Precision Tuning

This is a transcript of episode 245 of the Troubleshooting Agile podcast with Jeffrey Fredrick and Douglas Squirrel.

Squirrel describes how he deals with overly vague strategy descriptions: tune them until they’re specific enough that someone could disagree with them. He and Jeffrey discuss how this applies beyond strategy, the value of asking probing questions, and the productive conflict that can arise from getting past the “motherhood and apple pie” answers.

Show links:

• Podcast episode, Adversarial Analysis

Listen to the episode on SoundCloud or Apple Podcasts.

Introduction

Listen to this section at 00:11

Squirrel: Welcome back to Troubleshooting Agile. Hi there, Jeffrey.

Jeffrey: Hi Squirrel. So last time we were talking about “adversarial analysis” and we didn’t foreshadow this at the time, but when you and I were talking and you introduced that idea, you also introduced something else: “precision tuning.” I love that name and I love how they go together. I wanted to talk about them both last time, but you convinced me that we should talk about them separately. So here we are.

Squirrel: I thought that was your idea! I think we should have an adversarial analysis-no, I’m just kidding. Okay, let’s talk about precision tuning today, that sounds great.

Jeffrey: So what is precision tuning and how does it follow from the adversarial analysis if at all, or is it just something entirely different?

Squirrel: No, no. They’re very closely linked because they’re two cures for common errors that I find people make in their analysis of strategy. Something I’ve taken to doing in a more structured way recently is asking people I’m working with to summarize their company’s strategy, and it’s surprising how challenging they find that. That’s very insightful for me. It tells me a lot about how good other people are at communicating it. When they eventually come up with something, then I tell them, “Well, this part is overly specific and it’s got something wrong with it that you haven’t analyzed.” That’s what we talked about last time, understanding what you could mess up about this particular strategic goal. How someone could do the wrong thing following the instructions. But then they often have some other stuff I like to describe with a phrase very familiar in the United States, though not elsewhere: “motherhood and apple pie.” This refers to politicians when they say things like “my mother was wonderful,” and “we all should have the right to apple pie,” and “this is a land of opportunity!” They can talk for hours and never say anything.

Jeffrey: Right. Becasue no one’s against motherhood. No one’s condemning apple pie.

Squirrel: Mothers: terrible! Apple pie: poisonous!

Jeffrey: It’s safe, generic, meaningless.

“We’re Going to Profit”

Listen to this section at 02:30

Squirrel: Exactly. That’s what I often find people have in their company strategy. Often I’m talking to technical people, so they might have a more technically-focused action like, “we’re going to make our code better.” “We’re going to make sure that our application scales,” “we’re going to ensure that customers are happy.” These are all “motherhood and apple pie.” You couldn’t possibly disagree with them. “I want to write worse code. I want unhappy customers.” This just doesn’t make any sense. So the problem is that they’re in need of precision tuning. In other words, this is so vague as not to be a guide to action. You need to make sure that it actually tells you something. It needs to be something someone could disagree with.

Jeffrey: Hmm. I like that. Maybe it has to be specific enough to put some things out of bounds? Otherwise, you’re not really being very precise in what your strategy is, because you can’t then say what it isn’t, or something along those lines?

Squirrel: That’s a big part of it. Another ineffective approach might put something out of bounds that everyone would agree with. “We’re going to eliminate all bugs.” So, out of bounds is “leaving bugs in” or “writing more bugs.” But you haven’t said very much that someone could disagree with. If you said “we’re going to slow down our releases so that they only come out once a year and we only add 100 lines of code every every month” that that might cause people to disagree with you. That would be very specific. That’s what NASA does to build spacecraft software and it works really great for them. Probably not so good in your ecommerce t-shirt store.

Jeffrey: Right. I like this, I think I’m already pre-sold, which hopefully doesn’t make precision one of those motherhood and apple pie topics. I just so often find the problem is that people are imprecise in what they intend. I’m talking to people at all different levels, sometimes it’s about strategy, but it can be at all different layers. And the ones I’ve been dealing with recently have been product managers, and what I am telling them is “be very, very aware of and afraid of vaguely positive statements.”

Squirrel: Yeah, those are the worst. Vaguely negative, you’re maybe okay. Vaguely positive, motherhood and apple pie, it’s really dangerous because it puts you to sleep.

The Discomfort of Discovery

Listen to this section at 05:02

Jeffrey: I’ll give you an example. You’ll be working on a project and someone will say “we’re working on that and it should be done in two weeks.” Superficially this sounds very positive. But you notice that it’s not really much of a commitment. What do we mean by done? So I’ve been training the person to ask some more specific questions. So I might say something like, “okay, today is a Wednesday. So does that mean customers will be able to use it on the Wednesday two weeks from today?” Suddenly you’re being more specific. The answer could be “yeah, that’s what we expect.” Then you can have follow-up questions like, “well, that’s great. What’s the probability that you would put on that? Do you think it’s 90% or do you think it’s like 50%?” You start getting a more detailed conversation.

Squirrel: Which tests will it pass, what users will be able to use it, what geographies will it be rolled out in.

Jeffrey: Exactly. What exactly are we saying will be done and when? Also asking questions like, “that’s great, but do you have an idea of why it’s two weeks? Is there something else keeping it from being faster? What prevents it from being one week?” It could be like, “oh, we can’t do it in one week because we’re also trying to do X, Y and Z.” “Oh, so if we stop doing X, Y and Z, you say we can have it next week?” “Well yeah I suppose.” “Fantastic.” You’re getting into specific trade-offs that exist as opposed to what’s so dangerous about that vaguely positive message: it is reassuring.

Squirrel: It doens’t create any drama or worry or disagreement.

Jeffrey: Exactly. People relax. “Oh, that’s great. It’s being worked on. Fantastic. Oh, it’s two weeks? We even have a timeline!” What happens in practice is you get to two weeks later and it’s like, “so where is that thing that you said would done?” “Oh, it’s taking a little bit longer than we expected. Something came up and it looks like, it’s probably going be another week and a half or another two weeks.”

Squirrel: “Actually we released it. It’s great. It’s live to internal customers. It’s fantastic.”

Jeffrey: “To internal customers?”

Squirrel: “Yeah! It works for any user whose name starts with ‘Q.’”

Jeffrey: Yeah, yeah. “Just like we intended. Just like we expected.”

Squirrel: “That’s what I meant when I told you it’d be ready in two weeks.”

“What are you Doing?”

Listen to this section at 07:36

Jeffrey: That perfectly fits. That might sound dramatic to people, but that’s very realistic. “Yeah, it’ll be done and it will make no difference to the end users at all whatsoever. But that thing that we mentioned two weeks ago is actually done.” So that is if you just allow the vaguely positive thing to pass. This is why I’m really excited to talk about precision tuning. I think this is something people can go and apply immediately in their environment. They can listen to what people are saying, listen to what they themselves are saying. I remember when I first joined Tim Group where you were CTO. There was a phrase that some people would use: “the language of commitment,” and that was kind of an antidote to this imprecision, “who is doing what by when.” This is a similar idea. You can listen to what you’re saying, listen to what others are saying. Do we really understand what’s being committed to? Is anything being committed to or are we hoping that everyone feels good? Because that’s what I think we want to avoid.

Squirrel: I’ll just clarify here that the specificity doesn’t have to be in terms of numbers or something that’s digital. So it would in fact be perfectly fine to say “90% of our the customers that we talk to when we randomly phone them will tell us that the new the new login experience is better.” Now, that’s a qualitative piece of information, but it passes the precision tuning test because while you might say a bit more about how you select them and how many will you phone and so on, but it’s much more specific than “our customers will be happy,” which could be satisfied by anything. “I ran into one yesterday and they smiled. Therefore they’re happy.” I’ve given you something specific that you can target and you can ask, “Is it happening?” There was a test that someone told me about long ago. It must come from somewhere, if listeners know, they can tell me. This was a test for telling the difference between a virtual item and a real item, because it gets very confusing, you talk about servers and machines and so on. Are those real? So somebody said, “Well, if I dropped it on my foot, would it hurt?” That’s a good way to know. If it’s in the cloud it doesn’t hurt. If it’s a real thing in a server room, it hurts. So a similar test here would be if “I followed you around, could I see you doing that? Could I observe that? Is that something I could take in with my eyes or ears or nose?” If that’s the case, then I’ve got something that’s specific enough. I can actually test it. But if it’s motherhood and apple pie, how do you know?

Well-Meaning Pointlessness

Listen to this section at 10:40

Jeffrey: One thing that’s worth saying, is why these vague statements come up and why they get passed along. It so often is that people want to feel good and they don’t want to make other people feel bad. Doesn’t it kind of make for a bad relationship? If I’m asking someone all these questions, am I now turning adversarial with my colleagues? Isn’t that a bad thing?

Squirrel: Well, if you do it with the right intent and with the right communication, absolutely not. You may produce productive conflict, which longtime listeners will know we’re big, big fans of. So it may be that somebody says, “Man, I haven’t thought about any of that. Why don’t you go check?” Then you’ve got a really useful discussion to have. It may not be pleasant, but it will probably produce something like you going and being the one to ask all the customers whether they’re happy or not. That could be a very good outcome.

Jeffrey: Actually, I really like that. When they come back like, “no, I wasn’t going to do that, and that’s not my job. Of course I wasn’t planning to do that.” That’s actually a fantastic thing to find. I like that example because “there’s other people you should be talking to, not me,” is a great discovery. The idea of learning something that I didn’t know is exactly why I’m asking the question. This imagined other person is probably kind of annoyed, like, “how can you not know that?” But that’s the point. We don’t know what we don’t know, and the people we’re talking to don’t know what we don’t know. They’re making assumptions, assuming that when they say, “Oh, this will be done in two weeks, that of course we know that naturally there’s 15 other things that need to happen after that. Isn’t that obvious?” But no, it’s not necessarily obvious to us, but how do we discover this unless we’re willing to be brave enough to go ask these questions?

Squirrel: And the result of the productive conflict is usually that you have this learning of something that you didn’t know much faster. Because you will learn it. In two weeks you’ll come along and you’ll say, “Is it done?” They’ll say, “Oh, yes, the customers are really happy and everything’s fantastic. We asked four people in marketing. It was fine.” “Marketing!? What do you mean? How about actual customers?” You’ll discover it then. So it’s not like the information will never come to you. But would you like it today or in two weeks?

Jeffrey: Today. These things sound like you’re being very particular, asking people these questions. The person I was coaching said they actually found it helpful to the other people when they started asking them. It helped them to think a little bit more precisely and they’d go “Oh, yeah, you know what? I forgot this other thing. Oh, there’s also this other part.” And he would ask more of the questions. “Okay, so given those other pieces now, what do we think this does to what you’re expecting and what’s your level of confidence? Do you think there’s anything you’re missing?” “Oh, wait, no, there’s this other bit.” And in actually, it made for a much richer plan. It put everyone in much better footing. You end up actually in a collaborative space as long as as you act with the right intent. Your goal here isn’t to play gotcha. You’re not trying to catch people in a mistake and be like, “Aha. So you didn’t think this through. I knew it. You’re a bad collaborator.” That’s not the point. You’re trying to be productive in saying “let’s make sure that we have a clear shared vision of what we’re going to do.” If someone has already done that work and you ask these questions, guess what? They just answer them instantly. They’re like, “Yes, of course you can do it on a Wednesday two weeks from now, because we’ll have released it on on the Tuesday and here’s all the work we’ve done.” They’ll have those answers up to hand. It doesn’t cost anything. It doesn’t make for a negative interaction if they’ve already done the precise thinking themselves.

Finding Truth

Listen to this section at 14:24

Squirrel: And that brings us to something that we really like, which is the genuine question. So what we’re describing here is asking questions that you really want the answer to. A surprising answer would be valuable. So the scenario where you’re pounding on the table saying, “haven’t you thought of this? Aren’t you doing this? Why aren’t you doing this?” Those are leading questions. Those aren’t genuine questions, but a genuine question sounds like “I want to check that we’re on the same page. I think that it might mean that it’s released on Wednesday and that I can go and demo it on that afternoon. Can I book a demo?” That would be a genuine question to which the answer “no” would be really interesting. You would like to hear it and you would be supportive of it. That’s the attitude to take into the questioning that you might do as part of your precision tuning.

Jeffrey: Yeah, that’s right. The question here of level of confidence is one I find myself asking more often now, and it ends up being very valuable, because the difference of someone at a 50% level of confidence for an answer versus 90 or 95, that’s really useful data. An answer of 50% is totally fine. If someone says, “well, I think we’re like 50-50. We don’t know, we’re not 100% confident, but we think that we have a good shot,” that’s actually good to know.

Squirrel: I can work with that. But if I think it’s 100% and it’s actually 25%, not so good.

Jeffrey: That’s right.

Squirrel: Thanks, Jeffrey.

Jeffrey: Thanks, Squirrel.