How to Scale Paid Search Under a Hard ROI Floor

The short answer

When a board sets a hard ROI floor — "every dollar must return 1.5" — most teams treat it as a growth ceiling and stop pushing. That is almost always the wrong response, for two reasons. First, the floor is usually being applied to a number that is not trustworthy, so nobody actually knows whether the account is at the floor or comfortably above it. Second, a portfolio-level floor is the wrong rule for allocating each incremental dollar. Fix the measurement first, then govern with a marginal floor below the reported one, and you can grow profitably without ever breaching the mandate. I have done exactly this on an account the board believed was maxed out — it was not.

Step 1: Distrust the number the floor is applied to

A blended ROI of, say, 1.37 sounds like an account sitting just under a 1.5 floor. But before you accept that, audit the data feeding it. On the account I worked, five of six data files failed at least one validation check. Cleaning them changed the story completely:

Removing a few hundred duplicate booking rows lifted ROI materially.
Excluding an incomplete reporting tail (recent conversions that had not fully matured) lifted it again.
One undocumented cost line, when questioned, turned out not to belong to the account at all.

Cleaned up, the "1.37" was really 1.45–1.51. The account was at or near the floor, not below it — a completely different conversation with the board. The lesson: never accept a floor verdict on unvalidated data. The measurement fix is often worth more than any bidding change, because it changes what everyone believes is true.

Step 2: Fix what the bidder optimises toward

Here is the single highest-leverage change on most paid search accounts, and almost nobody makes it: stop letting Smart Bidding optimise toward gross conversion value when what you care about is net contribution.

On this account, Google's reported conversion value tracked gross booking value almost perfectly — but overstated net contribution by more than 10x, because margin varied enormously by product. Short rentals carried ~23% margin; long rentals ~7%. A single target-ROAS number cannot correct for that spread. The bidder was being paid to chase gross value and was therefore systematically mispricing the long-rental, low-margin inventory.

The fix is to feed the bidder a modelled net value per conversion instead of the raw platform value. This is the only change that improves every auction decision at once, rather than tweaking one segment. It is technical, it is unglamorous, and it is usually the difference between an account that scales profitably and one that quietly buys unprofitable volume.

Step 3: Separate the floor you report from the floor you allocate on

A 1.5 portfolio ROI floor is a sound governance threshold. It is a terrible per-dollar allocation rule.

Here is why. The portfolio floor protects the average. But the right decision for each incremental dollar depends on its marginal return. A dollar that returns 1.3 is still profitable — it still adds contribution — even though it is below the 1.5 average. Reject every sub-1.5 opportunity and you leave profitable growth on the table to protect an average that was never in danger.

The discipline I use: report against the 1.5 settled blended floor, but allocate new spend against a marginal floor of ~1.2–1.3. This protects the board's portfolio requirement while still capturing profitable growth below the headline number. On this account, applying 1.5 to every allocation decision would have forgone roughly $170K in annual contribution — profit sacrificed to a misapplied rule.

Step 4: Find where the money actually leaks (it is match type, not market)

When an account underperforms, the instinct is to rank markets — which country is dragging us down. On this account that was the wrong first cut. Within every market, the pattern was identical:

Exact match non-brand cleared the floor (~1.47).
Phrase hovered at breakeven (~0.99).
Broad match lost money everywhere (~0.44) and was the single largest spend block.

The drag was not geographic. It was structural: broad match was buying six different kinds of demand — high-intent city queries, off-category junk, informational traffic, competitor terms — all at one blended price. A large chunk of broad spend generated almost no conversions. Reallocating that waste into Exact lifted the blended return by ~0.1 with zero loss of volume, before adding a single dollar of new budget.

The lesson: diagnose by structure (match type, campaign type, product margin) before you diagnose by market. The averages hide the mechanism.

Step 5: Distinguish "budget-capped" from "rank-capped" headroom

When you are ready to grow, know why you are missing impressions, because the fix differs:

Lost to budget — the campaign would spend more if you let it. Fix: raise the budget. Direct and immediate.
Lost to rank — you are not winning the auction. Fix: improve bids or the value signal. Raising the budget does nothing.

Treating these as interchangeable is how teams pour budget into rank-capped campaigns and wonder why nothing changes. Separate them, and you know exactly which lever recovers which pool of demand.

The playbook, in order

Validate the data before accepting any ROI verdict. The floor may be applied to a fiction.
Feed the bidder net value, not gross conversion value. Highest-leverage single change.
Report on the portfolio floor; allocate on a lower marginal floor. Stop rejecting profitable growth.
Diagnose by structure first (match type, margin), then by market.
Separate budget-capped from rank-capped headroom and use the matching lever.

A hard ROI floor is not a growth ceiling. It is a discipline that, applied correctly, tells you exactly where profitable growth is hiding.

The bottom line

The brands that grow under strict ROI mandates are not the ones that spend more. They are the ones that fix what they measure, pay the bidder to chase the right number, and allocate on marginal return instead of a portfolio average. Every one of those is a decision, not a budget line.

If your paid search is stuck against an ROI target and you suspect the ceiling is really a measurement problem, that is the exact work of a Google Ads audit for ecommerce. Or run the DTC Paid Media Audit Checklist across your account first.

Frequently asked questions

Can you grow paid search without breaching a ROI floor? Usually yes. Report performance against the portfolio floor but allocate new spend against a lower marginal floor (around 1.2–1.3). A dollar returning below the portfolio average can still be profitable and additive, so applying the average to every decision forgoes real growth.

Why does my Smart Bidding overspend on low-value conversions? Because it is optimising toward gross conversion value, which ignores margin differences between products. Feeding it a modelled net-contribution value per conversion corrects the mispricing across every auction at once.

What is the highest-leverage fix on a paid search account? Usually the value signal you feed the bidder, followed by validating conversion data. Both change what the system optimises toward, so they improve every downstream decision — more than any individual bid or budget tweak.