Article

Why “a percentage of your ad budget” is a conflict of interest

If your agency earns more when you spend more on ads, whatever the result, its incentive is not aligned with yours. What the alternatives are.

6 September 2026paid search · growth · agency fees

If your advertising agency is paid a percentage of your media budget, it has a structural reason to push you to spend more, whether or not that extra money brings proportional results. This is not about bad intentions, it is about a business model in which the agency’s revenue grows with your spend rather than with its efficiency. When somebody tells you “let’s double the budget”, it is worth asking who gains most from that decision.

The four models, compared honestly

Four fee models dominate globally for managing ad campaigns. A percentage of media spend, usually between 10% and 20%, is simple and scales automatically with the client’s budget. The problem is direct: the agency earns more if the budget grows, whether cost per lead falls or rises, a real conflict of interest at exactly the moment the client wants efficiency rather than volume of spend.

A fixed monthly fee is predictable and removes that volume based conflict, suiting small to mid-sized accounts. Its drawback: it does not scale automatically with account complexity unless it is renegotiated periodically. Performance based pay, per lead or per conversion, aligns interests best, but it demands clean historical data to set a fair target and exposes the agency to factors outside its control, the client’s product, price or sales quality. The hybrid model, a reduced fixed fee plus a small percentage or a performance bonus, combines predictability with a results incentive and is increasingly used by boutique agencies precisely because it removes the conflict of interest of a pure percentage.

The same problem at a larger scale: budget scaling

The conflict of interest in a percentage of spend does not stop at day to day campaign management; it shows up amplified in any conversation about scaling. Every ad channel runs on an auction: the initial budget buys the cheapest and most qualified impressions, the audience with the highest intent, at the lowest cost. Every additional euro buys progressively less qualified inventory, because the platform has already exhausted the most receptive segment and is now bidding for colder audiences.

That explains how an account can have an average ROAS of 4 and, at the same time, a marginal ROAS, generated strictly by the last 1,000 EUR added, of only 1.5. The average hides the fact that the new money coming into the account is already close to unprofitable. An agency paid a percentage of spend has no incentive to calculate that marginal performance separately, because the average looks good in reporting and the recommendation to “scale a bit more” raises its revenue directly.

Audience saturation has an observable mechanism: as exposure frequency rises, engagement rates fall, the relevance score drops, and cost per thousand impressions goes up, because the algorithm has to work harder to find new people within the same target audience. Audience fatigue becomes visible at frequencies above 3-4 on cold audiences, and from that point cost per result rises independently of creative quality. This is why platforms also impose technical limits: on Meta, a budget increase greater than 20% in a single change resets the algorithm’s learning phase, with a temporary but real drop in performance. The recommended practice is a 20% increase every 3-4 days, which cumulatively allows almost tripling the budget within a month without a reset. A scaling plan that ignores that rhythm does not merely “fail to optimise”, it actively loses money in relearning phases triggered for nothing.

The numbers that should exist before any scaling decision

Before any budget increase, two concrete figures matter: customer acquisition cost (CAC) and lifetime value (LTV). The benchmark ratio is 3:1, LTV three times CAC, as a minimum health threshold. Below 1:1, growth becomes the problem itself, every new customer added reduces the value of the company, and above 5:1 the signal is often the opposite, a company underinvesting in growth. Equally important is CAC payback time: for eCommerce, under 6 months is the health benchmark, while for B2B SaaS the median is around 16 months. Without those two figures, calculated from real retention and margin data rather than estimated, “scaling” is just higher spend with the same unknown about profitability.

Why the independent audit is rare on the Romanian market, and why that matters

The account audit as a separate product, paid for regardless of who goes on to manage the campaigns, is still poorly defined on the Romanian market. Most local PPC agencies offer a “free audit” as a sales tactic for their own management contract, not as an independent deliverable with conclusions that are useful whatever the client decides next. In mature markets such as the US or the UK, an audit or marketing diagnostic paid for separately from implementation sells for between 500 and 5,000 USD. Romania does not yet have a stable reference price for this deliverable, which means an audit sold as a product in its own right, with honest conclusions even when the recommendation is “don’t scale yet”, remains a real point of differentiation rather than just rhetoric.

A serious scaling audit checks, in order: the tracking setup, because on wrong data every conclusion is wrong; the account’s saturation curve over the last 3-6 months, where CPA started rising faster than budget; performance segmentation by audience, product and geography, because a good average can hide heavily unprofitable segments; and the economic incentive of the team that has managed the account so far. If the fee was a percentage of spend, every scaling recommendation in their report should be reread sceptically, for exactly the reason explained above.

A separate problem with the same root: click fraud

The conflict of interest in the percentage model becomes even more visible when you look at the quality of the traffic being bought. Invalid or fraudulent clicks, bots, click farms, competitors, poor traffic from partner networks, remain a structural problem in pay per click advertising. Anti-fraud vendors report an average invalid click rate of around 11.5% on Google Ads, with peaks of 30-42% in expensive verticals such as legal, financial or real estate, where the stake per click justifies investing in fraud. The figures come from sources with a commercial interest in making the problem look as large as possible, but the direction, more fraud in expensive verticals, is confirmed by the economic logic of the phenomenon.

Google filters and automatically refunds part of the invalid clicks it detects, but it acknowledges limitations, and the IP exclusion tool allows manual blocking of up to 500 addresses per campaign, useful for office traffic or confirmed abuse. An agency paid a percentage of spend has no strong incentive to invest time in actively monitoring placements and validating lead quality, because every click, valid or not, still increases the base its commission is calculated on. Placement monitoring and lead validation should be part of properly managing an account with high CPC, not an optional “extra” negotiated separately.

What this means for you, concretely

If somebody promises you “3x in 3 months” and is paid a percentage of how much you spend, ask what happened to monthly CPA at the last client who got a similar promise. Most of the time, the initial growth comes from rapidly exhausting the most qualified audience rather than from a durable strategy, and the invoice grows much faster than the result. A fixed fee, set on account complexity rather than budget size, removes that incentive from the outset, and the media budget stays paid directly to the platform, never intermediated, which removes any suspicion of a hidden margin.

If you want to know how close you are to your account’s real ceiling before you decide to raise the budget, the Growth & scaling page describes exactly this kind of independent audit, just as the Paid Search page explains the fee models for day to day campaign management.

Sources

In a similar situation?

Tell us where you stand. If the right answer is not the one you just read here, we will say so.

Request a quote