The hidden cost of every new tool

Every tool that enters a company promises to save time. Almost none, though, says how much time it costs.

I am not talking about price. Price is the easy part: it is written on a quote, you compare it with other quotes and you approve it. I am talking about the time the tool absorbs after the signature, that is, the hours it takes to learn it, to connect it to what was there before, to decide who uses it and who doesn't, and to understand why the figure it shows doesn't match the figure the previous system showed. This time appears on no quote, and precisely for that reason nobody puts it on the table at the moment the decision is made.

What happens in companies

Let me take a case I followed personally. A manufacturing company in north-east Italy, about sixty people, healthy revenue. In four years it had introduced a CRM, that is, the system where customers and deals are recorded, a platform for e-mail campaigns, a system for booking sales appointments, two tools for analysing website traffic and, finally, a dashboard that was supposed to gather the data from all the others on a single screen. Each choice, taken on its own, made sense. Each supplier had shown a convincing demo and a success story.

When I started working with the management, the question we asked ourselves was not which tool to remove. It was another, more useful one: how much of the sales week goes into making all these systems talk to each other. So we counted. Between data entered twice, exports, contacts to reconcile and meetings to work out which number was the right one, the cautious estimate was one day per person every two weeks. Across eight salespeople that makes four days a week. Nobody had ever seen them, for a simple reason: no invoice carried them.

The count was nothing sophisticated. For two weeks each salesperson noted on a sheet, at the end of the day, how many times they had copied a figure from one system to another and how long they thought it had taken. The sheet was deliberately rough, because counting precisely would have required one more tool, and that would have been an elegant way of not understanding the problem. The number that came out was cautious by construction, since nobody writes down the thirty seconds lost finding a window on the computer again. And yet it was more than enough to see the problem.

There was also a cost the sheet could not record: who was paying it. The double entries were not done by the sales director. They were done by the two youngest salespeople, the ones with the least power to refuse and the greatest need to prove themselves. The time a tool absorbs, in other words, is not spread evenly: it flows downwards, where nobody sees it from the meetings in which the next purchase is decided.

The dashboard, the one that was meant to solve everything, was the most expensive tool in terms of time. Every figure it showed had to be checked against the source, because management did not trust what it saw. The monthly meetings on the numbers took twice as long, because half the time went into deciding which number to believe. It is a rule that always holds: a tool you do not trust does not remove work, it doubles it.

Why it happens

There are three pieces of evidence that help to understand why. They are not my inferences, but the results of scientific research on how we make decisions. My inferences come afterwards, and I flag them as such.

The first concerns how we estimate the future. Daniel Kahneman and Amos Tversky, two psychologists who spent decades studying how we make decisions, described a precise tendency: we underestimate the time and cost of projects we are about to start, even when we know perfectly well how badly similar projects went in the past. They called it the planning fallacy.

What does this have to do with tools? When we evaluate a new one we look at the demo, that is, the version of the future in which everything works. We do not look at the three previous tools, which took twice the expected time. This is not carelessness: it is how our brain builds a forecast, starting from the case in front of it and not from the statistics of similar cases.

Kahneman also proposes a remedy, which he calls the outside view. Before making an estimate, it pays to look for similar cases already lived through and start from how they went, rather than from your own plan. Applied to tools, the remedy becomes a simple question: the last three tools we introduced, how much time did they take compared with what was planned? The answer is almost always uncomfortable, and that is precisely why almost nobody asks it.

The second piece of evidence concerns attention. John Sweller studied what he calls cognitive load, that is, the amount of information we can hold in mind while we work. His research shows that this capacity is limited and that every switch from one system to another uses up part of it. This is not time you can measure with a stopwatch: it is the quality of the decision we make while moving from one window to the next, and it gets worse without our noticing.

The third piece of evidence explains why tools, once in, never leave. Hal Arkes and Catherine Blumer showed that we tend to continue an activity in which we have already invested money, time or effort, even when continuing no longer makes sense. It is called the sunk cost effect, because the money already spent does not come back either way, and yet it weighs on the decision. The person who signed the dashboard contract finds it harder to switch it off than someone looking at it from outside, and not out of pride: switching it off turns an investment into a mistake, and our mind prefers a cost that goes on to a loss that closes.

What changes for the decision-maker

If the initial estimate is always optimistic and the cost grows with every switch between systems, then the decision on a new tool cannot be made by looking at the tool. It has to be made by looking at what the company already has.

In the north-east company the final choice was not a new dashboard, better than the previous one. It was removing two systems, keeping the CRM as the only source of sales figures and accepting that some website analyses would no longer be done. Management recovered about three days a week of sales time. It did not reinvest them in another tool: it left them to the salespeople, to phone customers.

Switching off the two systems took one unpleasant meeting, in which someone had to say that the purchase made two years earlier had not paid off. It was the hardest part of the whole job, and technology had nothing to do with it.

The question to ask before every purchase, then, is not "how much time does it save me". It is "how much time does it cost me, and who pays it". The first question always has a reassuring answer, because the seller gives it. The second has no answer until you count the real hours of the people who will have to use the tool, and it pays to count them before the signature, not after.

Counting before the signature can be done, and it costs less than it seems. One week of counting, with the same rough sheet as the north-east company, tells you how much time the existing systems absorb today, that is, the base the new tool will be added to. If that base is already high, the new tool will not reduce the time: it will move it, usually towards the people with the least voice to point it out.

Then there is a decision worth taking on the day of the purchase, and not two years later: under what conditions the tool will be switched off. A date to review it, a result to reach, a person with the mandate to say it has not worked. Not out of distrust of the tool, but to take from whoever signed the burden of having to admit a mistake. If the rule was written beforehand, switching off is not a failure: it is the application of an agreement.

There is also a simpler criterion, which holds in most cases. If you cannot say which existing tool will be switched off when the new one comes in, the new one is probably not needed. What is needed is to understand why you cannot switch off the old one.