New customer quality is the first PPC KPI
ROAS, MER and nCAC all matter. But they only make sense once you know what kind of new customer paid media is buying.
The default PPC reporting stack is upside down. It starts with platform ROAS, then tries to explain the business from there.
The better starting point is new customer quality. Paid media should be judged by whether it creates valuable new demand, not whether a platform can claim revenue from people who were already warm.
ROAS is not a growth metric. It is an attribution claim.
The hierarchy
For ecommerce and serious lead-gen, I would rank PPC KPIs like this:
- New customer quality: are we acquiring the kind of customer the business actually wants?
- nCAC: what does it cost to acquire a new customer or qualified new opportunity?
- MER: what is the blended relationship between total revenue and total marketing spend?
- ROAS: what does the platform claim happened inside its attribution window?
That order matters. A strong ROAS number can still be commercially weak if the campaign is harvesting brand demand, returning customers or low-quality first purchases.
What new customer quality means
New customer quality is not one metric. It is a judgement about the customers being acquired.
For ecommerce, useful signals include first product purchased, gross margin, refund risk, repeat purchase rate, order count, cohort revenue and whether the purchase came from brand, non-brand, Shopping, Performance Max, paid social or email overlap.
For lead gen, the equivalent is whether the enquiry became a qualified opportunity, booked call, accepted quote, sale, retained client or high-fit pipeline. A cheap form fill is not a quality new customer signal.
Retention matters, but PPC does not own it
Retention is important. It just is not the primary operating job of the PPC account.
The CRM, email, product, sales and customer success teams own the work of turning a customer into a repeat customer. PPC owns the quality of the customers it introduces to the business.
That does not mean PPC can ignore retention. It means retention should be used as a quality signal for acquisition cohorts, not as a hiding place for weak acquisition. If a campaign buys customers who never come back, that matters. If the CRM team fails to nurture good customers, that is a different operational problem.
Brand, non-brand, new and returning must split at scale
When spend is small, you can sometimes tolerate blended reporting. When you want to scale, blended reporting starts lying by omission.
Brand and non-brand do different jobs. Brand captures and protects existing demand. Non-brand has to prove it can create or intercept demand profitably. New and returning customers also have different economics. Returning customers may produce high ROAS because they already know you. That does not mean the campaign created the demand.
At scale, split reporting at least four ways:
- brand new customers
- brand returning customers
- non-brand new customers
- non-brand returning customers
ROAS will differ across those groups, and it should. The mistake is blending them into one number and pretending it explains growth.
Why in-platform ROAS is not truth
Google Ads and Meta are not lying in the simple sense. They are answering a narrower question: which conversions can this platform claim under its attribution rules?
That is not the same as asking whether spend created incremental profit. Platform ROAS can include warm users, brand demand, returning customers, remarketing influence, view-through behaviour, email halo effects and purchases that would have happened anyway.
There is also a time-window issue. Google Ads conversion reporting depends on the conversion window you choose, and conversions can be reported up to 90 days after the click. That is not lifetime value. It is an attribution window.
Use in-platform ROAS for what it is good at: operating inside a channel. Do not use it as the source of commercial truth.
The useful metric stack
The spend stress test
A useful question is simple:
If spend increases by 20%, does total revenue, new customer count, margin or qualified pipeline move meaningfully?
This is not perfect. Seasonality, stock, promotions, lag, competitor behaviour, pricing, margin mix and CRM activity all matter. But it is much closer to truth than staring at platform ROAS in isolation.
If paid media spend rises and the only thing that improves is attributed revenue inside the platform, be suspicious. The platform may be harvesting demand that already existed.
Use judgement
The framework is good, but it is not a mechanical scoreboard. You still need judgement.
A high nCAC can be acceptable if the first-purchase product creates strong repeat customers. A low nCAC can be bad if the customer refunds, churns, never repeats or wastes the sales team's time. MER can look weak during a deliberate growth push. Platform ROAS can still be useful when comparing similar campaigns with similar attribution exposure.
The point is not to worship a new dashboard. The point is to stop letting one platform-claimed ROAS number decide whether growth is real.
Operating rule
For a scaling account, the reporting should answer four questions:
- Are we buying new customers?
- Are those new customers any good?
- What does it cost to acquire them?
- Does total business performance move when spend changes?
If the dashboard cannot answer those questions, the account is not ready to scale confidently.
Sources
Need better PPC reporting?
If your dashboard cannot separate brand, non-brand, new customers and returning customers, it is not ready for scale.