Profit on Ad Spend: the metric, where it came from and what a real implementation requires
POAS measures the gross profit a campaign returns for every euro of ad spend. BigConvert was the first agency in Romania to use it as the main bidding signal, starting in 2018 in our own affiliate accounts. This page explains how it started, why Google and Meta now build their own versions of the idea, and why most profit-based offers on the market stop at a calculator that the advertising algorithms never see.
Written by Ovidiu Golea, Co-Founder of BigConvert. Last updated .
What POAS is, and what ROAS misses
POAS (Profit on Ad Spend) is a performance marketing KPI: gross profit generated by a campaign divided by the cost of that campaign. ROAS (Return on Ad Spend) divides revenue by cost. The two numbers move together only when every product in a catalog has the same margin, which is never the case in a real store.
Two orders can carry identical revenue and very different profit. An algorithm optimizing on ROAS treats them as equal and spends the same to get either one. An algorithm optimizing on POAS pushes budget toward the order that actually pays. That is the whole idea, and it only works if the algorithm receives profit, not revenue, as the value of each conversion.
| Product A | Product B | |
|---|---|---|
| Sale price | โฌ100 | โฌ100 |
| Cost of goods, shipping, fees | โฌ40 | โฌ85 |
| Gross profit | โฌ60 | โฌ15 |
| Ad spend for the sale | โฌ20 | โฌ20 |
| ROAS | 5.0x | 5.0x |
| POAS | 3.0x | 0.75x |
Same ROAS. One order makes โฌ40 after advertising, the other loses โฌ5. Revenue-based bidding cannot tell them apart.
It came from affiliate marketing, where only the commission counts
In affiliate marketing the advertiser does not see order value as income. The only number that matters is the commission. From 2018 I imported commission value, not order value, as the conversion value in Google Ads, and let the bidding algorithms optimize on it. Every bid decision was made on the money we would actually keep.
The results are measurable. Today the BigConvert affiliate operation (together with Glew Solutions) runs at over 5,000 sales per day across roughly 150 Romanian stores on 2Performant, where it holds the first position in the network ranking, plus over 1,500 stores across Europe, exclusively through Google Shopping via BigConvert's own Comparison Shopping Service, on more than โฌ7M of our own ad spend. Nobody is paying us to be careful with that budget. The model has to work or we lose money.
The step from affiliate to retail is short. If optimizing on commission works for an affiliate, optimizing on margin works for a store. Commission and margin are the same thing seen from two sides: what is left after the cost of the sale. Optimizing on sales value is outdated. We optimize on profit.
First in Romania, and the platforms followed the same logic
BigConvert was the first agency in Romania to implement POAS as the primary bidding signal in client accounts. In 2020 I presented the model to Google while applying to list our profit-based optimization in their partner directory. At that point no major platform offered profit as a native optimization target.
Google later launched Conversions with Cart Data, a feature that lets advertisers send cost of goods sold with each conversion so Google Ads can report gross profit and campaigns can optimize on it. Meta's Conversions API accepts a custom value for each purchase event, so profit can be passed as the value that value-based bidding optimizes on. Both platforms adding a profit variant to their core products is the clearest confirmation that the signal matters. It also confirms the direction we took two years before those features existed.
In Romania, POAS has since become a common term and many agencies now describe a profit-based approach. We take that as a good sign for the market. It is also where the differences start.
Why a margin calculator next to the ad account is not POAS optimization
Most profit-based offers we see today work the same way: a dashboard or spreadsheet imports orders, applies a margin per product or per category, and shows profit after the fact, in parallel with the ad account. That is useful for reporting. It is not optimization, for one simple reason: the bidding algorithm never receives the number. Google and Meta keep optimizing on the value they are given, and that value is still revenue.
A number the algorithm never sees cannot change a single bid. For POAS to do anything, profit has to be imported into the campaigns and the advertising accounts as the conversion value, per order, close to real time, and corrected when orders are cancelled or returned. That is the part that takes engineering, and that is the part most parallel calculators leave out.
Three layers. Most offers stop at the first.
Calculate profit
Import orders, apply margins, show a profit figure next to the ad account. Any spreadsheet can do this. It describes what happened; it changes nothing about what the algorithm does next.
Where most "POAS" offers stopFeed profit into the ad accounts
Compute margin per order (COGS, shipping, payment fees, discounts), send it server-side as the conversion value to Google Ads and Meta, retract cancelled and returned orders. Now the algorithm bids on profit.
Where optimization actually startsAct on the catalog with price intelligence
Study the prices set in the market for each product, combine them with margin and demand, and adapt per catalog what gets pushed and at what price, within the margin the client sets.
What BigConvert adds on topPOAS plus market prices: sell what is demanded and what produces profit
Profit per order answers one question: did this sale pay? It does not answer the question that decides a catalog's results: which products have potential at the price the market currently accepts? For that we monitor the prices competitors set for the same products, continuously, and combine that signal with each product's margin and demand.
The combination lets us adapt per catalog, not per account average. A product priced above the market with a healthy margin gets repositioned, within the client's guardrails, so it can actually sell. A product that only sells below its cost at market price is deprioritized rather than fed more budget. A product that is both demanded and profitable at the current market price gets the push. The outcome is a catalog that sells what the market is asking for and what produces profit, at the same time.
-
check_circle
Competitor price monitor ยท Market prices for the same products, tracked live and fed into both bidding and the store's price rules.
-
check_circle
Per-product potential ยท Margin, demand and market position evaluated per product and per catalog, not as an account average.
-
check_circle
Margin guardrails ยท Price rules never cut below the minimum margin the client sets. The profit target is protected by design.
-
check_circle
Closed loop ยท Every change flows back into the conversion value the algorithms optimize on, so the system corrects itself.
What a full POAS implementation includes
-
check_circle
S2S tracking ยท Server-side conversion tracking, independent of browser cookies and ad blockers.
-
check_circle
Per-order margin ยท COGS, shipping, payment fees and discounts computed per order, in real time.
-
check_circle
API exports ยท Profit values pushed to Google Ads and Meta Ads as the conversion value.
-
check_circle
Cancellation retraction ยท Cancelled and returned orders automatically removed so algorithms learn from real profit.
-
check_circle
Competitor price monitor ยท Live competitor pricing feeds both bidding and the store's own price rules.
-
check_circle
Margin guardrails ยท Prices are never lowered below the client's minimum margin.
-
check_circle
EU data residency ยท Data on Hetzner servers in Germany, kept 540 days, never leaves the EEA or goes to third parties.
What it looks like in client accounts
POAS 8.26x on Google, ROAS 19.90x on Meta
Over 81,000 orders in 8 months, with profit as the bidding signal across both platforms.
+77.1% net profit
Net profit up 77.1% on a budget 58.9% higher. Spend grew, but profit grew faster, because bids followed margin.
+165.5% orders
Growth in orders on a catalog where price position and margin decide which fragrances get pushed.
Questions about POAS
Who introduced POAS optimization in Romania?add
Is POAS a product or a brand?add
How is POAS different from ROAS?add
Why is a margin calculator not enough?add
What does Google offer natively?add
What is price intelligence in this context?add
See how BigConvert applies POAS in a live account
The full methodology, the technical setup and the onboarding steps are documented on bigconvert.ro. If you want to check whether your catalog data allows profit-based bidding, book a call.
Back to the homepage POAS sectionPOAS (Profit on Ad Spend) is a performance marketing KPI, a metric we track and optimize on. It is not a brand or a product name we sell under. Google Ads, Conversions with Cart Data, Meta and Conversions API are trademarks of their respective owners.