Share
Why Your ROAS Is Lying to You | eCommerce’s Biggest Blind Spot
Welcome to From Clicks to Customers, a three-part series from Klaviyo and Adcore built around one idea: the brands that win Q4 are the ones that stop running paid media and lifecycle marketing as two separate jobs. This is Post 1 of 3.
Picture two eCommerce brands: same category, same monthly spend on Google and Meta, same average order value, same target ROAS. On paper, they should perform similarly. Often, they won’t, and the difference usually isn’t creative, bid strategy, or budget.
Table of Contents
What your ad platforms actually know about your customers
Google and Meta are exceptional at delivery, finding the person most likely to click or convert, in real time, at scale. But their intelligence stops at the edge of their own platform. They don’t know that the shopper who just “converted” returned the item a week later, or that another customer has a support ticket open over a bad delivery experience and shouldn’t be getting a “come back and shop again” ad. They definitely don’t know that someone who bought a walnut bed frame doesn’t want to see birch nightstands next, but without better signal, that’s exactly the ad they’ll get.
That gap gets filled one of two ways. Either a brand feeds its ad platforms a richer signal, flagging VIPs, at-risk churners, recent purchasers to exclude from prospecting, and best customers worth cloning into a lookalike audience, or the algorithm falls back on blunt data like “anyone who’s ever visited the site.” Same platforms, same budget, often different ROAS: not because one brand’s campaigns are built better, but because one brand feeds the machine better inputs. ROAS as reported inside Google Ads or Meta Ads only tells you how a campaign performed against the audience it was given, not whether that audience was the right one.
What richer signal actually does to performance
This isn’t theoretical. When Adcore tightened audience inputs for one retail client – sharper segmentation, owned channel data fed back into the platform, creative aligned to where each customer actually sat in the funnel – spend dropped 24% and impressions climbed 118%. Same budget, same platform, different information going in.
Most brands haven’t touched that lever yet. Email, SMS, and purchase history are among the richest audience signals available – and they’re the ones most paid campaigns still aren’t using.
The part your ad platform can’t see
What happens to a customer after they convert usually lives somewhere else entirely – in the systems tracking email, SMS, purchase history, and lifecycle behaviour. Whether they come back for a second order or what their real lifetime value turns out to be – that data rarely makes it back into the platform that acquired them. Tracing that full path changes how you’d judge a campaign, relative to trusting reported ROAS in isolation.
To be clear: Google Ads and Meta Ads remain the system of record for spend, delivery, and platform-level performance. Nothing about closing this loop replaces that. What changes is the layer running alongside them – pushing sharper, customer-data-informed audiences in for suppression, retargeting, and lookalikes, while pulling funnel and revenue data back out. That loop runs both ways. Consent captured through Meta and Google lead ads should land back in the same customer record, not sit stranded in a platform export nobody checks.
What to watch instead of ROAS: revenue per customer
The fix isn’t a fancier attribution model, it’s a different question. Instead of asking what a campaign’s conversions cost, ask what those customers are actually worth over the next 60 to 90 days. Revenue per customer, run alongside day-one ROAS, is the number that answers it.
A campaign with mediocre 30-day ROAS but strong 90-day customer value is often a better outcome than one with strong ROAS and retention that stops at the first order, the full-funnel view ROAS alone can’t give you. Part 2 of this series covers the specific audience moves; Part 3 turns revenue per customer into the metric that keeps paid and email pointed at the same outcome across BFCM.
Why this matters heading into BFCM
Klaviyo’s BFCM 2025 data shows shoppers who received both email and text messages placed 11% more orders, added 34% more items to cart, and viewed 71% more products than single-channel shoppers, while email revenue grew 15% year-over-year and text grew 25% (Klaviyo, “How cross-channel orchestration and the rise of text messaging shaped BFCM 2025”). Adcore’s own client data points the same direction: accounts running on richer, customer-informed audiences have consistently outperformed accounts running on platform defaults, well before BFCM volume even hits. BFCM raises the stakes further: competition for the same audiences drives media costs up, and the customers acquired over BFCM weekend set the size of the repeat-customer base a brand carries into the next year. Getting audience quality and revenue-per-customer measurement right before Q4 ramps up compounds, getting it wrong compounds too, in the wrong direction.
Three things to check in preparation
These three checks line up with Phase 1, Audit + Prepare, the first of four phases in the 2026 BFCM Calendar Adcore and Klaviyo built together, running from lots of post-BFCM audit work.
- Audit what audiences you’re actually sending to Google and Meta. Real segments, VIPs, churn-risk customers, high-LTV lookalike seeds, recent-purchaser suppression lists, or just “all visitors” and “all purchasers”.
- Check whether lead-ad subscribers are flowing back into your customer database. Consent collected through Meta and Google lead ads should land somewhere your team can actually follow up from.
- Map one full journey end to end, ad click, site behavior, placed order, and what that customer did afterward. If that means pulling data from three different tools, that’s the gap worth closing before Q4 spend ramps up.
The takeaway before Q4 planning starts
None of this means paid media is underperforming, or that Google and Meta’s own reporting should be ignored, it’s still the right place to judge delivery and platform-level performance. It means ROAS reported in isolation can only tell you about the audience you gave the algorithm, not whether it was the right one, or what happened to those customers afterward.
Brands that treat customer data and paid media as one connected system, feeding sharper audiences in, pulling revenue and lifecycle outcomes back out, go into Q4 with a clearer view of where their budget is actually working. Adcore has generated over $1.03B in revenue for clients running exactly this playbook. If any of the three checks above turned up a gap, it’s worth raising before Q4 planning locks in.
Get the full 2026 BFCM Calendar. See all four phases, Audit + Prepare, Ad Testing + Segmentation, Execution, and Retention + Reporting, mapped month by month. Download the calendar.
Thinking through your Q4 strategy? Adcore and Klaviyo work with ANZ retailers on exactly this kind of paid-and-lifecycle alignment. Get in touch with Adcore or explore Klaviyo’s platform.
Trending Articles
Get exclusive CMO tips that I only share with email subscribers.
Don't forget to share this article!
Back to top