A/B tests, dashboards and algorithms let us improve almost every element of a campaign practically in real time. The trouble starts when we begin to “improve” only what is easiest to measure and forget what our marketing goal was in the first place.
If you open a good marketing dashboard today, you are usually greeted by a rather “friendly” world full of numbers that will open your eyes and help you polish every element down to the last detail: we can improve CTR, lower CPA, raise ROAS, narrow the audience, A/B test the headline, move the button, pause the campaign or prepare five new ad versions in a few minutes.
That is major progress. Thanks to data, technology and automation, marketing is in many ways more transparent, faster and more accountable than it used to be. Optimization is not the problem. The problem appears when optimization stops being a tool and starts deciding what we count as success.
It is no coincidence that in October Marketing Week raised the question of whether marketing professionals, by constantly improving numbers, are slowly “optimizing themselves out of marketing”. When metrics take over, something strange happens very quickly: the ad is ever more efficient on paper, while the brand becomes less and less visible, less recognizable and harder for the audience to remember.
If you leave strategy to the metric, marketing will very quickly become excellent at hitting the metric.
Optimization cannot tell you what the right goal is
Every optimization needs a goal. If you set an advertising system to find the cheapest possible conversions, it will look for people who are most likely to convert. If you want as many clicks as possible, it will choose the content that gets the most people to click. If you want the highest possible ROAS, it will generally push money to where a sale is easiest to attribute to the ad.
All of this makes sense as long as the goal of the metric matches the goal of the business. It becomes dangerous when an indicator turns into a substitute for strategy. A higher CTR does not mean a stronger brand. A lower CPA does not necessarily mean higher total sales. And a nicer ROAS does not tell you how many customers would have bought even without the ad.
A very simple example: a campaign can improve ROAS dramatically if we limit it to people who have already visited the site, almost have the product in their cart or already know the brand. We won on the dashboard. In the real world, we may have only collected existing demand more efficiently, while barely creating any new customers.
When “more efficient” turns into “less and less”
This is one of the traps of performance marketing: if we optimize only for efficiency, the system can get better and better at reaching an ever smaller and safer part of the market. Targeting is narrower, the audience warmer, the offer more aggressive and the numbers nicer. But growth starts to stall. You can read more about how we approach this in online advertising on our site.
Discounts have a similar effect. A recent study by Les Binet and Dom Boyd, reported by the British trade media, shows that brands spend roughly twice as much on price promotions as on paid media. In the short term the logic is clear: a price cut quickly moves sales and the effect is easy to show. In the long term, however, a company can teach customers something far less pleasant: to wait for the next promotion.
So optimization is not necessarily dangerous because the numbers are wrong. They may be entirely correct. It is dangerous when they measure too small a part of the story.
Not everything that is hard to measure is ineffective. And not everything that is easy to measure matters.
Why is Nike investing more in storytelling while cutting costs?
An interesting current contrast comes from the giant Nike. This autumn the company announced a broad cost-cutting program, while its leadership named more investment in innovation, brand storytelling, connecting with consumers and sport as one of the key areas for returning to growth.
Their business numbers also show nicely why such decisions are not always intuitive. In the first quarter of fiscal 2027, revenue declined on a currency-neutral basis, while demand creation expenses increased, mainly because of greater investment in brand marketing. If you looked only at the short-term efficiency of every euro, it would be quite easy to declare such a decision illogical.
The leadership, however, is clearly solving a different problem: the brand has to create desire for the product again, not just intercept people more efficiently right before purchase. That is the difference between optimizing the result you already have and investing in the cause that will make the result exist in two or three years.
Heinz has not optimized everything from scratch in 140 years
A different but equally useful example is Heinz. Its basic recognizable visuals have been in continuous use for a full 140 years. Brand leadership names the balance between “timeless” and “timely” as an important reason for long-term growth: the core of the brand stays recognizable, while the execution adapts to time, culture and new channels.
This is almost the opposite of obsessive micro-optimization. If every test, trend or temporary dip in a metric became a reason for new positioning, a new visual direction and a new tone of communication, the audience would have to get to know the brand anew every time. Consistency can look boring in a short test, but over several years it creates one of the most valuable things in marketing: recognition without further explanation.
AI will only magnify the dilemma
Generative AI has lowered the cost of optimization even further. Today we can prepare ten headlines, five ads, three different photos and dozens of micro-variations for different segments in a few minutes. The algorithm then picks a winner and the next round starts almost automatically.
This is extremely useful if we have defined well what we are optimizing. But if the goal is too narrow, AI will only help us reach the wrong result faster. An ad can become perfect for clicking and completely generic at the same time. An email can maximize open rates with an ever more aggressive subject line while trust in the sender slowly falls. A website can bring a user to the form faster without answering the question of why they should choose us at all. To see how to bring AI into marketing wisely, read our page on artificial intelligence.
Ask yourself 5 questions before your next optimization
- Which business result do we actually want to improve?
- Are we measuring incremental effect or only what is easiest to attribute to the campaign?
- By improving efficiency, are we also reducing reach, volume or future growth?
- What important thing might this metric overlook: recognition, trust, differentiation or future demand?
- What must stay consistent enough in our brand for the market to be able to recognize us at all?
When to optimize and when to do something new?
Optimization is excellent at improving a system that has a good foundation. If you have a clear offer, a good product, a wide enough audience and a recognizable brand, it makes sense to look for better creatives, a faster website, smarter budget allocation and less friction on the way to purchase.
But if growth is stalled because the offer is not interesting enough, because the market cannot tell you apart from competitors or because you only talk to people who already know you, an extra five percent of optimization will not solve the problem. You do not need yet another button variation. You need a new idea, a new perspective or a decision that the dashboard cannot yet suggest.
Good marketing therefore needs both: the discipline of measurement and enough courage to sometimes do something whose result cannot be fitted in advance into a neat Excel cell.
Optimize the execution. But do not reduce strategy to a single number.
The lesson of the story: better numbers are a means, not a goal
Marketing without measurement is guesswork. But marketing that believes only what it can measure immediately can become very precise and at the same time very short-sighted.
The best teams use data to make better decisions, not as a substitute for them. They optimize where the goal is clear, test when they can learn something from the result, and at the same time protect the things that take time: the idea, recognition, trust and future demand.
So the next time a dashboard shows a green arrow, do not ask only whether the number is better. Also ask: are we better because of it as a brand and as a business?
