recovery programme, client-reported from their own finance data. That is cash recovered, not contribution and not profit.

Neither result isolates the effect of POAS bidding on its own; both are reported alongside their limits in the case studies. The metric isn't the win. The way the metric drives weekly decisions is the win.

Frequently asked questions

What to do this week

If you've never calculated POAS for your account, here's the 30-minute audit:

1. Export a recent period of Google Ads spend and attributed orders, long enough to cover your returns lag
2. Export your commerce platform orders for exactly the same period
3. Note your average contribution margin % (if you don't know, ask your CFO - they almost certainly do)
4. Multiply attributed revenue by (1 − average variable cost %) for a rough contribution margin estimate
5. Divide by ad spend

If the result is materially lower than your reported ROAS would suggest, you have a measurement gap that's quietly costing money.

For a deeper audit - variant-level COGS integration, returns adjustment, SKU job classification, the full bid restructure - that's what we do at JudeLuxe.

See what your ROAS is hiding

If your account cannot show contribution margin by SKU, the reported ratio is not a commercial answer. A commercial review takes 30 minutes and looks at your Google Ads spend, your margin structure and where contribution is leaking. No preparation required.

Typical client spend is around £10k/month, and we welcome enquiries below that. Fixed fees from £2k/month. If we are not the right fit, we will say so.