The story below follows a small invented team, the sort of group you might find behind any modest online business. Its members ran a regional gift-box shop with a handful of staff, and in one busy autumn they faced the same question that hits most growing companies: which of our pending decisions can we afford to get wrong? What follows is a narrative about reversible decisions, told with the figures kept out of it, so you can concentrate on the reasoning and see how the evidence changed their minds.
The setup
The owner, Marta, had three open questions on the whiteboard. The shop’s storefront software was creaking and might need replacing. A courier had offered a generous rate in return for a long exclusive agreement. And a popular supplier of hand-made candles wanted to become the shop’s only candle source. Everyone on the team had an opinion, and for two weeks nothing happened.
The team’s operations lead, Joris, finally suggested a different way of discussing the list. Instead of asking “which option is best,” he asked “which of these can we undo, and how fast?” That one change turned a messy argument into a tidy sorting job.
Step one: sort by the price of reversal
Joris wrote each decision on a card and asked the team to estimate three things: the cost of undoing it, the time needed to undo it, and how soon they would see whether it was working. The results were revealing.
- Storefront software: moderate cost to undo, but their product data could be exported at will, and they could run a second storefront in parallel for a trial. Reversible with effort.
- Courier agreement: the exclusive term would run long, with penalties for leaving early. Slow and expensive to undo. Close to a one-way door.
- Candle supplier: trivial to undo if the first order was small. The risk was only a few unsold boxes. Reversible, cheaply.
Notice what the cards did. Before sorting, the courier deal and the candle deal had looked equally urgent. After sorting, they belonged in different categories. The team’s own habits had been treating all three as the same weight.
Step two: act on the cheap one right away
The candle supplier was the obvious place to begin. The team placed a small first order, built it into one seasonal box and watched. They learned something useful within ten days: the candles sold well, but a few arrived with chipped lids. That was exactly the sort of information that a long discussion could never have produced. Because the first order was small, the chipped lids cost almost nothing, and the team could ask the supplier for tighter packaging before placing a larger one.
It is worth pausing on the evidence here. The team did not trust the supplier’s sample photographs or the glowing words on a sales call. They trusted what happened in their own warehouse. A reversible step is a way of buying that sort of evidence cheaply. Many small businesses make the mistake of paying for confidence in advance, with long negotiations, when they could have purchased it in arrears with a trial.
Step three: stage the middle decision
The storefront question took longer. Marta did not want to move the whole shop at once. Instead the team cloned the product catalogue to a new platform, sent a small share of visitors there for a few weeks and compared how the checkout behaved. They kept the old storefront alive the whole time. When a payment bug appeared on the new platform, they simply stopped sending visitors and fixed it. No customers were stranded and nobody had to apologise.
For teams in similar spots, the guidance on how smaller firms plan to automate their routine work echoes this staged style, with early steps small enough to take back.
A side benefit appeared that nobody had planned. Because the old storefront stayed live, the team’s anxiety dropped. People stopped dreading launch day, since launch day was no longer a single moment with no return. Staff who had been privately worried about the switch started pointing out small problems early, rather than hiding them until it was too late. Reversibility, it turned out, is also a morale tool.
Step four: slow down on the permanent one
The courier agreement was the one-way door, so Joris treated it very differently. He asked for the contract in writing, requested a short pilot period at the same rate, and listed the evidence the team still lacked. How did the courier handle damaged parcels? What happened during a peak week? Could the shop leave if service slipped? The courier refused to shorten the term, which was itself a piece of evidence. The team declined the exclusive deal and chose a shorter arrangement at a slightly higher rate.
That choice looks conservative, but it was actually the bolder one in terms of staying flexible. They paid a little for the option to walk away. Months later, when a regional competitor of the courier improved its service, the team was free to move some parcels over without penalty.
What an outside lens added
One of Marta’s friends, who builds software for a living, pointed out that engineering teams do this routinely. They release new features behind switches, test on a fraction of users and keep a quick way to return to the previous version. She mentioned an Amsterdam consultancy that advises and trains teams on DevOps, SRE and cloud delivery, simply as an example of where that discipline is taught. The team did not need to hire anyone. They only needed the idea, and it translated neatly into parcels and candles.
What the evidence says about their approach
Looking back, the team noticed a pattern in the outcomes. The quick, reversible decisions produced fast learning. The staged decision surfaced a genuine checkout defect while only a sliver of traffic was exposed to it. The permanent decision was delayed, not abandoned, and the delay purchased better terms. Nothing here required special intelligence. It required honesty about which decisions were which, and a willingness to buy information before buying commitment.
Their approach is not free of cost. Running two storefronts at once took extra work, the candle trial delayed a big seasonal order by a week, and the shorter courier contract was pricier. Each of those costs was small and visible, which is the point. The alternative was a hidden, potentially large cost that would have appeared only after everything was locked in. If you want a companion read on watching money closely while you do this, the article on finding spending leaks and protecting your cash buffer covers the habits that make small experiments affordable.
Lessons you can reuse
- Sort before you debate. A card per decision, with the cost and time of reversal written beside it, calms most arguments.
- Run the cheapest reversible test first. It teaches you more than another meeting.
- Keep the old arrangement alive until the new one has survived a busy week.
- Treat a supplier’s refusal to offer flexibility as information.
- Pay a modest premium for exits when the alternative is a long, penalty-heavy term.
- Write down what you would do to undo each step, and who would do it.
The shop’s whiteboard looks different now. There are still three columns, but they are labelled “try this week,” “stage over a month” and “needs evidence first.” New ideas land in the first column by default and have to earn their way to the third. If you want a planning frame to go with the sorting habit, the walkthrough of mapping digital growth in manageable phases fits nicely.
Closing thought
The lesson of Marta’s autumn is not that caution is good or boldness is bad. It is that the right amount of either depends on the price of reversal, and that you can usually lower that price with a little design. Shorten terms, shrink the first order, keep the old way alive, and write down the exit. Then, when a truly permanent choice arrives, you will meet it with real evidence rather than nerves.
