The problem with one blended ROAS number
On the ad platforms, brand campaigns and performance campaigns look like the same thing. Both are paid media. Both drive traffic. So most founders manage them as one budget. One number judges it: the blended Return on Ad Spend (ROAS) on the agency report.
But they are two different jobs. asks for the sale today. Paid search. Retargeting. Shopping ads. Its value shows up as tracked sales within days. builds trust over months. Video. Sponsorships. Influencers. Its value shows up later. More people search your name. Clicks on everything else get cheaper.
Blend them into one number and you get the wrong answer both ways. The brand spend drags the blend down. So you cut brand ads that may be quietly working. Or the performance side props the blend up. So weak campaigns hide inside a healthy-looking average. One number. Two wrong calls.
1. An example showing you the numbers
Your store sells wine and spirits direct to customers. A typical order is $100. A good bottle, or a two-bottle set. You spend $1,000,000 a year on paid media. Paid social, paid search, and retargeting. Plus video ads and influencer deals.
Start with your . The five non-ad costs eat $70 of every $100 order. That leaves $30 to pay for ads. So ads break even at $100 ÷ $30 ≈ 3.3x. Your agency reports a blended 3.3x. The dashboard’s verdict: a million dollars of ads, earning nothing.
Now split the same spend into its two jobs and look again.
$1,000,000 of ad spend, the blended number vs the split view
Annual figures, whole dollars. Break-even ROAS on this store is 3.3x.
| Line item | Blended view | Performance only | Brand only |
|---|---|---|---|
| Ad spend | $1,000,000 | $800,000 | $200,000 |
| Tracked sales | $3,300,000 | $3,200,000 | $100,000 |
| Agency ROAS | 3.3x | 4.0x | 0.5x |
| Ad cost per $100 order | $30 | $25 | $200 |
| Verdict vs 3.3x break-even | ‘ads earn nothing’ | earns $5 per order | wrong metric entirely |
The blend says break-even. The split says two different things. Performance is truly working. At 4.0x, the ads cost $25 of each $100 order. The order leaves $30. So $5 of profit remains. Across 32,000 orders, that is $160,000 a year. That is a case for scaling further. Not for standing still.
The brand side shows 0.5x. It looks like a disaster. It is not. It is the wrong ruler. Brand ads are not built to win a tracked sale this week. Their value lands later. More branded searches. More sales from people who already trust you. Cheaper performance clicks. Judge them on brand-tracking data over 6 to 12 months. Not on a 7-day window.
One honest note. Brand advertising really is harder to measure. Branded search, recall surveys, and direct traffic are slow and noisy. A ROAS dashboard is fast and clean. That is why blending is tempting. The blended number is easier. Easier is not the same as right.
The sentence that changes how you think about ad budgets
Performance ads must pay for themselves now. Brand ads pay you back later. One number cannot judge both.
Split the spend into two budgets. Measure performance against your break-even ROAS every week. Measure brand on brand-tracking data every quarter. A blend hides your winners and shoots your slow-burners.
2. How to split brand and performance in your ad account
One week of work once you decide to do it. The output: two budgets, two yardsticks, two review rhythms.
- Tag every campaign by its main job. Performance: direct sales, conversions, shopping ads, retargeting. Brand: reach, video views, awareness, ads that court brand-new shoppers. Some campaigns sit on the line. Tag them by the main job they were built to do.
- Make the split a choice, not an accident. Add up what each bucket spends today. Young stores usually run heavy on performance, around 90/10. Grown brands drift toward 70/30. There is no right split for everyone. But an accidental split is always wrong.
- Give each budget its own yardstick. Performance: agency ROAS against your break-even, 3.3x on this store. Check it campaign by campaign. Brand: branded search growth, recall surveys, direct traffic. And watch whether your performance clicks get cheaper over time. Never put both on one report line.
- Set the brand budget as a share of revenue, not a share of the ad budget. Brand is a long-term bet. So it should grow with the size of the business. A common range is 3 to 8 percent of revenue. Tie it to the performance budget instead, and it jumps around with every weekly tune-up.
- Review each on its own clock. Performance weekly, against break-even. Brand quarterly, against the brand-tracking goals you set. Weekly reads on brand data are mostly noise. A quarterly trend is a signal. Revisit the split itself once a year.
3. One warning before you act
None of this makes brand spend worth its price by default. Some brand campaigns are simply waste. Badly aimed video. Sponsorships that do not fit. Influencers with the wrong audience. The rule is not ‘defend all brand spend because it is brand.’ The rule is ‘judge brand spend on brand numbers.’
So hold brand to account on its own ruler. Give it 6 to 12 months. Branded search flat? Recall not moving? Performance clicks not getting cheaper? Then the brand campaigns are failing, fairly measured, on their own numbers. Cut or rebuild them. Exactly as you would a losing performance campaign.
4. Frequently asked questions
What share of my budget should go to brand?
It depends on your stage. Early brands usually put 85 to 95 percent into performance. They need sales now to pay the bills. Grown brands with real recognition often move 25 to 40 percent into brand. Whatever you choose, choose it on purpose. The split should be a decision. Not a leftover.
How do I measure brand ads without tracked ROAS?
Three signals. Branded search volume: are more people typing your name into Google or Amazon? Aided recall: in regular surveys, do more target customers know you? And knock-on savings: is your Cost of Acquisition (CAC) on performance campaigns falling as more people know you? Slow signals. But honest ones.
Should brand campaigns show any tracked ROAS at all?
Some, but small. A few viewers click through and buy inside the tracking window. So brand campaigns often track at 0.3x to 0.8x. Treat that as a bonus. Not the point. Defending a brand campaign by its tracked ROAS? You are back to the wrong ruler.
How long before I judge the brand budget?
Six months at least. Twelve is a fairer read. Brand moves slowly. Short windows give you noise, not verdicts. Cannot commit to six months of steady spend at a real level? Then skip brand advertising for now. You would be buying data too noisy to use.
5. Quick reference: what to avoid and what to apply
What to avoid
- Managing brand and performance as one budget with one blended ROAS.
- Cutting brand campaigns because a 7-day dashboard shows 0.5x.
- Letting weak performance campaigns hide inside a healthy-looking blend.
- Setting the brand budget as a leftover of the performance budget.
- Judging brand data weekly, it moves in quarters, not days.
What to apply
- Tag every campaign as performance or brand by its main job.
- Measure performance against your break-even ROAS, campaign by campaign.
- Measure brand on branded search, recall, and cheaper clicks over time.
- Set the brand budget at 3 to 8 percent of revenue, on purpose.
- Review performance weekly, brand quarterly, and the split yearly.
Want your six costs mapped for you?
Our free 31-Point Profit Diagnostic puts a dollar figure on where your profit is leaking. Or get a quote for done-for-you eCommerce accounting.
Definitions, modeling notes & rate-basis disclosures
Definitions
- The six profit levers
- (1) Discounts, (2) Refunds, (3) Product cost (landed COGS), (4) Sales channel and payment fees, (5) Shipping and 3PL fees, (6) Advertising spend.
- Return on Ad Spend (ROAS)
- Sales divided by ad spend. A blended ROAS mixes every campaign into one number, whatever its job.
- Break-even ROAS
- The ROAS level where ads exactly pay for themselves. The math: 100 divided by the share of the selling price left after the five non-ad costs. On this store, 100 ÷ 30 ≈ 3.3x.
- Performance advertising
- Paid media built to win a tracked sale now: paid search, retargeting, shopping ads. Judged against break-even ROAS.
- Brand advertising
- Paid media built to make more people know and trust you over months: video, sponsorships, influencers. Judged on brand-tracking data.
- Branded search volume
- How many people search for your brand by name. An early sign that brand advertising is working.
- Aided recall
- The share of surveyed target customers who know your brand from a list. It moves slowly. Regular surveys measure it.
- Contribution per order
- Selling price minus all six costs. What one sale really leaves behind.
Modeling notes
- This article uses the series’ standard teaching store: a $100 product, product cost $40, a standing 10% discount, shipping and 3PL $12, refunds 5%, and payment and channel fees 3%. That leaves $30 per order for advertising and a break-even ROAS of 100 ÷ 30 ≈ 3.3x. Only the advertising lever varies here.
- Performance: $800,000 of spend at 4.0x = $3,200,000 of tracked sales = 32,000 orders. Ad cost per order $100 ÷ 4 = $25; contribution $30 - $25 = $5; 32,000 × $5 = $160,000 per year.
- Brand: $200,000 of spend at 0.5x tracked = $100,000 of sales = 1,000 tracked orders. The tracked window is, on purpose, not the yardstick for this budget.
- Blended: $3,300,000 ÷ $1,000,000 = 3.3x. Check: 33,000 orders × $30 = $990,000 of non-ad contribution against $1,000,000 of spend, about break-even (-$10,000, inside the rounding of the true 3.33x).
Rate-basis disclosures
- Baseline: Direct-to-Consumer wine and spirits brand on Shopify, $100 average order value, $1,000,000 annual paid media.
- Split: 80% performance ($800,000 at 4.0x agency ROAS), 20% brand ($200,000 at 0.5x tracked).
- Non-ad variable costs: 70% of the selling price (product 40%, discount 10%, shipping and 3PL 12%, refunds 5%, fees 3%).
- Working benchmarks: tracked ROAS on brand campaigns 0.3x-0.8x; brand budget 3-8% of revenue; brand review windows 6-12 months.
- All figures in whole dollars; break-even ROAS 3.33x rounded to 3.3x.