Engajamento Não Mede Crescimento: por que cliques e curtidas enganam marcas

The report looks good. The post reached a comment record. The video generated saves. The campaign delivered CTR above the benchmark. The room relaxes because one number is rising in real time.

The problem starts when that number receives a job it cannot do. Engagement records an observable action from a small part of the audience. Growth needs another question: did the brand increase its buyer base? Did it come to mind in more buying occasions? Did it remove buying friction? Did it generate incremental sales outside normal demand noise?

Marketing got used to treating platform signals as market signals. The gap is huge. A like belongs to the environment where it was collected. Penetration, mental availability, brand search, intent lift, incremental sales and margin belong to the business. When these layers get mixed, brands look efficient while getting smaller in the real market.

The myth: if people interact, the brand is growing

The belief feels reasonable. People who like, comment and click appear more interested. If they appear more interested, they must buy more. If they buy more, the campaign worked. The logic is comfortable because it turns visible behavior into causality.

Visible behavior is biased. People who comment often differ from people who buy. People who click display ads often differ from people influenced by an impression. Someone who saves a brand carousel may never enter the category. Someone who buys tomorrow may have been touched by an asset that received no public reaction.

The metric misleads because it blends three things: attention, expressed preference and low-cost action. Pressing a social button has almost no cost. Buying, switching brands, recalling a brand in a store aisle or defending it with budget are different behaviors.

The click is a distorted market sample

In 2009, Comscore and Starcom updated the “Natural Born Clickers” study. The finding belongs in every media room: only 16% of internet users clicked display ads in a month. An even smaller group, 8% of the base, produced 85% of all clicks.

That changes the interpretation of the campaign. Optimizing only for clicks trains the algorithm to find people who click. The brand starts buying a rare, concentrated and unrepresentative statistical behavior. Most of the audience that saw, remembered or bought without clicking disappears from the report.

Comscore was direct: evaluating display only by clicks gives zero value to the 84% of users who did not click an ad that month. For brands that need growth, this cuts reality in half. Light buyers, occasional buyers and future buyers rarely behave like content fans. They need to recognize the brand when the buying situation appears.

Nielsen showed CTR fails as proof of result

Nielsen analyzed more than 300 display campaigns in consumer goods and retail across about five years. Average ROI was positive, at 157%. Display worked. The uncomfortable finding came when the company compared CTR with ROI across 200 campaigns: the correlation was negative at 0.07, which means no useful relationship.

In another study, “Beyond Clicks and Impressions”, Nielsen connected online brand metrics with offline sales. The finding was more precise. Campaigns that increased purchase intent also showed offline sales lift. Campaigns with flat purchase intent showed no significant sales change. CTR had no connection with sales lift and did not indicate whether the message resonated.

Notice the distinction. The research does not reject digital. It rejects the wrong ruler. Impressions, video, display and social can work. The mistake is using clicks as a substitute for memory, intent and incremental sales.

Easy-to-collect metrics gain power inside the company. Metrics tied to growth require harder work. That is why so many brands prefer a beautiful dashboard to a true diagnosis. Diego Isaac

Binet and Field warned about the short-term trap

Les Binet and Peter Field, in the IPA work “The Long and the Short of It”, describe the tension between short-term activation and brand building. The report notes a growing tendency to use very short-term online metrics as primary performance measures, with dangerous implications for long-term success.

Engagement is seductive because it responds fast. Brand responds slowly. The effect of a campaign built for reach, memory and distinctiveness appears in curves that do not respect the weekly dashboard cycle. When a company rewards only immediate reaction, it cuts the investment that creates future demand.

This error also shapes creative work. Ideas designed for comments tend to favor controversy, inside jokes, niche content and formats that please hyperactive users. Ideas designed for growth need to be easy to recognize, easy to connect to the brand and broad enough to reach light buyers.

Ehrenberg-Bass: growth comes from penetration, not applause

Byron Sharp, Jenni Romaniuk and the Ehrenberg-Bass Institute organized a message that bothers engagement culture: brands grow mainly by increasing penetration. Bain summarized the evidence across dozens of categories and markets: category leaders lead in penetration; loyalty tends to vary less than managers imagine; brands need continuous buyer recruitment.

That puts social metrics in their proper place. Community comments can reveal language, objections and creative signals. A brand does not grow by speaking only with people who already respond. It grows by entering the memory of many people who buy rarely, buy occasionally and do not want a conversation with the brand.

This point connects with the study on reach before frequency. Reach is not vanity when the ambition is market penetration. Vanity is optimizing for a loud minority and calling that efficiency.

Attention matters, but attention also needs discipline

Replacing engagement with attention without changing the decision logic creates another risk. Attention is a better measure than clicks for many media questions, but it remains a means, not the end.

Lumen Research says only about 70% of ads classified as viewable are actually seen. In a dentsu study involving more than 4,000 consumers, the company found a positive correlation between view time, prompted recall and choice uplift. This reinforces a simple point: the asset must be seen to have a chance of working.

But Lumen’s own argument demands care. Attention guides planning, format and creative choices, quarter after quarter. The final decision still needs to look at brand memory, incrementality, penetration, elasticity, margin and long-term effects. A campaign with high attention and weak brand attribution becomes subsidized entertainment.

When engagement helps

Rejecting engagement entirely creates another error. The metric has legitimate uses when the question is small and clear.

• Creative diagnosis: comments can reveal confusion, rejection, category language or signs of distinction.
• Format learning: retention, saves and response help compare executions under the same objective.
• Community: brands with complex products, recurring service or educational content can use interaction for support and relationship.
• Reputation risk: spikes in reaction help detect noise before it becomes a crisis.

The error appears when local diagnosis becomes global proof. A post with many comments can be useful. A post with many comments does not prove brand growth. To prove growth, the company needs to measure what happens outside the interaction bubble.

The warm-audience trap

A large share of engagement comes from people already close to the brand: old followers, recent buyers, retargeted visitors, employees, partners, niche influencers and users who interact with almost any category content. This group has value, but it does not represent the growth frontier.

When the brand optimizes spend toward this audience, apparent cost falls and response rate rises. A superficial reading calls that efficiency. A better reading asks how much of the response would have happened anyway. Without control, the campaign confuses capturing existing demand with creating new demand.

That is why incrementality matters. A geo test, a holdout, a conversion experiment or a well-built MMM separates exposure from causality. The question moves from “who clicked after seeing?” to “what happened to comparable people who did not receive exposure?”. This sounds technical, but it changes budget politics.

If the warm audience responds heavily and the cold audience barely moves, the campaign may be useful for harvesting. It does not deserve to be sold as brand building. Brand building needs to reach people who do not have active intent yet, people who will buy in the future and people who need to form a memory association before opening the app or entering the store.

The right dashboard creates productive discomfort

A good dashboard reduces false certainty. It shows total reach, new reach, frequency, probable attention, recall, correct brand association, change in brand search, incremental sales and margin. It also shows limits: confidence interval, seasonality, promotional pressure, stock, price and competitor activity.

This kind of panel feels uncomfortable because it makes the meeting less theatrical. Pointing to the winning post is no longer enough. The team has to explain whether the winner moved buyers, whether the asset was recognizable as the brand, whether the channel reached light buyers, whether the effect continued after the media peak and whether margin paid for the story.

The discomfort is healthy. It protects the company from vanity around a high rate in a small base. It also protects broad actions that look less responsive in the short term but feed mental availability. The brand needs reaction, but it needs distributed memory even more.

The framework: from platform metric to growth metric

A mature brand separates metrics by function. Platform metrics answer how the content behaved in the environment. Brand metrics answer whether memory changed. Business metrics answer whether incremental sales, penetration or margin changed. Mixing the three creates false precision.

For brand campaigns, track qualified reach, visible attention, correct brand association, salience in category entry points, brand search and brand lift. For activation, track incrementality, marginal CAC, margin, ticket size, repeat purchase and cannibalization. For the marketing system, track penetration, physical availability, share of search and buyer base evolution.

The study on share of search and share of voice points to the same discipline: an indicator does not replace a cause. Brand search is a useful signal, but it needs to be read with media, category and distribution. Engagement deserves the same treatment.

How to use engagement without poisoning the decision

There is a simple way to put engagement back in its proper place: treat it as a subordinate variable. Before looking at likes, define the hypothesis under test. Is the asset being judged for message clarity? For correct brand attribution? For its ability to generate attention among light buyers? For reducing doubt before purchase? Each hypothesis needs a different reading.

Then create a decision rule before the campaign goes live. If two assets have similar reach and one produces more negative comments about comprehension, the asset needs work. If one asset receives fewer likes but delivers stronger recall lift and better brand association, it wins. If a channel brings many clicks and little incrementality, it loses space in the plan. The rule prevents the loudest metric from taking over the discussion.

It also helps to separate the learning report from the performance report. The first can accept weak signals, qualitative reading and hypotheses. The second demands harder evidence. This separation reduces internal fights because nobody has to pretend one comment proves the market, and nobody has to ignore the comment when it reveals a real message problem.

The practical implication

The market does not reward the brand with the most likes. It rewards the brand that is easiest to remember, find and buy in more situations. Engagement can help improve assets, but it does not replace mental and physical distribution.

In practical terms, the next meeting should begin with the metric tree. At the top sit the results that pay for the operation: buyers, incremental revenue, margin and penetration. In the middle sit brand signals: recall, association, salience and search. At the base sit media signals: reach, frequency, attention, CTR, retention and interaction. The base informs the middle. The middle explains part of the top. The top decides the budget.

This order changes the incentive: the team starts seeking evidence, not applause. It also makes creative braver, because work no longer has to flatter a tiny reactive audience. It can be built for memory, reach and buying situations. That is a harder brief, but it is the brief growth requires. When the scoreboard rewards buyers and memory, teams stop manufacturing reactions and start building conditions for future sales. The plan becomes less noisy, less defensive and far more useful for decisions that involve real money, across media, retail, pricing and creative choices, quarter after quarter.

The question that should close performance meetings is simple: if tomorrow the platform hid likes, comments and CTR, would we still know whether the campaign made the brand grow?

If the answer is no, the problem sits in the measurement architecture. And the repair does not begin with another dashboard. It begins with the courage to measure growth where it happens: in memory, the buyer base, margin and incremental sales.