This text is written by Elina Chung, a founder of Elina Chung Oy and a member of Crazy Town Jyväskylä and the EriCa ecosystem.
A decision made on taste
A few years ago I watched a product decision get made because someone liked the way Instagram looked.
Someone senior wanted it to feel more like a consumer app. He would point at the Instagram logo as the model for ours. It was a B2B tool for documenting processes and organising how teams structured their work. At the time, nobody had settled who the ideal customer was, or fully agreed on what the product was for.
Wanting a well-designed product is reasonable. What stood out was the reasoning behind the specific choices: not what would help a customer use the thing, but personal taste. Decisions got made on preference, not on what the product could do, or what it was worth to the people meant to buy it. The question of who it was for had been left open. The strongest opinion in the room filled it.
The company was a scale-up, still testing the market, and it later went through a significant round of layoffs.
What was missing wasn’t design sense. It was anyone asking what we actually knew. The influence was real, and it had been earned somewhere: in starting the company, in putting money behind it. But it was being spent somewhere it hadn’t been earned, on what an undefined buyer would want from a product we hadn’t finished defining.
What engineering already does
Engineering teams have a process for exactly this. After an outage or a failed experiment, a blameless postmortem asks what was known at the time versus what was assumed, and nobody asks whose fault it was. Plenty of companies run that process for engineering. I have never seen one run it for a lost deal, a launch nobody noticed, or a market they entered and quietly backed out of.
I have seen the other version too. At a company I joined later, the ad and agency spend was substantial, but leads either weren’t coming in, or the ones that did weren’t worth much. The agency’s answer was to spend more. Nobody was asking whether the leads were any good.
We stopped spending. Before touching anything else, we worked out who we were selling to, and found three groups sitting under one label, ”IT decision makers”.
The person evaluating it was reviewing it on technical merit. The people who would use it every day were asking something narrower: does this fit the way I already work, and how does it sit alongside the stack I already have. They rarely thought about how it would benefit the business. The person signing it off understood the product too, but valued it differently again, on what it did for the organisation’s governance and risk position, and ultimately the bottom line.
Same product, three different questions, and we had been sending all of them the same words.
We split the messaging to speak to the evaluator and the person signing it off, and deliberately left the users out. In firms that size they weren’t making the decision. They were stakeholders in it, and part of the feedback loop once the product was in, which is a different job needing different words. Deciding who not to talk to is part of the same exercise.
With that settled, we worked directly with the agency on the copy. Qualified lead volume improved by roughly 30% over the first year. Not a dramatic number, but a real one.
The three questions
Every time I have done this, it has come down to the same three. Who are we really selling to. What are we selling them. Where are those people looking. Answer those honestly and most of the operational decisions, the channels, the budget, the agency, follow from them. Get them wrong and nothing downstream rescues it.
The honest part is the difficult part. The teams I’ve asked can answer all three quickly. That speed is the problem, because the quick answer is usually the assumption rather than the knowledge. A useful review separates the two: what did we know, and how did we know it, against what did we believe because nobody had questioned it. Then one follow-up for each belief. What would we have to see to know this is wrong?
None of this needs to be a project. An hour with whoever was in the room, no names, no fault, one page of notes. The output isn’t a document. It’s one assumption you have agreed to test before the next decision.
A version, not a conclusion
One more, and this one ends better than it starts. A company with a premium product and a strong position at home expanded into a new region, using largely the playbook that had worked there, slightly reframed. Positioning and messaging had never been worked out for the new market specifically. Locally, the sales team was under constant pressure to discount and undercut competitors, because the commission structure rewarded the number of deals closed, not the value of each one. Over time that wore down the premium positioning, and marketing’s return fell to almost nothing.
Nobody set out to damage the brand. The incentive structure rewarded one thing, volume, while the actual goal, healthy and valuable deals, was never built into how success was measured.
You could hear it in how people described their own work. When sales raised the quality of the leads, one marketer’s answer was that their job was to bring in volume, not to qualify it. It landed as arrogance. It was also an accurate description of the role as it had been defined. Volume was what the work was measured on, so volume was what the work produced. That’s what an incentive problem sounds like from the inside: no decision anyone can point to, just people doing the job they were given.
The correction was expensive, and it was quick. Half the marketing team went, along with some of sales. It’s worth noticing where the cut landed. The function that couldn’t show a return was the one that shrank. I was in it.
Then the direction changed. The discounting stopped. The money that had gone into undercutting competitors went into building credibility with voices the industry already trusted, to build the standing it had assumed it already had. The company repositioned itself and carried on. Getting it wrong wasn’t fatal. Not noticing could have been.
Some companies never do any of this and survive anyway. If you sell something inexpensive to a very large number of buyers, you can afford to guess repeatedly and let volume find the answer for you. Unfortunately that isn’t available to you if the sales cycle is long, the pilots are expensive and the number of possible buyers is smaller. Then you get a limited number of attempts. Decide beforehand what each one is meant to test, or it will teach you the wrong thing as easily as the right one.
The one question worth adding
Treat your view of the market the way you already treat your product: as a version, not a conclusion. A product still finding its buyers is in beta. So is everything you believe about those buyers, including the price you put in front of them. A price set for a market you haven’t entered yet is an informed estimate, not a fact: you make the best call you can, then test it and adjust quickly when the evidence says otherwise.
Three companies, the same root cause each time: a belief about the market treated as settled fact instead of something to check. The method isn’t complicated, and it isn’t new to you. Pause, ask what was known, review without blame, adjust. You already run it on the technical side of the business. The harder question is what stops it reaching the room where decisions about buyers, positioning and price get made.
Next time something you were sure about turns out to be wrong, the useful question isn’t whose fault it was. It’s whether you knew, or you assumed.
Elina Chung runs Elina Chung Oy. She works on international B2B market strategy: who you sell to, what you sell them, and why it lands. Twelve years in-house in B2B technology, including SaaS, cybersecurity and health technology, in Australia, Asia-Pacific and EMEA. She mentors on EriCa Reactor.
