ROI vs ROAS vs POAS: Which Metric Proves ROI?

Olamide Sule, founder of Dolphin Analytics: a London-based analytics specialist who sets up and audits GA4, Google Ads and Shopify tracking for agency and in-house teams.

We get asked which metric to report almost every week, usually by a marketing lead who has to defend a budget to a board or a client. This is the explainer we give them: what ROI, ROAS and POAS each measure, when to reach for which, and why the number sitting on the ad dashboard is rarely the one to trust.

ROI vs ROAS: what is the actual difference?

ROAS measures the revenue a campaign earns for every pound of ad spend, while ROI measures the profit that campaign makes against its total cost. ROAS judges the media buy; ROI judges the business decision. The gap between the two is everything you spend that is not the ad itself: the cost of the product, shipping, payment fees, and the ad spend counted as cost rather than just divided into revenue.

That gap is why the two numbers tell different stories. A campaign can look strong on ROAS and thin on ROI at the same time, because ROAS never sees margin. Marketing teams live in ROAS because it moves daily and maps onto campaign decisions. Boards and finance live in ROI, because profit is the only number that pays wages.

ROI vs ROAS vs POAS at a glance

POAS sits between the two. Here is how the three line up before we work through an example.

Metric	What it measures	Formula	Best for
ROAS	Revenue per pound of ad spend	Revenue from ads ÷ ad spend	Steering campaigns day to day
POAS	Gross profit per pound of ad spend	Gross profit from ads ÷ ad spend	Ecommerce with thin or variable margins
ROI	Profit per pound of total cost	(Profit generated minus cost) ÷ cost	Proving whole-channel return to a board

None of the three is the “right” one. They answer different questions, and a team reporting only ROAS is answering the media question while the board is asking the profit question.

What does POAS mean, and why does it matter?

POAS means profit on ad spend: the gross profit a campaign generates for every pound of ad spend, rather than the revenue. Where ROAS divides revenue by spend, POAS divides profit by spend, so it strips out the cost of the goods before it judges the channel. Adjust’s marketing glossary defines POAS as profit on ad spend for exactly this reason.

POAS matters most in ecommerce, where two products can post the same ROAS and return wildly different profit. A candle on a slim margin and a skincare set on a fat margin can look identical in the ad platform. They are not identical to the business. POAS is the metric that tells a performance team to push budget toward the products that actually pay.

A worked example: one campaign, three very different numbers

Take a single campaign and run all three metrics across it. The numbers below are illustrative, chosen to be easy to follow rather than typical of any one account.

The campaign spends £10,000 and drives £50,000 in tracked revenue. Shopify sets out the ROAS formula as revenue divided by ad spend, so that is a ROAS of 5:1, five pounds back for every pound spent. Reported on its own, it looks like a clear winner, and this is the number that usually reaches the weekly campaign review.

Now bring cost in. If products, shipping and payment fees take 70% of that revenue, the campaign generates £15,000 in gross profit, so POAS is 1.5:1. Google frames ROI as net profit over cost, which here means the £15,000 profit minus the £10,000 spend, divided by the £10,000 spend: an ROI of 50%. The campaign still makes money, but the honest return is 50%, not the 5:1 the dashboard led with. Same campaign, same period, three numbers that would each start a different conversation.

How do you measure return on marketing investment?

Measuring return on marketing investment is less about picking a formula and more about trusting the inputs. The formula is the easy part. Getting the inputs right is where most reporting quietly falls apart. Work through it in order:

1. Match the metric to the decision. ROAS for campaign steering, POAS when margins vary, ROI for the board.
2. Agree what counts as a conversion. Decide the attribution model before you pull numbers, not after, so nobody relitigates it in the meeting.
3. Reconcile against a source of truth. Check ad-claimed revenue against what Shopify, the payment processor or the CRM actually recorded.
4. Subtract real costs. Cost of goods, fulfilment, fees and the spend itself, so the number is profit, not revenue wearing a nicer label.
5. Report one number with its assumptions. State the model and the costs alongside the figure, so it survives the first hard question.

Agencies we work with as a white-label analytics partner usually report ROAS for campaign steering and ROI for the quarterly review, which keeps the client conversation honest in both directions.

Why is your ROAS probably wrong before you even start?

The most common reason a return number is wrong has nothing to do with the formula. It is that the revenue feeding it was overclaimed. Ad platforms count conversions they only influenced, including view-through conversions where someone saw an ad and never clicked it, so Meta Ads Manager and Google Ads can each take credit for the same order. Google’s own attribution documentation shows how much the model chosen changes which touchpoint gets the credit. Add the platforms up and you can “sell” more than the business actually made. We wrote about how ad platforms overreport against real sales in more detail.

So a ROAS of 8:1 in the dashboard might be 4:1 once you strip out the sales the ad never really caused. Every metric downstream inherits that error. Fix the metric debate all you like; if the revenue input is inflated, ROI and POAS are inflated too.

How we make these numbers trustworthy

When we audit a client’s reporting, the first job is reconciling ad-claimed revenue against a source of truth like Shopify or the CRM, never the ad dashboard on its own. It is not always bad news. Reconciling one subscription box brand’s book-launch campaign against Shopify confirmed it really had returned 9.4x in 26 days, a number the team could take to the board with confidence because it came from the till, not the ad account.

Getting to a single, defensible number is what a reporting layer is built to do: one reconciled figure, with its attribution model and costs stated, instead of three dashboards that disagree. If your dashboards and your bank statement disagree and you are not sure why, talking to us is a quick way to get a considered read on the problem from a person, with no account access and no sales call. When you already know the numbers need fixing inside GA4 and the ad platforms, that reconciliation is the paid audit, scoped per property.