--- title: "When the goal is clear and the process is not" description: "Plenty of companies know exactly what result they want from an AI agent and have never measured the problem. The four questions that reveal it in five minutes." url: https://studiochat.io/blog/el-problema-sin-dimensionar locale: en translations: { es: https://studiochat.io/es/blog/el-problema-sin-dimensionar.md, pt: https://studiochat.io/pt/blog/el-problema-sin-dimensionar.md } site: https://studiochat.io --- [ Back to blog](/blog) There is a type of sales meeting that starts very well and ends with nothing. The customer explains what they want and explains it perfectly. They want to recover more debt. They want to sell more. They want their team to stop answering the same thing twenty times a day. There is no ambiguity at all about the result. And then you ask four questions and there is no answer to any of them. ## [The four questions](#the-four-questions) They are always the same, and they are not technical questions. **How many are there.** How many cases, orders, complaints or queries come in per month. Not a range from a thousand to ten thousand: a number. **How long each one takes.** From arriving to closing, and separately, how many minutes of someone's time it takes. Those are two different numbers and both matter. **Who does it today.** By job title, not by department. If the answer is that several people do it when they can, we already know something important. **How it ends.** What share reaches the result you are after and what share falls away along the way. A company that can answer all four has a sized problem. A company that cannot answer any of them has a goal, which is not the same thing. The first two are enough to see the size, because the problem is an area: cases per month across, minutes per case up. The problem is an area: cases by minutes Cases a month: 240 Cases a month: this number is measured Minutes a case: 18 Minutes a case: this number is an estimate Cost of an hour: USD 14 What you said: 72 hours a month, USD 1,008 a month, USD 12,096 a year. With one side unmeasured, the problem sits between 42 and 122 hours a month. End to end that is 2.9 times over, and you do not know where in the range you are standing. That is not a small problem or a large one: it is an unsized problem. Measuring the missing side costs a week of writing things down and is worth more than any proposal you ask for now. The amber dashed axis is the side you marked as an estimate. Tap that control's button and watch the dashed rectangle shrink until it disappears. Both sides are yours. The unmeasured range assumes an estimated number can be 70% above or below, which is what shows up when somebody finally counts. You can switch off either side, to mark the one that is not measured. What appears then is a second dashed rectangle, which is the range of what is not known. When that rectangle is bigger than the solid one, there is not yet a big problem or a small one: there is an unsized problem. The numbers are whatever you put in: there is no measurement of a client of ours here. This object also has [a page of its own](/the-problem-rectangle), in case you want to send just that. ## [Why this is not the vendor's problem to skip](#why-this-is-not-the-vendors-problem-to-skip) The temptation, on the selling side, is to treat this as an objection and carry on anyway. The meeting continues, the proposal gets written, it gets signed. And the problem turns up whole six months later. Without a baseline there is no way to know whether the agent improved anything. If nobody knew how long it used to take, nobody will be able to say whether it now takes less. The argument about whether the project worked ends up settled on impressions, on who spoke loudest in the renewal meeting, or on whether somebody liked the tool. The vendor loses that argument every time, even when the agent performed well. Because the only thing that can be demonstrated is what was measured beforehand. ## [The awkward case: a clear goal hides the gap](#the-awkward-case-a-clear-goal-hides-the-gap) What makes this hard to spot is that a well formed goal sounds like a well built process. When someone says **recover the overdue debt**, the sentence is so clear that you assume there is a process behind it. Sometimes there is none. There is no list of who owes what, no order of who gets written to first, no text that gets sent, no place where the reply gets recorded, no payment link. What there is is a person doing what they can and a goal written on a slide. An agent on top of that does not automate the process. It gives system shape to something that never had one, and the decisions nobody made end up being made by whoever writes the instructions. Sometimes that turns out well. It almost never turns out the same as what the company would have decided had it sat down to decide. ## [What to do if you are the one who has not measured it](#what-to-do-if-you-are-the-one-who-has-not-measured-it) If you are on the company side and you read the four questions without being able to answer two of them, the conclusion is not that you are not ready for an agent. It is that there is a week of work worth doing first, and it is cheap. You write down four things per incoming case: when it arrived, who took it, how long it took and how it ended. Two weeks of that gives volume, time per case and resolution rate. With those three numbers, any conversation with a vendor changes tone, because you are no longer arguing about whether the problem is big: you are discussing which part of it can be taken. And that week is cheap next to what it costs to find out, six months later, that there is no way to tell whether the project was worth it. Frequently asked questions ### What does an unsized problem mean? It means knowing what result you want and not knowing the size of what has to be solved to get there. A company can say with total clarity that it wants to recover more debt or sell more, and be unable to say how many cases that is a month, how long each one takes today, who handles them or what share of them end well. ### Can an agent project start without the problem being measured? It can, and it usually ends badly for a concrete reason: without a baseline there is no way to know whether the agent improved anything. Six months later the argument about whether it worked gets settled on impressions rather than data, and the project is not renewed. ### How long does it take to measure a process before automating it? Far less than it looks. In most cases two weeks of writing down four things per incoming case is enough: when it arrived, who took it, how long it took and how it ended. That already gives volume, time per case and resolution rate, which is nearly everything you need. Share [ X](https://x.com/intent/post?text=The%20goal%20is%20perfectly%20clear%20and%20the%20process%20does%20not%20exist%3A%20how%20to%20spot%20it%20before%20the%20meeting&url=https%3A%2F%2Fstudiochat.io%2Fblog%2Fel-problema-sin-dimensionar) [ LinkedIn](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fstudiochat.io%2Fblog%2Fel-problema-sin-dimensionar) [ WhatsApp](https://wa.me/?text=The%20goal%20is%20perfectly%20clear%20and%20the%20process%20does%20not%20exist%3A%20how%20to%20spot%20it%20before%20the%20meeting%20https%3A%2F%2Fstudiochat.io%2Fblog%2Fel-problema-sin-dimensionar)Copy link Ignacio Puig Moreno Ignacio Puig Moreno · Co-founder · Business, StudioChat Want agents like these working for your team? [Talk to us now](https://cal.com/studiochat)