By the Lenoretech SEO Strategy Team · Reviewed by a senior SEO strategist · Last updated: June 2026
If your in-platform ROAS looks fine but your bank balance disagrees, the problem is almost never the creative. It is usually a measurement gap, an attribution overlap, or an optimisation target that is quietly working against your actual margin. Below are the 11 mistakes we see most across audited accounts, each paired with the exact number that reveals it.
1. Trusting in-platform ROAS over blended ROAS
Meta says 4.2x. Google says 5.1x. You add them up, feel great, and still miss your profit target. That is because both platforms claim the same conversions. The metric that exposes this is blended ROAS: total revenue divided by total ad spend across every channel, pulled from your own finance data rather than the ad dashboards. We routinely see accounts where the summed in-platform ROAS is 4.5x while blended ROAS sits at 2.1x. That gap is your real-world performance, and it is the only ROAS number worth reporting to a founder or CFO.
2. Attribution double-counting across channels
When a user sees a Meta ad, then clicks a branded Google search, then converts, both platforms book the sale. Multiply that across thousands of conversions and your reported revenue can exceed your actual revenue by 30 to 60 percent. The exposing metric is platform-reported conversions divided by Shopify, GA4 or CRM actual orders. If that ratio is above 1.3, you are double-counting and scaling budget on phantom returns. We audited a fashion brand reporting 1,840 monthly purchases across Meta and Google while their order management system showed 1,190. They had been increasing spend for two quarters against revenue that did not exist.
3. Optimising CPL while CAC silently climbs
Lead-gen teams celebrate a falling cost-per-lead, but cheaper leads are usually worse leads. Lead-to-sale rate drops, the sales cycle lengthens, and your true customer acquisition cost rises even as CPL falls. Track CPL and CAC on the same chart. We audited a B2B account where CPL dropped 22 percent over a quarter while CAC rose 41 percent, because broad-match expansion pulled in unqualified traffic. The dashboard looked like a win. The P&L did not.
4. Ignoring new-customer ROAS
Blended ROAS can stay flat while you slowly become a remarketing machine that pays to acquire purchases you would have got for free. The metric is new-customer ROAS: first-time buyer revenue divided by total spend. If blended ROAS is 3x but new-customer ROAS is 1.1x, your paid budget is mostly harvesting existing demand, not growing the business. In one supplements account, 71 percent of "conversions" attributed to paid were repeat buyers already on an email flow. Once we re-scoped budget to first-time acquisition only, headline ROAS dropped from 3.4x to 1.9x, but actual new revenue per rupee spent climbed within six weeks because the spend was finally pointed at growth instead of harvesting.
5. Reporting ROAS on revenue instead of margin
A 3x ROAS on a product with a 25 percent contribution margin loses money. The fix is to track MER against contribution margin, then set a break-even ROAS from that same margin. The formula is break-even ROAS = 1 divided by contribution margin. Use contribution margin, not gross margin, so the number nets out shipping, payment fees, returns and the variable cost of fulfilling the order, all of which scale with each extra sale paid media buys. At a 25 percent contribution margin you need above 4x just to break even before overheads. Most teams never calculate this single number, so they scale campaigns that are structurally unprofitable.
6. Letting tCPA and tROAS bidding learn on broken signals
Smart bidding is only as good as the conversion data feeding it. A misfiring pixel, a deduplicated event, or a server-side tag passing the wrong values teaches the algorithm to chase the wrong users. The exposing metric is a comparison of platform-recorded conversion value against your back-end value, event by event. A consistent gap means the bidder is optimising toward inflated or phantom value, and no creative test will fix that. We have seen a single hardcoded default value (every purchase logged at the same amount) quietly train an algorithm for months to favour low-value buyers.
7. Scaling budget faster than the auction can absorb
Jumping budget 100 percent overnight resets learning and pushes you into more expensive auction inventory. CPMs spike, frequency climbs, and ROAS craters even though nothing about your offer changed. Watch marginal ROAS: the return on each incremental block of budget, not blended ROAS. When marginal ROAS drops below your break-even line, you have found the efficient spending ceiling for that audience. Our paid teams typically scale in 20 to 30 percent steps to keep marginal returns above break-even.
8. Counting branded search as performance
Branded campaigns post huge ROAS because the user was already going to buy. Folding them into your blended number masks weak prospecting. Segment branded ROAS from non-branded ROAS. If pausing branded barely changes total orders, that spend was defending traffic you already owned, not generating new revenue. In one home-goods account, branded search reported a 14x ROAS that propped up the whole portfolio average; a two-week pause test moved total orders by under 4 percent, proving most of that "return" was demand the brand captured for free through organic and direct. This is one of the fastest profit leaks to find and the easiest to argue about internally, because someone always wants to keep the shiny number.
See our performance marketing services or book a free audit →
9. Optimising for the click instead of the cohort
A campaign with a great 7-day ROAS can have terrible 90-day economics if it attracts one-time discount hunters. The metric that matters is LTV-to-CAC ratio by acquisition cohort. Healthy paid acquisition usually sits at 3:1 or better within 12 months. If your discount-led campaigns sit at 1.4:1, you are buying revenue that never repeats, and that shows up in ecommerce accounts constantly.
10. Treating landing pages as someone else's job
You can have flawless targeting and still bleed ROAS at a 1.1 percent conversion rate when a 2.2 percent page would double returns at the same spend. The exposing metric is conversion rate by landing page versus ad CTR. High CTR with low conversion rate means the ad oversold and the page underdelivered. This is where conversion-focused web design often returns more than any bidding tweak, because the math compounds across every campaign feeding the page.
11. Running paid in a vacuum, ignoring organic lift
When paid and organic are measured separately, you cannot see that strong SEO is quietly subsidising your paid ROAS, or that aggressive paid bidding is cannibalising free clicks you would have won anyway. The metric is incrementality, measured via geo holdout or conversion-lift tests. A simple two-region test (paid on in one matched market, off in another) tells you how much of your reported ROAS is genuinely incremental versus demand you would have captured for nothing.
How to fix this without guesswork
The pattern across all 11 is the same: the dashboard is optimising for a number that does not match your bank account. Start by reconciling platform-reported revenue against your actual finance and CRM data. Then rebuild reporting around three honest numbers: blended ROAS, new-customer ROAS, and break-even ROAS on contribution margin. Once the measurement is trustworthy, the rest of the work finally moves real profit, whether that is creative, bidding, or audience structure. If you want a second set of eyes, our team runs structured audits as part of our PPC management service, and we will show you the exact leak before recommending a single budget change. Tell us your numbers on the contact page and we will tell you which of these 11 is costing you the most.