How to Make Brilliant Decisions: Matt Hancock

0:00I'm Matt Hancock and I am working to make sure that technology around the world can be used to make lives longer, to save lives.

0:08Matt Hancock:Making decisions under pressure. Yeah. How do you think about that? So I'm a quick decision taker, so I know that on balance I will take decisions quickly and I err on that side. But taking decisions too quickly is a weakness. Is that ego, Matt? Like you've got all of the information, all of the data points at this decision, but actually I'm more important. Yeah, I think that was the wrong decision. I like to think of myself as a rationalist and an optimist. Humans find any group over 150 impossible to consider. I'm a big fan of Mr. Beast, right? In Beast games, people form into groups and they will stick together and they will act irrationally for themselves in the better interest of a smaller group.

0:54But they find it hard to act against themselves in the interests of the whole group.

0:59Matt Hancock:Do you have any examples of situations where you had lots of information and there was still a struggle to maybe make those decisions? Let me try. Matt Hancock, welcome to the show. It's great to be with you. We always start the podcast off the same way. So who are you and what problem do you solve? So I'm Matt Hancock and I am working to make sure that the technology around the world can be used in the health system and can be used to make lives longer, to save lives. And I come from a tech background. I've obviously got deep experience in health. And I think that it's one of the areas that there's a huge opportunity, but it's hard.

1:45That's what I'm doing. What do you think in your lived experience makes you perhaps uniquely qualified to be able to talk on that subject? Well, I wouldn't say uniquely qualified, but I come from a technology background originally. The first job that I had was in the family data business. It's data business, you know, and back then it's had the strapline intelligent use of data for all 45 years of its existence. So my folks were ahead of their time. and that's in address management. You know, when you type your postcode into the internet and it brings up your address. Yeah. And we were the first to do that.

2:22Oh, wow. Yeah. So there's, and now that business is going strong in address data expertise. My first real job was decoding the Y2K bug in COBOL, the programming language, which is now pretty out of date. But that was, so I come from that background And then obviously I had a long time in government and I mostly did tech related jobs in the business department and then in the digital department and DCMS. And then I became health secretary being asked to bring the improvements in technology into the NHS and into the health system. And that's why I was given that job. And then obviously the pandemic struck.

3:10But we use technology a huge amount all the way through. So the combination of experience of the health system, but coming at it from a tech background, that's what I bring.

3:21Matt Hancock:Wow. Nice. So we want to get your expertise today when it comes to decision making. If you strip decision making right down to its first principles, where do you take it? Where do you take it? So I think about the ability to make decisions is one of the most important things in business and in government, for that matter, and in life. And there's so many facets to it, aren't there? But I think at the core of it is that you have to choose to make a decision. and thinking about how you make a decision is actually a really important topic for being better at making decisions and therefore getting better outcomes.

4:11Nobody makes decisions that are 100 % right, obviously. So taking a decision is a risk. But also not taking a decision is itself a decision when you're faced with an issue. And so thinking about how you go about that is I think critical and some people say I can't take that decision until I have all the information available but one of the things especially accelerating the modern world is you never have all the information you have to assess when you've got enough information and when you should when the benefits of taking a decision and closing down an option because if you don't say no to something then you're not taking a decision at all actually you're just keeping your options open and so the first decision about a decision is when to take the decision if that doesn't sound too convoluted.

5:02Matt Hancock:It's very simple but if you had all of the information available it probably wouldn't be a decision. Like the decision is already made for you. Well that's true in some cases when a decision becomes obvious but not all because lots of times there's trade-offs and it's a judgment call and in a way the more one of the things i found is the more senior you get the less obvious the the decision is because if a decision is obvious then it gets taken and somebody else if you know if a if a system if a company if an organization is structured well enough then the easy decisions will get taken at a lower level and it's the harder decisions that get raised to the top Do you have any examples of situations where you had lots of information and there was still a struggle to maybe make those decisions?

5:55Others. Endless. Endless. So let me try to think of an example. You know, a tough balancing act would be, for instance... Oh, let me try to think about it. OK, so in health tech, when we were taking the Data Protection Act through Parliament, right, I was the digital secretary at the time, but this matters for health tech. We had all the information on who should be responsible for a patient's data. That is the judgment there of who should be responsible and who should have the legal responsibilities. We didn't need any more sort of evidence or information to make the judgment on that. And the way that the GDPR was structured, it gave the authority for that and the responsibility to the GP.

6:59Now, with hindsight, I think that was the wrong decision. Right. I think that data should be owned by and controlled by the patient, not the GP. And the system, obviously, the health system at the NHS in England, obviously has responsibility for holding the data because the patient doesn't hold it. But it shouldn't be the GP who controls access to that data. It should be the patient. And that is an example of a decision that was a wrong decision of what went into that legislation. But it wasn't because of a lack of information. it was that it was a judgment call because there's pros and cons two sides to it and uh the wrong call where does bias come into decision making so someone's perhaps lived experience could be completely different maybe they're more risk of us or have a greater risk appetite yeah you've probably seen a lot of examples where those two things are perhaps butted heads when two people at the decision making table together yeah what have you learned so i think so i think that i I mean, there's so much in that question, right?

8:01So most decisions are a factor of the rational at the core of it, you hope, right? And then there's the human psychology, behavioral psychology of how a human takes a decision. And then there's the group dynamics. Because when we take a decision, we often take into account what somebody else has done or said and often deny that element of the decision-making even to ourselves, right? And some of the most interesting things in marketing are about how to get others to influence somebody's buying decision. So I love the example from, was it Panasonic? One of the TV makers in the 80s, right? And they sold through radio rentals.

8:55and radio rentals would do their own deliveries. It was days before that. You know, the delivery system was all structured as it is now. Radio rentals would intentionally go and deliver to the neighbour. And the neighbour would say, hello, I'm here for Mr Smith. He's got his brand new Panasonic television. Oh, wow, I didn't know that. And they'd say, no, I'm sorry, I'm not Mr Smith, I'm Mr Jones. Mr Smith is next door. That is an advertising signal, really strong. Yeah, yeah. Marketing signal, actually, yeah, to the neighbour. So that's one example. You know, there's some people who are brilliant on this, like Rory Sutherland, right?

9:35He talks about this stuff. And the behavioural psychology that goes into not just the... So building from the rational to the emotional, behavioural impulses of an individual, then out to the group dynamic. You know, we've all been in group decisions where one person saying something can flip the decision. You know, 12 Angry Men, the play, is a brilliant example of how 11 people can assume that the decision should go one way. And there's one holdout. And that holdout that gradually grows through the group until they come to the opposite decision. And if you hadn't had that one person, it would have gone in the opposite direction.

10:25And that's a sort of case study of this group behavioral dynamics. Wow, that's a really interesting insight. Almost like the interesting insight that not enough people like and subscribe to this podcast.

10:40Matt Hancock:If you want to help us fulfill our dreams of becoming full-time podcasters, please like and subscribe. That way, we can be happy. I wonder what you could tell us about almost like a lesson-sharing exercise between the way government makes decisions and the way business makes decisions. Are there things that both sides could perhaps learn from each other? Yeah, definitely. So at the senior level, politics is essentially non-hierarchical. There's 650 people elected to parliament. They are each elected for their constituency, and they have that seat in Parliament until the next election because they've been elected and it's their name on the ballot paper.

11:23There's then, of course, the party overlay on top of that and then 80-odd of those people end up in government running the country as ministers. Ultimately, although there's obviously a hierarchy from the Prime Minister down, to get something through you have to get a wide range of people to agree with it. And this is why some of the decision-making can be slow. And unless the prime minister is involved, can be relatively unambitious, right? There's this process called right-rounding government, where if you have a policy that impacts on any other department, you have to then write to every secretary of state.

12:06And in your red box at the end of the day as a secretary of state, There's normally four or five right rounds of other departments wanting to do something. And in most cases, if your department doesn't have a particularly strong view and isn't impacted, you tick it. And that's kind of the deal in return for them ticking your idea. But your department might be impacted. And so then those considerations can be taken into account. That's why it happens. The consequence of all of this, of everybody having an individual seat that they're personally elected on, and of the nature of cabinet government, meaning that you have to basically get sign-off from 20-odd people for every decision that you want to take that's significant enough, means that it is essentially non-hierarchical.

12:56Business is essentially hierarchical. and therefore some of the dangers in business are in fact the other way around that if you don't have the checks and balances and the decent governance people can just take decisions and say this is what we're doing and there's nobody to stop them. Look at Meta which is an entirely hierarchical business with control in one person they spent 800 billion I recently saw on trying to develop the metaverse and have shut it down as a mistake. Now, I'm sure if Mark Zuckerberg was sitting here, he'd say that wasn't a mistake to go for it because he didn't know that it wasn't going to work out.

13:42But that was a clear example of a decision taken by an individual to just absolutely make something happen. And, you know, the decisions taken by the cabinet and decisions taken in Meta are probably two ends of the spectrum. And all businesses are somewhere on that range. A publicly listed business has much more governance and operates more like a government department than an owner-manager-run business where they're the majority shareholder. But nevertheless, getting that balance right So you don't just get sclerotic indecision through attempts at consensus on the one hand, all the way across to not making the mistakes of lack of challenge on the other.

14:39That is all part of structuring an organisation well to take good decisions.

14:44Matt Hancock:So we run a sales agency. So like within that sales skill set is obviously influence. And when you were talking about it there in government and in business, one of the key skills that you have to have is the ability to influence, whether it's getting people to come on board with your decisions or like to buy something. Can you talk about Rory Sutherland and marketing? Yeah. So when you think about if you had a new idea that you were pushing with it, whether government or in business, how would you approach starting that influence process? So obviously the first thing is to know the audience, right?

15:17Because that's the person or the people whose decisions you're trying to influence. So understanding the audience is the central thing. And then understanding who influences the audience. So aiming for direct influence over a decision taker and then trying to work out who else is going to influence that decision. Those principles apply whether it's a B2B sale of something really valuable where you've got to put a huge amount of effort into any single decision or whether it's a B2C thing where you're looking for numbers, but the techniques are different. The avenues, the channels are different, but the goal is still the same.

16:07So working out what's going to change that person's mind or get them to make the right decision and then working out who else is going to influence them. Both absolutely critical.

16:18Matt Hancock:And then we obviously have a team of employees. And one of the things, especially with not to talk down, but like especially with younger people, one thing that I find myself saying is like, just think about it. Because I think so many people are quick to make decisions without actually just the first step is like thinking about the decision and the process. I'm a big fan of Shane Parrish and the different mental models when it comes to thinking. And one of the things I say is like, okay, if that option existed and that option existed, what is the decision that could come from that so it's not as binary?

16:51Matt Hancock:What mental models, Matt, have you found helpful for you when making decisions? When making decisions? Well, I think that the thinking fast and slow model is really helpful because a decision that you can take and an action that you need to do that takes a very short amount of time, but will take the same amount of time now as in the future, there is a strong bias towards making a quick decision and getting on with it. Because, say, a mundane decision that you have to take, so you can't delegate it, and it has to happen. If you put it off for two weeks, it'll just happen, but two weeks later, so why not just get on with it?

17:35So the ability to spot a necessary quick decision and take it is, I think, really important for efficiently running an organization. And this comes back to the information gathering. If you can see a decision, it is low consequence. It won't get taken elsewhere if you do nothing. and you've got as much information as you need to take that decision, then do it and take sometimes one minute, two minutes to make that decision. Then the decisions that require consideration and might need more information, but even if you've got the information that you need, might need thinking through, then allow yourself the time for that.

18:31And, you know, this is a this is time worn, right? There's idioms in English, you know, sleeping on a decision is an absolutely standard way of putting something into that second bracket. And then the other thing, one of the things I've learned about myself is how quick you are to make a decision is is important in knowing and trying to get more decisions right. so I'm a quick decision taker so I know that on balance I will take decisions quickly and I err on that side and everybody's weaknesses are overdone strengths so quick decision taking is a strength but taking decisions too quickly is a weakness and to get these things right you've got to know that about yourself and think it through so at time of recording we're in an interesting probably intersection a lot of people talking about things like claude co-work yeah where you can not only gather information now but claude co-work can say i've got all the information do you want me to actually do the work for you yeah and i imagine in health tech there's probably similar opportunities yeah what do you think is a uniquely human decision that still needs to remain human yeah where do you think the opportunities are to like outsource the tech for a greater result that is a great question and it one that almost everybody in business leadership positions are grappling with right now yeah um so i use ai as a junior member of staff right that's how i'm a junior member of the team that's how i see ai they it claude i use claude um it's absolutely amazing it brings things to me but i would not send out an email to drafted without reading it myself.

20:19And so the decision-making it can do is the obvious and the mundane. And then the judgment is the thing that you then apply on top, which is the difficult decision-making. And so it's a bit earlier I was saying that junior members of a team will take the decisions that are obvious, right? So the more senior you get, the more difficult the judgments are. And that's because if they were straightforward, they wouldn't be brought to you. AI is exactly the same. So asking AI to, say, draft an email once you've given it the context, absolutely fantastic. And that is decisions, right? It's making a decision what to write in the email, basically.

21:11But I would then go and check that. before sending it because I don't yet think it's good enough to be able to represent me. But I think this is rapidly changing, right? We're in March 2026. I think it's now normal to receive a draft of something from your lawyer that has obviously had AI in it, right? And whereas a year ago, you'd be like, what are you doing? You know, I'm paying you to write something. What are you doing using AI? that norm is shifting very quickly to of course they're going to use ai because then it means the service is cheaper uh and um but so long as they've read it and agree with it um and i imagine in a year's time humans will find it much harder to spot that something's been written by ai at all uh and we'll go we'll we'll start to get to the next step what what do you think within uh within health tech yeah what are the opportunities say in the next two years so in health tech this is a a really important and sensitive subject where is the human judgment um and there's been some really good work by the uh us and uk regulators recently um to say that if something is a uh an information gathering tool then it is of a lower sensitivity to a decision making tool if that makes sense and that's all about where the human is and where the human judgment is in the decision train if you think about health care about 80 percent of spend on health care is spent on people and many of those people are involved in essentially information gathering and assessment and these are the things that ai does increasingly well now there's the other half of people who are irreplaceable in healthcare, which is the human interaction.

23:08But those elements of information assessment, and the best example of this is understanding scans, which can now be understood better by AI than by many pathologists. That is where the technology will be brought to there and then for something that is of consequence for the patient you need to have the human in the loop because you need to check like me checking reading an email before pressing send you need the the judgment of the doctor but the that my view is that judgment will get better and better the more and more the humans can use AI. So at the moment in health tech, the big area of advance and some recently regulated technology is very impressive to record notes.

24:09It makes life easier. It's quite hard to measure the productivity improvement because actually often what it does is it just makes the doctor's life easier. They can spend more time talking to the patient and less time writing up the notes afterwards. They've got to read the notes after the AI has written them to check that that is what the discussion reflected. So that's AI 101, entry-level AI in the health system. The next level is when the AI makes recommendations to the doctor and says, based on this, here are the following, essentially, prognoses or diagnoses. And we are still miles away from that being reliable enough to be acted on.

25:03that needs to go to a doctor as a decision maker. But if you think about it, that decision should then be better because the doctor has the benefit of the AI having read everything that's available on the subject as opposed to having been taught and the lived experience of the doctor. Bring those two things together, the human experience and the judgment, and the AI's ability to assess information at a much greater scale, and you would hope you get a better decision. So here's something that I want to see in the health system. When a piece of AI makes something safer, then that should be rewarded in the way that we judge hospitals, GP practices, et cetera.

26:01the health system. So an organisation like the CQC, which goes and looks at our hospitals and checks that they're doing the right thing, and gives them a rating. If you use a piece of AI, say, that can spot the onset of sepsis, because there's now AI tools that spot sepsis before it becomes a, before a human could spot it by monitoring the vitals and having seen the vitals ahead of a patient actually getting sepsis, right? If it's clinically proven that that technology spots sepsis early and saves lives as a result, then when a hospital puts that piece of technology in, they should be rewarded in the CQC write-up for having mitigated risk.

26:52And that's not about... That's flipping a sort of tech optimism on its head. That's saying this is about reducing risk of harm by using technology that is available. The critical thing is that there's several steps in that that need to be measurable and measured, but once they are to a clinical standard, then we should be weighing the decisions that the people who run hospitals make in favour of these assessment tools. I've just used that as one example, right, of the use of technology in a life-saving manner.

27:33Matt Hancock:Zach Thompson, do you ever get sick of calling kebab houses and nanas when you're trying to speak to qualified decision makers? I do. Well, I think I might have found the solution. 98 % correct qualified data, which means when you're cold calling, you're always going to be speaking to the right person. Do you know what the solution is? It can't be Prospio, can it? Prospio, let me spell it for you. P-R-O-S-P-E-O dot I-O. Al, if you're looking for great qualified data, then you know what the solution is. Prospio.io. It feels like when you're talking there about the doctor and the notes, it feels like the biggest thing stopping AI from taking over is context and the human lived experience.

28:19Matt Hancock:If we keep accelerating at the rate that we're going, surely we get to a point where actually that lived experience is also inculcated. Yeah. So you mean, I will read all the novels that are ever written and understand the human condition. Yeah. Yeah. And there's a challenge here, which is that what we want. Is not necessarily rational. So even if the AI can go down that route perfectly. I am an optimist about the human condition. I don't think that it will ever quite get it. And that's because I give you an example.

29:00When we replace drivers in cars with driverless cars, we as a society, as groups of humans, essentially demand that they are much, much safer. So driverless cars are now safer than human drivers. Yes. But we don't just say, therefore, let's just switch over. we demand that the technology is almost 100 safe and there's a lot of people in the tech world who find this incredibly frustrating it's like look you know you don't need drivers in fact it's safer but if a driverless car is marginally safer then society as a whole rejects that right and you get all the stories that you get about the you know one crash of a driverless car gets far more coverage than a crash of a human driven car and um that isn't just the fear of change it is that is actually that is a collective group decision it's a reality that when humans replace a human decision with an ai decision they want a higher standard and demand a higher standard obviously You know, X, Twitter and the Daily Mail amplify that point of view.

30:22Right. But nevertheless, that point of view is real. Whereas if you ask an AI, should we have our driverless cars? is to say, most likely do an assessment and say, well, driverless cars lead to fewer deaths and injuries per mile driven than human driven cars and have done since maybe whatever, September 2024. I don't know the date, but reasonably recently. And therefore now is the moment to change everything over. But that's not what we want as humans. It's just not what we want. We want driverless cars to be almost perfect before we're relaxed about them going on the roads. And that's okay. But is that ego, Matt?