Why this matters
Retailers with a wide price range often run one target across the account, copied from last quarter's average, which hides loss-making product groups behind profitable ones. The target CPA versus target ROAS comparison sets out when each target type fits and why either should come from margin, and the break-even ROAS definition gives the floor beneath any return target. Here you apply both to one account where the right answer differs by campaign.
The retailer and its margins
Lensmoor Eyewear is an invented online retailer of prescription glasses, designer frames and contact lens subscriptions. Its finance lead has worked out contribution margin for each group after product, glazing, delivery, payment fees and expected returns.
Lensmoor Eyewear: product groups and margins
| Product group | Typical order value | Range of order values | Contribution margin |
|---|---|---|---|
| Own-brand glasses | £110 | £60 to £190 | 55% |
| Designer frames | £260 | £180 to £420 | 25% |
| Contact lens subscription | £24 per monthly box | Almost identical | £9 margin per box |
Profit rule for glasses and frames: keep at least 10 percentage points of revenue after ad cost.
Profit rule for subscriptions: the owner's cash-flow policy says advertising for a new subscriber must be repaid by the margin on the first three boxes.
Home try-on: customers can request five frames to try at home for free. From Lensmoor's own order records, 35% of try-on requests become an own-brand order averaging £95, and each try-on costs £6 to send and restock.
Lensmoor Eyewear: campaign results over eight weeks
| Campaign | Current bidding | Spend | Conversions | Conversion value | Reported ROAS |
|---|---|---|---|---|---|
| Shopping – Own-brand glasses | Manual CPC | £6,000 | 240 orders | £26,400 | 4.4 |
| Shopping – Designer frames | Manual CPC | £5,500 | 60 orders | £15,600 | 2.8 |
| Search – Contact lenses | Manual CPC | £3,600 | 150 subscriptions | not tracked | n/a |
| Search – Home try-on | Manual CPC | £2,400 | 200 requests | not tracked | n/a |
Shopping order values are passed to the platform and match the order system. Search campaigns pass no value.
Choosing the target type
Three questions decide the target type for each campaign. Do conversions differ meaningfully in value? Is that value tracked accurately at the moment of conversion? And is there enough conversion volume for automated bidding to learn from, which the manual versus automated bidding comparison covers? A campaign that passes all three can bid to a return target; one whose conversions are worth about the same, or whose value arrives weeks later, bids to a cost target derived from what a conversion is worth. Where current results sit far from the economic target, the starting target is a step towards it, not the destination.
Task
- Calculate the break-even ROAS and the target ROAS for own-brand glasses and for designer frames.
- Calculate each Shopping campaign's contribution after ad cost over the eight weeks.
- Work out the maximum and target cost per acquisition for a contact lens subscriber and for a home try-on request.
- Choose a target type for each campaign and give the reason in one sentence.
- Set a starting target for each campaign and say how you would move it over the following weeks.
Worked answer
The arithmetic first, using the break-even calculator or by hand.
Lensmoor's targets and contribution
| Product group | Break-even ROAS | Target ROAS |
|---|---|---|
| Own-brand glasses (55% margin) | 1 ÷ 0.55 ≈ 1.8 | 1 ÷ (0.55 − 0.10) ≈ 2.2 |
| Designer frames (25% margin) | 1 ÷ 0.25 = 4.0 | 1 ÷ (0.25 − 0.10) ≈ 6.7 |
Contribution after ad cost:
- Own-brand Shopping: £26,400 × 0.55 = £14,520 margin, less £6,000 spend = £8,520.
- Designer Shopping: £15,600 × 0.25 = £3,900 margin, less £5,500 spend = a loss of £1,600.
Cost targets:
- Contact lenses: maximum CPA = 3 × £9 = £27. Current CPA = £3,600 ÷ 150 = £24. Target £22, leaving room under the cash-flow limit.
- Home try-on: value of a request = 0.35 × (£95 × 0.55) − £6 ≈ £18.30 − £6 = £12.30 maximum CPA. Current CPA = £2,400 ÷ 200 = £12. Target £10.
Starting targets:
- Own-brand Shopping: target ROAS 3.8, stepped down towards 3.0 if marginal return holds above 2.2.
- Designer Shopping: target ROAS 4.5, held while volume settles; pause any brand whose margin cannot reach 4.0.
- Contact lenses: target CPA £24, moved to £22 after the learning period.
- Home try-on: target CPA £12, moved to £11 and then £10.
Own-brand glasses: target ROAS. Order values vary widely, values are verified and the volume is healthy, so a return target lets bidding favour the larger baskets with lens upgrades. It earns well above its target, so start just below the trailing return to protect volume through the switch, then loosen in steps while the marginal return, not the average, stays above target.
Designer frames: target ROAS, and a smaller campaign. Values vary and are tracked, so the target type is return, but the campaign reports a ROASRevenue generated per unit of ad spend. below its floor and loses money on every attributed order. Here a target above the trailing figure is the point, because lower spend is the intended outcome. Set it above break-even, accept the volume drop, and pause brands whose margin cannot clear the floor. The thin order count means predictions will swing, so keep watching after the learning period.
Contact lenses: target CPA. Every subscription is worth the same when it converts, so a return target adds nothing. The current CPACost of each conversion. sits under the cash-flow limit, so start there and tighten once.
Home try-on: target CPA now, value later. The request has no value when it happens. A cost target derived from the order rate is honest today; importing try-on orders with values would allow a return target later.
Common mistakes
- One target across the account. The blended return looks healthy while designer frames lose money.
- Setting the target from the trailing average. It would lock designer frames into a loss.
- Using order value, not margin, for try-on. It overstates what a request is worth.
- Forcing a return target on subscriptions. Identical values give value bidding nothing to work with.
Next, test your judgement in the targets and bidding checkpoint, or return to the intermediate performance marketing level.
