Reading and improving ROAS honestly

0:00Welcome to the Leaders Insights Podcast.

0:07Today's episode, reading and improving rows honestly.

0:11Everyone quotes ROAS, return on ad spend, revenue divided by what you paid for ads.

0:17You say most marketers are lying to themselves with it.

0:20Prove it.

0:21Happy to.

0:22A team runs a meta campaign.

0:23The dashboard says four to one, $4 back for every dollar in.

0:28They celebrate.

0:29But half those buyers were already coming.

0:32Facebook took credit for a sale that would have happened anyway.

0:35That's not return.

0:36That's a receipt for demand you already had.

0:39So the numbers inflated.

0:40By how much realistically?

0:42On branded search, people literally typing your company name into Google.

0:47I've seen reported ROAS of eight or ten collapse to under one when you pause it and watch

0:52what actually disappears.

0:55Google's own dashboard will happily bill you for customers who knew your name before

0:59the ad loaded.

1:00You're saying the vendor's number is self-serving?

1:02Shocking.

1:03Of course it is.

1:05The platform is both the referee and the team.

1:07Meta, Google, TikTok.

1:10They grade their own homework and market generously.

1:13Anyone who takes their attribution at face value deserves the budget they're about to

1:17waste.

1:18Fine.

1:19So how do you build a ROAS you can actually trust?

1:21Three moves.

1:22First, strip incrementality.

1:25The sales that only happen because of the ad.

1:27Not the ones you'd have gotten regardless.

1:30You test it.

1:31Turn a channel off in a region.

1:32Watch what really drops.

1:34People hate turning off spend.

1:36It feels like setting money on fire.

1:38It's cheaper than lighting it on fire quietly for two years.

1:42Uber did this famously.

1:44Killed a chunk of app install advertising, something like $120 million annualized, and demand barely

1:50moved.

1:51Two-thirds of it was fraud or dead weight.

1:53That's the whole game in one story.

1:55So test one.

1:56G.O. holdouts.

1:57What's two?

1:59Fix your revenue number.

2:01Most teams count top line order value.

2:03You should count contribution margin, revenue minus the cost of the product, shipping,

2:08payment fees, the discount you gave.

2:11A four-to-one on gross revenue can be losing money once you subtract what the sale actually

2:17cost you.

2:18Give me the trap in numbers.

2:19While a $100 item, reported RO's four, means $25 spend.

2:25Sounds great.

2:26But your product costs $60.

2:28Shipping and fees $15.

2:30You gave a 10% coupon.

2:32You clear maybe $15 of margin on a $25 ad cost.

2:37You paid to lose money, and the dashboard threw a party.

2:39Brutal.

2:40And three.

2:41Pick a consistent window.

2:44Attribution window.

2:45How long after seeing an ad a sale still gets credited?

2:48Meta defaults to seven-day click plus one-day view.

2:52TikTok pushes longer windows because longer windows flatter TikTok.

2:56If your channels use different rulers, you're comparing nonsense and reallocating budget

3:01on it.

3:02Okay, so I've got an honest, deflated number.

3:04Now how do I actually move it up?

3:06You stop worshipping the blended average and go to the edges.

3:10Segment by new versus returning customers.

3:13Returning customer ROAS is often three times higher.

3:17You're paying to rereach people who already love you, feels efficient, does almost nothing

3:21new for the business.

3:23So the flattering segment is the useless one.

3:26Usually.

3:27The number that matters is new customer acquisition cost against what that customer's worth over

3:31their life.

3:33If a fresh buyer costs $40 and brings $200 over two years, a bad one-to-one first order

3:40ROAS is actually a bargain.

3:43Judging that campaign on day one revenue would get you fired for making the right call.

3:47You've just told marketers their best-looking metrics are the biggest liars.

3:52What's the pushback you get?

3:53My CEO wants one number by Monday.

3:56Fine.

3:57Give them one number, but make it the honest one.

4:00I'd rather report a blended two-to-one I'd bet my job on than a seven-to-one

4:04I know is fiction.

4:06The fiction always gets discovered, usually right when you've scaled spend on it.

4:10Give me the one thing to do tomorrow morning.

4:12Run a geo holdout on your single biggest channel.

4:16Turn it off in two comparable regions for three weeks.

4:19Keep it on everywhere else.

4:21The gap between what the dashboard claims and what actually drops is your real ROAS.

4:26Everything else is decoration until you know that number.

4:28That's a wrap.

4:30Fresh CMO briefings drop daily at MBA-training.com.