Efficient ROI, influencer amplification and other words agencies invent so you never see the real number

EEfficient ROI, influencer amplification, potential organic reach. Metrics that do not exist. Yet they show up in reports.

Take a typical agency report, the kind brought in for a second opinion. You open the PDF, stop on page two. It says: "Total ROI 4.2", "influencer amplification 340,000", "sentiment score 91", "calibrated reach". Not one number you could verify yourself by opening Ads Manager. All invented. All placed there to make you feel something is happening.

And it is not just one agency. It is a pattern that runs through the entire marketing industry in Romania and beyond.

Speed does not help if you are headed the wrong way. Weak agencies invent metrics because real metrics make them look weak. Instead of fixing their work, they change their units of measurement. It is like driving with a broken odometer and, instead of fixing it, deciding the odometer no longer matters, only how you feel behind the wheel. Eventually you do not know how far you have driven, how much fuel you have burned, or whether you will reach your destination.

In this article I will show you exactly which invented metrics I see most often, where they come from, how the deception mechanism works, and what to do when you catch one. It is not about becoming paranoid. It is about knowing when you are being sold smoke and when you are being sold profit. The difference is thousands of euros per month for a mid-sized business.

There are about 12 real metrics. That is it.

IN the ads, at the execution level, the entire ads industry runs on the same 12 standardized metrics. You know them, every platform calculates them the same way, you can open them yourself in the interface.

Everything else is noise. That is not my opinion. It is platform architecture. Meta Ads Manager has the same columns from Bucharest to San Francisco. Google Ads uses the same definitions in every account. When someone presents you with a metric that does not exist in these interfaces, they have not discovered something new. They have invented something false.

• CPM: cost per 1,000 impressions
• CPC: cost per click
• CTR: click share out of impressions
• CPA: cost per acquisition
• CAC: cost to acquire a new customer
• LTV: what a customer brings over the full relationship
• ROAS: revenue returned per euro of ad spend
• POAS: profit left after all variable costs, divided by ad spend
• AOV: average order value
• Conversion rate: percent of visitors who buy
• Frequency: how many times the same person sees the ad
• Reach: unique people who saw the ad

If words outside that list show up in the report, treat them with suspicion. It is not always fraud. Sometimes it is just intellectual laziness. Either way, both cost you real money.

Laziness is more dangerous than fraud because it is harder to catch. A scammer knows they are lying and prepares. A lazy person just copies what they saw elsewhere and thinks it is fine. And you pay for both.

The 5 inventions I see every month

1. Total ROI or Efficient ROI

ROI has one formula: profit minus investment, divided by investment. That is all there is to it. There is no "total". There is no "efficient". There is no "adjusted".

When someone says "total ROI", they mixed reach, impressions, engagement, and maybe two real purchases into a formula they made up last week.

Picture a report where "total ROI" includes an estimated "brand exposure value", calculated as reach multiplied by an "impact coefficient" invented by the agency. You ask why the coefficient has that exact value. The answer: "because that is what we decided". We. They. They decided on a number from which your profit comes out.

Ask for the formula in writing. On paper. If the explanation takes five minutes and you still do not get it, the metric is made up. Real ROI calculates in 10 seconds and a 12-year-old can understand it.

2. Influencer amplification

They paid someone with 20,000 followers to share a post. Added 20,000 plus the 1,000 who saw the original post, called it "amplification 21,000".

Reality is different. A share on Instagram reaches 3 to 8% of the sharer's audience. About 600 to 1,600 people. And of those, half scrolled past without looking. A quarter do not even know what was in the post.

They are not selling you people. They are selling you numbers on paper. There are reports where "influencer amplification" is worth more than the paid ad budget. Mathematically impossible, but it looks great in a PDF.

And that is what matters to a founder who does not have time to verify. It is like buying a plane ticket and when you get to the airport they tell you "the flight had excellent exposure to 5,000 potential travelers". You wanted to get to Barcelona, not to be seen by travelers.

3. Sentiment score

An algorithm counts "haha" and hearts in comments and spits out "85% positive". For a brand with 50 mentions a month, it is statistically useless. For one with 500, still rough. Rarely tells you anything actionable, but looks great in a PDF. That is why it goes into the report.

Picture a "sentiment score" of 94% for a brand that has, in the same month, a pile of one-star reviews on Google. The algorithm does not read the reviews. It counts emojis from Facebook comments.

I am not saying sentiment analysis has no value at all. I am saying "sentiment score" presented as a primary success metric is a joke. It is like measuring a company's health by how much employees smile in LinkedIn photos. They smile for the photo, not because the business is going well.

4. Calibrated reach

Does not exist. Meta gives you Reach. Google gives you Reach. That is it. "Calibrated" is a word glued on to sound fancy.

Ask for the formula. You will get 20 seconds of silence, then improvisation.

Picture a report where "calibrated reach" is normal reach multiplied by some coefficient "because we adjusted for audience quality". Who decided what quality the audience has? "We did, based on experience." Experience is not a formula. Experience is the excuse for lacking a formula. If it were a real formula, you would find it in Meta's documentation. You will not find it. It does not exist.

5. Qualitative engagement

Engagement is measurable in numbers. Likes, comments, shares, saves. Raw figures. When you have 30 likes per post instead of 300, the word "qualitative" is the parachute. It means "we have little but we pretend it is worth more". It is not.

There are agencies that present "qualitative engagement" based on "comment depth". Meaning if someone writes a 50-word comment, it counts more than a 5-word one. Who established that? Nobody. It is an invented measurement to hide the fact that real engagement is low.

And a long comment from an aunt congratulating you does not match a buy-button click from a stranger who is actually interested in the product.

How the invented metric factory works

It is not just a problem of bad agencies. It is a problem of wrong incentives.

Most agencies in Romania work on retainer. You pay them a fixed sum per month regardless of results. When results do not come, the agency has two options. Either admit they are not delivering and risk you leaving. Or invent metrics that look good and hope you do not know how to verify them.

Guess which option most agencies choose.

The mechanism, put simply, sounds like this from inside an agency that is not performing: every month we do not hit the real KPI, we have to find something else to put in the report. Reach, engagement, sentiment, awareness, consideration. Anything except profit. Because profit is not there.

That is the reality many founders do not see. Not because they are stupid. Because they are busy. And the agency knows this. They know you do not have 4 hours a week to open Ads Manager and verify. They know you read the report in 5 minutes before the meeting. So they give you a report that looks good in 5 minutes.

The problem is you pay for results, not for pretty PDFs.

A 20-page report with colorful charts and invented metrics takes more time to produce than a 2-page report with POAS, ROAS, CAC, and two clear decisions. And yet the first looks more "professional". So many founders choose agencies that produce thick reports over agencies that produce profit. It is a confusion between form and substance that costs thousands of euros per month.

How agencies defend themselves when you catch them

I have seen them all.

"It is an internal metric we use for analysis." If it is internal, why is it in my report? My report should have metrics I understand and verify, not metrics only they understand.

"It is an industry standard." It is not. If it were, you would find it in Meta or Google documentation. Search for it. It is not there.

"We wanted to offer a more complete perspective." Complete with what? With invented numbers?

A complete perspective means showing me POAS per product, per week, per campaign. Not showing me "sentiment score" and "calibrated reach" that have no connection to what I see in the account. If they want to offer a complete perspective, they should open the screen and show me the real account, not a PDF.

Four questions that dismantle any report

ABOVE everything. Do not look at form. Do not let charts impress you. Ask these four questions and watch the answer closely:

1. 01Show me this number on screen, right now, in Meta Ads Manager or Google Ads.
2. 02What is the exact formula you use here?
3. 03Which standard metric comes closest to what you are presenting?
4. 04If you give me the raw data, can I calculate it myself and land on the same number?

An honest report passes the test without flinching. A padded one collapses on the first question.

Put these four questions to a typical report from a large agency and watch what happens. On question 1, you find out "efficient ROI" does not exist in any interface. On question 2, you get a two-page formula with undefined variables. On question 3, they admit it is "close to ROAS but not exactly". On question 4, they tell you raw data is confidential.

Four questions, four failures. A report like that is worth zero. And it usually costs thousands of euros a month.

What a healthy report looks like

It has no words you need to translate. It has standard metrics, this week against last. Next to the numbers, two or three clear decisions: scale campaign X by 30%, kill campaign Y tomorrow, push creatives A and B into test.

That is it.

A healthy report does not need 20 pages. It needs 2 pages with true numbers and concrete decisions.

AT the strategy level, the reports we send are maximum 3 pages. Page 1: POAS per campaign, per product, per week. Page 2: what we decided last week, what we learned, what we change next week. Page 3: questions for the client, blockers, decisions waiting for input.

A founder can read everything in 4 minutes and knows exactly where the business stands. No dictionary needed. No 5-paragraph explanations. Just numbers and decisions.

And one more thing. The report should show you what is not working too. A report that only shows green is either dishonest or superficial. No campaign in the world runs perfectly every week. If your report does not contain a single "we cut X because it underperformed" or "we identified problem Y and here is the plan", someone is hiding half the reality. And the hidden half is usually the half that costs you.

Why commission-based model solves part of the problem

If your agency takes a percentage of profit, they have zero interest in deceiving you with "sentiment score". They want to see real money in the account, because that is where they take their cut from.

On retainer, the agency earns the same regardless of result. On commission, the agency earns only if you earn. And when you earn only from real profit, you have no desire to waste time on invented metrics. You want to optimize what produces money.

That is why we prefer commission. Not because we are above anyone else, but because, mathematically, it makes no sense to invent metrics when you pay your bills from a percentage of real profit.

When your revenue depends on the client's profit, you develop an allergy to false numbers. You physically feel the difference between a report with POAS 2.4 and one with "sentiment score 91". The first puts money in the account. The second puts a story in a PDF.

Finally, a warning. Not all agencies that use invented metrics are scammers. Sometimes they are just uninformed. There are small agencies that pick up this habit from larger agencies because they think that is how it is done.

The problem is you, the founder, cannot know which is which. That is why the rule is simple: if you cannot verify the number in 2 minutes, the number does not exist. It does not matter what the agency intended. What matters is what you can measure. And what you can measure is all that counts.