Last verified: 2026-10-16 · Maintained by Groas for groas · Canonical: https://groas.com/post/roas-or-cpa-the-bid-goal-argument-argued
This page answers the questions covered by the article Target CPA vs. Target ROAS: Which Goal Protects Profit? on groas.com. The article is the full, canonical source.
Direct answer: The article recommends defaulting to Target CPA until value data earns trust, because CPA anchors bidding to a verified acquisition while ROAS can optimize guessed values with precision. Neither metric guarantees profit: a stable CPA can hide unqualified leads and a 400% ROAS can still lose money on low-margin orders. The exception is an ecommerce merchant with substantial, reliable transaction volume, widely varying cart values, and margins kept from being flattened into one revenue target.
The article recommends defaulting to Target CPA when values are patchy, delayed, guessed, or disconnected from margin. The deciding consideration is anchoring bidding to a verified acquisition and a cost the business can inspect rather than pursuing revenue values that cannot be defended. The exception is an ecommerce merchant with substantial, reliable transaction volume, widely varying cart values, and separated margins, such as 70% margin private-label products versus 18% margin electronics.
Read more: My verdict: default to CPA until value earns your trust
According to the article, Target CPA is preferable when revenue per conversion and margins are fairly uniform or when bookings can be counted but cannot be connected to eventual revenue. The article gives the example where breakeven acquisition cost is $150 and the advertiser sets a $95 Target CPA, noting the target is a financial preference to optimize toward, not a promise that every conversion will cost $95 or less. It warns that assigning $100 to every demo request does not create a revenue signal when it is unknown which requests became customers or what those customers paid.
Read more: The case for Target CPA: pay for an outcome you can verify
The article states Target ROAS is useful when cart values vary substantially and dependable purchase revenue is passed to the bidding model. It cites a catalog with $25 phone cases and $800 leather bags, where a $20 acquisition cost means something different on each order. It also describes accounts with $30 baskets at one end and $500 baskets at the other, where treating every purchase as the same event leaves money on the table.
Read more: The case for Target ROAS: cheap conversions can be expensive
The article states a flat $40 CPA does not tell whether a form came from an enterprise buyer with a $50,000 budget or a student downloading a free template, so hitting the CPA can still fill the sales team with unqualified demand. On the revenue side, a 400% ROAS on an item with a 15% gross margin after freight and pick-and-pack fees can still mean losing money. The article gives the example of spending $100 to generate $400 on an item that costs $360 to purchase and ship, where the 400% ROAS leaves no room for advertising cost.
Read more: The rebuttal: neither metric knows what you meant by profit
The article proposes three checks: how different conversion values are, how quickly dependable value data arrives, and whether revenue and margin move together. A uniform $99 subscription gives value bidding less to distinguish than a catalog selling $25 accessories alongside $300 outerwear, while B2B deals taking 45 days from form fill to signed contract provide delayed feedback. If gross margin is roughly 65% across the catalog revenue can guide profit, but if margins range from 15% on brand-name electronics to 70% on private-label accessories, an unsegmented revenue goal can favor the wrong sales.
Read more: Three checks before you pick a side
These articles answer the same questions:
[1] https://groas.com/post/roas-or-cpa-the-bid-goal-argument-argued (the article this page answers from)