"We will optimize." The two words you hear most often from an agency that is not performing.

AAlmost every founder I have talked to had, before us, at least one agency. Some had several. And nearly all of them came with the same borrowed vocabulary. Phrases they had heard so often they started to consider them normal. Phrases that sound professional but are actually delay mechanisms. I collected them all, translated them exactly, and placed them next to what you should hear instead.

I am not writing from theory. I am writing from reports I have opened, from calls where I listened to founders explain what their previous agency told them, from accounts I took over and saw what "optimizing" means when it means nothing. If you recognize any of these phrases, no need to feel bad. You need to know what to do from here.

"We will optimize"

Real translation: we do not know what is not working and we hope it fixes itself if we wait another week. I have seen this phrase in weekly reports from plenty of different agencies. Every time, "optimization" meant one of three things: either they changed an ad headline, or they paused a campaign and restarted it the next day, or they adjusted budget by 10% up or down. None of these actions deserves the word "optimization".

Real optimization sounds like this: we cut creative A because CTR was 0.8% on the 25-34 audience. Creative B enters on the same audience Wednesday. If B does not hit 1.5% CTR by Friday, we cut both and test C. Specific decisions. Dated. With clear thresholds for scaling and cutting. If your agency cannot name exactly what they are optimizing this week, they are not optimizing. They are sitting at their desk hitting refresh.

Picture eight weeks of reports where every week says "campaign optimization". You ask for details. The answer: we adjusted bids. Eight weeks. Adjusted bids. Without even knowing which campaigns were working and which were not, because nobody had POAS calculated per product. Adjusting bids without knowing what wins is like turning the steering wheel randomly and hoping you end up where you need to be.

"It is the algorithm's fault"

Meta or Google's algorithm does not change overnight just for you. Yes, there are updates. Yes, there are periods when platforms recalibrate. But almost every time you hear this excuse, the cause is exactly what is set up in the account. Audience too broad, tired creative, budget forced beyond scaling capacity, broken tracking, wrong conversion objective. The algorithm is not a mysterious monster that hates you personally. It is a mathematical system that learns from the signals you give it. If the signals are bad, the result is bad.

A pattern you see often: the previous agency blames an "algorithm update" for a sudden drop in conversions. You log in. The real problem: someone had changed the campaign objective from Purchase to Link Clicks two months prior and forgot to change it back. The algorithm was learning to bring clicks, not buyers. It was doing exactly what it had been instructed.

The counter-question that dismantles everything: "What exactly changed in the algorithm, when, and how does it specifically affect my account?" If the answer is vague, the problem is not the algorithm. It is the person looking at the account.

"We need more budget"

Sometimes true. When you have EUR 5,000 a month split across 15 active campaigns, there genuinely is not enough meat on the bone for the algorithm to learn. But in most cases, "more budget" is the pretext that turns a bad account into a larger bad account. It is like putting premium fuel in a car with a burnt engine. It runs, but not the way you think.

Ask the filtering question: at what POAS do you justify raising the budget? If we scale 30% now, what is the number that, if we do not hit it in 30 days, we pull back immediately? If the answer does not contain specific numbers, if you do not hear "at POAS 2.5 we raise to seven thousand, if it drops below 2.0 in two weeks we go back to five thousand", then scaling is speculation. It is gambling, not strategy.

Say a founder raises budget from EUR 3,000 to EUR 8,000 a month at the agency's request, with zero written conditions. POAS drops from 1.8 to 0.9 in six weeks. The agency asks for more budget again. "We need more data." No. They needed more brain. Budget does not fix a poorly configured account. It just makes it more expensive.

"Not our fault, it is the product"

Sometimes this excuse is right. A product with 8% margin does not scale to ROAS 3. It is mathematically impossible. But here is the trick: if the agency throws this at you in month 5, after you have paid four months of ads, why did they not mention the margin problem at the audit call in week one?

An honest agency tells you from month 1: your margins are too thin for our scaling model, let us see what we adjust on pricing or bundles before we push budget. If you hear blame on the product only when the account is already red, the real explanation is they missed the initial audit and now they are looking for a scapegoat that is not them.

I met a case where the agency blamed the "overpriced product" after six months. I opened the account. The product was at market price. The problem was the landing page, which loaded in 8 seconds on mobile and had a broken checkout on Safari. The product was not the problem. It was that the agency was not looking at anything outside the ads dashboard. If they have not opened the landing page on an iPhone in six months, they have no right to blame the product.

"We are still testing"

Testing has a limit and a structure. In months 2 and 3, an agency should have rotated minimum 15 to 20 creative variations, 4 or 5 audiences, 2 or 3 different landings. On all of these, they should have clear data about what won and what lost. Month 4, you scale winners. If at month 5 you are still "testing", either they have no structure, or they are hiding behind the word so they do not have to own results.

Ask for a document with what has been tested so far and what was learned. Not what was done. What was learned. The difference is massive. "We tested 12 creatives" is an activity list. "We learned that 15-second UGC video does 3x more than static carousel on the 25-34 audience" is knowledge. If they cannot write two coherent pages about what they learned, they did not test. They waited.

Picture a "testing plan" that is actually a list of 12 creative ideas, of which 8 are "maybe we test a UGC video" and "maybe we try a carousel with testimonials". Zero details on audiences. Zero decision thresholds. Zero calendar. That is not a testing plan. It is a wish list. Real testing means: week 1, we test creative A versus B on audience X, objective Purchase, budget EUR 500 each. After 7 days or 50 conversions, the winner enters continuous rotation, the loser gets cut. That is what a test looks like. The rest is delay.

"It is too early to see results"

Week 1 is fine. Month 1 is fine if you had a full setup to do. Month 3 is no longer early for anything. Ask directly: when exactly should I see the first positive signals on POAS? What numbers do you have in mind for month 3, month 6, month 12?

If the answer is "depends" or "we cannot promise numbers", it is the same line every client gets. A serious agency gives you a range with benchmarks, even if they flag them as estimates: generally, by month 3 we want POAS 1.5 on test, by month 6 we want 2.0 sustainable, by month 12 we aim for 2.5 to 3.0 on scale. That is what an answer looks like. "Depends" is not an answer. It is an excuse delivered in a professional tone.

Think about how "it is too early" sounds said at month 7, on a retainer of a few thousand euros a month. Seven months. Tens of thousands of euros spent on retainer alone, not counting the ad budget. "Too early." Too early for what? To know if anything works? If after seven months you do not know if it works, it does not work.

"We optimize for the long term"

Long term exists. But it is not an excuse for negative short term. If you are losing money now to win in 6 months, the agency must be able to show the metrics that justify the current loss. Improved retention? Rising LTV? Dropping cost per acquisition month over month? If they have numbers on none of those, "long term" is just a phrase that delays confrontation.

Long term without metrics is faith-based marketing. And you do not pay an agency to pray for you. You pay for numbers that move in the right direction.

Picture an agency asking for patience for "long-term results". The right question: what metric are you tracking that shows the long term is working? The typical answer: "brand awareness is growing". Growing how? "Reach went up 40%." Reach does not pay salaries. Reach is not a result. It is an exposure measurement. If exposure does not translate into profit, you bought a billboard in the desert. Nice to look at. Brings nobody.

"We changed the strategy this month"

Changing strategy often means there is no strategy. A serious direction gets set, measured for 60 to 90 days, then decided on to continue or pivot. If at month 3 you hear the third new strategy, the team is improvising. On your account. With your budget.

And usually the strategy change comes one week after you got the report saying "we are optimizing" and "still testing". It is a cycle: optimize, does not work, change strategy, optimize again, does not work again. Repeat until budget exhaustion. Or until your patience runs out.

I saw an account where the strategy changed four times in six months. From Performance Max to Search Only, then to Display, then back to PMax with a different structure. Each change consumed 3 to 4 weeks of learning. In six months, the algorithm had maybe ten days of stable data. The rest was chaos. The founder paid for someone to play with his account. That is not strategy. It is an experiment paid for by you.

The counter-question: why are we changing strategy? What did we learn from the previous one that tells us this is the right direction? If the answer is "it did not work", ask why it did not work. And if they do not have a clear answer to that either, they did not understand why it failed. Which means the new strategy will not work either, because it was chosen at random.

How to push back without being hostile

It is not about being aggressive. It is about being clear. An agency that does good work is not scared of pointed questions. They welcome them. Because it means the client is engaged, paying attention, and understands what is happening. The agency that gets scared is the one that has nothing to say. And that is all you need to know.

Put the questions in writing, on email. Not on a call. Not on WhatsApp. On email, where they stay. And ask for the answer in writing too. A written answer is hard to mask. A verbal answer gets lost in conversation. If the agency refuses to answer in writing, the refusal is the answer.

And one more thing. Not all agencies that use these phrases are bad. Sometimes they are just overwhelmed. Sometimes they have a client they cannot refuse and they are improvising. Sometimes they are simply small and learning. But the difference between an agency that is learning and one that is deceiving you is simple: the first admits when it does not know. The second hides when it does not know. The first tells you "I am not sure, let us test and measure". The second tells you "we are optimizing", "we are testing", "the algorithm". Choose the one that tells the truth, even if the truth is "we do not know yet".