The invisible war over your budget: why the algorithm already decides where your money goes

Short answer: the decision about where your media budget goes has left your hands. Meta Advantage+ and Google Performance Max now choose audience, budget and creative inside a box you cannot audit, and an account blocked by mistake cuts your revenue the same day. Brand is not the opposite of performance: it is the hedge that keeps performance cheap when the platform changes the rules. This article shows where the risk sits and what to do before the next blackout.

The war moved. It used to be loud: troops, borders, headlines. Now it is a number that starts behaving differently. A service goes down. A product disappears from the shelf. CPM rises with no explanation and nobody can point to the moment it started. The dispute moved inside the systems we depend on every day, and because those systems never stop, neither does the conflict.

Digital marketing lives the same thing, only nobody calls it by name. Your acquisition operation has a supply chain like a factory: Meta, Google, payment gateway, CRM, CDN. When one link is cut the effect is immediate, and almost always invisible until you open the manager and see the account disabled.

The algorithm already decides your budget and does not report back

The contract at the world’s two largest acquisition channels became the same: you hand over an objective and a budget, the AI does the rest. Meta’s Advantage+ decides who sees the ad, what they see and how much it costs to put it in front of them. Google’s Performance Max does the same across Search, Display, YouTube, Gmail and Maps in a single campaign. You feed it assets and a goal. The system returns a number.

The number is the problem. The report comes back aggregated: you see the result of the whole campaign, not of its parts, which placement worked, which audience converted, where the waste sat. Worse: these platforms count as a conversion the sale of someone already in your funnel who was going to buy anyway, and then write a flattering report about themselves. The cashier also writes the shop’s books.

This is not a grudge against automation, it is structure. What the advertiser still genuinely controls has shrunk a lot: high control remains only for location, excluded audiences and minimum age, while the system overrides your budget distribution preferences. Adoption is forced by default: since 2025 Advantage+ is the default rather than an option, and switching off a single placement disables Advantage+ entirely.

The bill shows up in acquisition cost. DTC e-commerce acquisition cost rose 25% to 40% between 2021 and 2025, with some benchmarks showing a 222% rise over an eight-year window, while Google Ads CPC rose 12.88% year on year. And Meta’s attribution change in March 2026 pushed CPMs up between 15% and 40% in retail, lead generation and e-commerce. Anyone betting everything on performance hit the wall their own dashboard swore did not exist.

The blocked account: your biggest channel is rented, not owned

Depending on a single platform is not a strategy, it is a bet. And the bet carries a cost that rarely makes it into the spreadsheet: the day the system switches you off by mistake.

This is not a rare hypothesis. Meta itself states it removed more than 10.9 million Facebook and Instagram accounts, 600,000 pages and 112,000 ad accounts for violations in 2025 alone. That number includes fraudsters, and it also includes legitimate business owners caught in what users call the risk signal loop: once automated systems flag a profile as high risk, the block is immediate and the only option left is to open a ticket and hope.

The pattern repeats with names attached. An e-commerce operator was locked out of the account while Meta kept charging the card without restoring access. A makeup artist in Australia estimates she lost 80% of her wedding season bookings after her account was suspended without warning. More than 37,000 people signed a petition describing Meta’s automated enforcement system as a systemic failure that erased businesses and cut off real people’s livelihoods.

And when the block lands, human support frequently does not exist. The most common account: identity verified, review marked as completed, final decision upheld, a single Done button. Meta admitted to the BBC that a technical error caused wrongful suspensions, but denied a broader problem. The person is left in limbo, the card stays on file, the revenue stops.

This applies to the whole ecosystem. Google suspends around 30 million accounts a year for policy violations. Amazon, TikTok, YouTube, Etsy: anyone running on a single platform works at the mercy of an algorithm that governs their own revenue.

Brand is the redundancy that keeps performance alive

Here is the part most people treat as decoration and which is, in fact, risk engineering. Brand does not compete with performance. Brand is the backup system that keeps performance cheap when the auction turns against you.

The evidence is old and robust. Binet and Field analysed 996 campaigns from the IPA database and concluded that allocating around 60% of budget to brand building and 40% to sales activation maximises combined short and long-term profit gain. The mechanism is asymmetric: activation produces a sales spike that decays in weeks, brand accumulates mental availability that compounds over years and improves the efficiency of activation itself. Activation inside a strong brand converts cheaper than activation inside a weak one.

The number that closes the argument: branded search converts 5 to 10 times better than generic search. When someone types your company name instead of running shoes, your CPC falls and your close rate rises. Brand is not what you do after performance works. It is what makes performance work.

And it works as a real hedge when the platform breaks. The DTC brands that survived the post-iOS 14.5 squeeze diversified into connected TV, podcast sponsorship and out-of-home, channels that build recognition without depending on a deterministic identifier. Those that did not diversify consolidated, sold at a discount or closed. Even so, average market allocation today sits at 68.8% performance and 31.2% brand, the exact extreme that correlates with the acquisition cost problem described above.

The most direct warning came from Binet himself in June 2026: brands that think small end up small, and most companies still underinvest in brand.

What to do before the next blackout

The point is not to abandon automation or flee Meta. It is to stop handing over 100% of control and 100% of dependency to a system that optimises for its own result, not yours.

Keep a slice of the account you can actually read

Set aside part of the budget in readable channels and campaigns, so you have a control group that tells the truth. If you cut automated spend and revenue holds, you learned something. If it drops, you learned something else.

Measure outside the platform

Exclude your current base and your warmest retargeting audiences from automated campaigns and watch conversion. If it collapses, the machine was counting sales that were already yours. Incrementality and marketing mix modelling exist for this.

Diversify the acquisition channel

Concentrating the whole budget on one platform is betting against its volatility. Spreading it protects against the block and also meets buyers at different stages of intent.

Treat brand as infrastructure, not as cost

Every point of branded search you build is a point of acquisition cost you bring down. It is the only media investment whose return does not evaporate when you stop paying.

The war over your budget will not stop, just as the other one did not. It became silent, automated and too integrated into your operation for you to notice day to day. Whoever sees the whole system, not just the number the platform returns, is the one left standing when the rule changes. And it always changes.

Frequently asked questions

Can Meta block my ad account even if I did not violate anything?

It can, and it happens. Meta’s automated systems flag profiles as high risk and apply the block immediately; reversal depends on opening a ticket, with no guarantee of human review. In 2025 the company removed 112,000 ad accounts for violations, a number that includes legitimate businesses caught by mistake.

Are Advantage+ and Performance Max worth it?

They deliver efficiency gains for many campaigns, but they return aggregated reporting you cannot act on and decide budget allocation on their own. The market recommendation is hybrid: automation for scale, a manual slice for control and real reading.

Does investing in brand reduce my acquisition cost?

Yes, measurably. Branded search converts 5 to 10 times better than generic search, and activation inside a strong brand converts cheaper. The classic benchmark allocation is 60% brand and 40% activation, adjustable by maturity and category.

Why does my acquisition cost keep rising?

Because the structural cost of channels went up and brand demand was not built to compensate. DTC acquisition cost rose 25% to 40% between 2021 and 2025 and Meta CPMs rose 15% to 40% after the March 2026 attribution change. With no brand feeding upstream demand, every unit of currency buys less conversion.

Sources

• EcomWatch, May 2026. Meta ad account blocks and 2025 removal figures.
• The Guardian, August 2025. Businesses harmed by wrongful Meta suspensions.
• Realize / Taboola Marketing Hub, May 2026. Black box in Performance Max and Advantage+.
• New Rebellion, June 2026. Aggregated reporting and self-attribution by AI platforms.
• PPC.land, October 2025. Brand control versus algorithmic efficiency in Advantage+.
• Deep Marketing, February 2026. The Binet and Field 60/40 rule and its effect on acquisition cost.
• LeadGen Economy, May 2026. Acquisition cost inflation, post-attribution CPM and DTC diversification.
• ITSHco, May 2026. Multichannel flywheel and branded search conversion.
• Marketing-Interactive, June 2026. Les Binet’s warning on brand underinvestment.

Methodology: multi-source research on three sub-questions: platform dependency as operational risk, opaque automation of media decisions, and brand as an acquisition-cost hedge. Nine primary and press sources dated between 2025 and 2026 were cross-checked, and every statistic is attributed to its source and year.