Analysizing Systems Part 2
The way you overcome something's happening in your environment that change the results you get, you can measure that you can compensate for it the way you do that is called norms. So a norm is a measure that uses historical data to provide context. So in the case of our pizza restaurant, if they looked at what late august early september looked like last year, they could probably c o our sales went up thirty percent over that period last year, too, and we weren't doing any advertising because it's back to school, you can use historical data toe, add context to a measurement you can compete, and this this is every time you look at a profit number or a revenue number remember the company that made a billion dollars? And who knows whether that's, good or bad the way we figured out, whether it was good or bad as we compare that to historical number, how did we do last month? How did we do last year? How did we do last week? That comparison is really important important caution when you're ...
using norms if the way that you're calculating something changes, you can't rely on the historical norm anymore, so, for example, when I was working in online advertising for procter and gamble. We had a method of testing we would we would have a piece of advertising and we would test it to see is this likely to perform in a way that is going to produce a positive result? We're not we would throw it to you know ah panel and they would test it and they would give us a number and the number was either good about we had a large database of historical information about how ads typically performed and we could use that as a norm to compare this piece should we run it or not really really valuable tool until we switched companies and used a different method to go through that process of calculating the result and we no longer had that historical database of information to compare to and they're wass a great tendency of well, the score kind of looks the same right it's a score on a ten point scale, so we have this new method and it got an eight point six and it says five point four in the system is our historical norm yea, apples to oranges comparison you're not comparing the same thing if you're using historical norms, you need to be very, very careful that what you're comparing to is calculated the exact same way if it changes it's no longer a valid comparison cool make sense proxy ah proxy is measuring something by measuring something else which sounds kind of weird and so you think about how this works in the real world so for example many forms of government in the world use a system called representative democracy trying to elect leaders by figuring out the quote unquote will of the people yes we can't take every single voter into a little booth that scans their brain to figure out exactly what their preferences are and aggregate that in a way that we can directly measure what the will of the people means so we don't do that we don't measure directly you know we measure a vote's a vote is a proxy for a preference this person prefers this this candidate to go to fill this position in politics we count up the votes and that count becomes a proxy for the thing that we're trying to measure how it works other examples of proxies things that you khun you can think of as trying to measuring one thing that's hard to measure by by measuring something else any ideas anything subjective how much your boss likes your performance is usually measured by proxy how much a customer like something is usually measured by proxy how much something is working or not if you can't directly quantify that is measured by proxy proxies are everywhere get some suggestions coming and we've got lala who says product ratings as for product quality and customer satisfaction is a proxy uh guma real estate values? Yep uh, m b seven stockholder boats? Yes, exactly. Senorita diary is tv ratings tv perfect? You know how tv ratings air calculated it's actually a sampling strategy right? There are certain people who have attached their tv a little box and a little box measures what channel, how long they have their tv on what channels they flipped to, how long it's on that channel and what they see or what they don't see. And then they aggregate all of that information up into a sample and that sample becomes a proxy for how many people are watching this particular piece of television entity of important time. And the proxy very crucially, is used to negotiate advertising rates. So the more popular shows the ones that get the best ratings, the ones that have the biggest audience. Those are the ones that companies spend millions and millions of dollars to make sure their commercial appears on that show at a particular network at a particular time tell works ok, so that's a proxy segmentation segmentation is the idea that splitting data into well defined subgroups adds context and gives you insight that you wouldn't normally have. So when you're collecting data about something it's important to understand that not every person so particularly about groups of people, right, if you look at let's, see you run a large business and you're looking at your customers you can analyze all of your customers all at once it's value you valuable you can learn a lot from the process of doing that, but you could learn even more if you say ok here all of my customers let's look at everybody under the age of thirty five and everybody at the over the age of thirty five and see how beat the behavior of group a compares to group b no, you didn't really add anything into your data set you're not collecting any more information the informations are you there? You're just artificially separating those measurements into groups and using that separation as a basis of comparison. How else might you segment, david? Uh, customers, but demographic? Yes. Oh, so male versus female age where they live in the world. And so, you know, for example, here at creative live, this course is being viewed by people in many, many, many, many countries you could you could collect the data about who is viewing and what they're doing, and segment that out by the country that people live in and there's probably some differences in behavior that's really cool? How else can you segment byproduct, use product usage? So who is using feature a versus who's using feature bee is are there any other differences and how you do that? What else? Purchasing customers versus non purchasing exactly what two people who buy from us do versus what do people who not buy from us to you or what is what is true about their life or their business that distinguishes them by purchase behavior? Yeah, what else exposed to marketing versus those haven't exactly people who have seen some type of advertisement for you versus not other types of segmentation you could do this a billion ways it's really fun I'd love to hear what some of your personal like within your actual businesses what are what are you, uh specifically for your your practice? How do you segment your customers? And I was just thinking about that because I'm different. I guess uh, one of the ways is which kind of treatment they get. Oh, and another one is, uh, levels off, I guess in a way, levels of pain or or no maybe the way they come in, like what sort of injury they had or yeah, I think that's totally because that once we know that when we put them into a particular plan uh individualized climb customized plan but it is based upon what they came in totally so so let's actually run with that for a minute because you have you have lots of customers potential customers coming into the clinic and they have different types of injuries right. They go all go on different plans. They have different types of behaviors. One of the things that you can do to analyze is how intense is the treatment for that particular thing? How long does it take to provide care for that particular type of treatment versus how much can you bill for? In terms of how much each new client is worth to your particular clinic, and then you can use our old friend the ratio, right? How many hours has spent with the client versus how much does that client bring into in dollars in the worth of the business and you can start to compare serving different types of customers and what that does to the end result of the particular business. Right. So this this is one of the things that a tangible example here is one of my clients and a dear friend. His name is tim girl. He runs ah, web design company called the outfit group. Uh, and when we started working together, tim was building websites for anybody who wanted tio wanted a website, so somebody would say, give me a website, he would build it for them and it could be anybody, so his client base was really, really varied. We were working together and started to do some analysis of how long it took to serve a particular customer in terms of ours, how you know and and some subjective estimates of how much do you like working with certain types of business versus what is the financial reward for serving that particular type of customers who were segmenting by customer type? And what he found in the process of doing the segmentation is there was a certain type of customer that represented what you can call an ideal customer it's the person who needs what you have values, what you've got doesn't take a whole lot of time or energy to sell because they're really excited about working with you. You can provide a fantastic result to them, and they pay you enormous sums of money. It's the ideal situation, they love you, they value you, they shower you with money, everybody's happy tim founded that his ideal customer was authors who are in the process of launching a book and want to use the website is a tool to market the book ideal customers had all the characteristics that he was looking at, so tim decided to make a really hard choice, but, well, I should say a choice that felt very difficult at the time, tim said. My company now serves authors who are in the process of launching a book it doesn't serve anybody else, so if you are not in this situation there are lots of web designers out there I'm happy to refer you but I don't build websites for anybody except this ideal target customer that we've defined through segmentation you know what happened his profitability went up ten times he was working less he was making more he was happier at work his customers were thrilled with him and those customers when it works they have lots of other author friends who want the same thing and they talked him up and he wins business through it so within the past couple months I think tim had five of his clients in the new york times best seller list at the same time because if you're an author in the process of launching a book you want to work with tim because he's the world class expert and how to do this right that's what could really come out of segmenting your customer base in released trying to see who gets the most out of what you do and focus on them because it made me think of something that just happened to my clinic yeah so my first and primary the most primary thing always as the patient should get the care that they need on the other hand it is a business and so speaking businesswise um we were kind of all over the place and treating pretty much anyone that came through the door but then a couple of months ago I decided that I was realizing that the patient I really love to tweet was somebody who had come out of a car accident because they just needed the care really fast they had to get back into their life and we were the perfect clinic for that because we had all the practitioners that treat that and so I started like zoning in on that group of people and uh I am exactly that now much happier um actually working a little less but working on people that really need it I really value that care so it's exactly what your justice right hello and you know, the wonderful thing is you get the focus on the types of customers that you do an amazing amazing job with and you don't really need to serve everybody there may be other people who could do any equally amazing job at serving other types of customers that you don't specialize in so it's better for everybody it's better for you it's better for the customer and it's better for the business that you're referring the customer too if you can send non ideal people for you to somebody else who can take care of the much better it's a win all around at least when I started doing that it was really hard because a prospective patient would come in and I would be like so conflicted about I have c that patient and I really should just see them but then I was like, I have to focus my attention and energy on a particular group of patients, but it was very hard to like say no to some people and do you know why it's heart, why it feels hard loss aversion, right? There's a customer here who has money that they want to pay me and I don't want to lose that, so it feels really uncomfortable to say, you know what? I'm not the right person for you you should go to this person because they could take care of you better. It feels hard, right? So but it's really important that's how you create space enough to make sure that you conserve your ideal customers as best as possible in case people don't know tim actually will be coming to creative life s e I didn't know I seriously I'm like tim real quick if you have ever thought about writing a book or marketing a book, tim stuff is amazing and think group dot com tim, he actually just wrote a book called your first thousand copies where he talks about what you really need to do in order to market a book, but actually sneaky sneaky kind of thing, the systems that he talks about really market anything on the internet, justus well, um tim is really amazing and and knows his stuff and I've been privileged to work with tim tim was actually one of my very first clients when I opened up the consulting arm of personal mba and it's it's amazing to see where he is right now well it's hell yeah absolutely hell his course is actually called how to sell your first one thousand books on he'll be here september ninth or tenth so go ahead and roll for that but just one asking questions here that's okay? Absolutely eso actually can we ask what on segmentation that maybe go back a little bit as well? Because some measurement stuff yeah so we've got simcox who says just as you can measure insignificant things in your business can you segment inappropriately or ineffectively or ever in such a way as to mislead you? Um yes you can and it's miss lee, if your data is good misleading is the biggest risk of misleading is paying attention to a measurement that is actually not going to help you change behavior and by spending your time and energy focusing on that measurement and ignoring something that actually is important and will change your behavior that's the risk so if your data is good, the measurements that you creates through segmentation are probably going to be valid that's not the risk it's the risk of spending your time and energy looking at something that's not gonna help you the when you analysis off there segmentation goes wrong in the film business day the story is they do all the announced a demographic so they do our ages or they do it's ethnicity and that creates a problem because sometimes they think they begin thinking they analyzed the data being thinking they teach twenty five years only the only liked to watch the certain movies or this group of people hearing like that in this a huge issue because our steelmakers thinking they're crazy right? And uh so what do you think about that? How else make can you segment ok so we have a population here the total population of moving people who watch movies in theaters and a particular segmentation approach which is demographics it's only one way to segment how else could you potentially segment that population to get different information? Yeah see, I don't have a problem with how they do I have a poem with the analysis of it because once they get to that data they they determined that that group only likes that kind of movies and what you begin seeing its movies then you don't want to kill us anymore because every movie it's it's a stupid movie designed that they think that this was our people want to see that you're short so a certain type of segmentation is providing a certain types of measurements that's providing that's creating certain types of decisions that keep being made over and over and over again so one of the things you can do is take the same population, segment it differently and do it in a way that may give you additional information to make different decisions and here's one thing you might do what if you look at the total population of movie goers and you look for every person in that population how many movies on average they go to see in the theater and you segment that population by that right? So the people who are in theaters three times a week because they love movies what do those people like versus the people who on lee come to a movie wants or twice a year? But are there any differences in the behavior of those sets and so we can define so the people who come very, very often those air your frequent purchasers? What do your frequent purchasers what do they like? How do you measure their responses? And if they really like a movie are they more likely to go recommend it to their friends than somebody who only goes since he's something once or twice a year? Yeah, probably, but the different segmentation allows you to get more information, get a different perspective on the behavior of that particular group then you would just relying on classic demographics it's like people who likes what for to have dinner with people like food. The first group goes to see any movies come on, there's a lot of movies, right? The second group is the mystery because they watch and there I was soon the people were like restaurants there the same way they go to they hear about a new russia, and they go to see a theatre now, so that's that's just the moral of the story is you can collect a lot of data and the way that you slice and dice that data will really affect the types of information and the types of decisions that you get out of it. We actually we had a question that was more related to numbers and analytics, so we can go ahead and throw it out now or band apart. Asked if you would comment on big data and predictive analytics as an illustration of correlation industries, they're more interested in correlation for behavior rather than cash casualities that was a question yes, and so big data is really just doing all of the things that we've been talking about on ly just collecting a lot of information about pretty much everything that you can get your hands on and keeping it in a huge database that you're continuously adding to so you could do exactly what we've been talking about. You can slice and dice it in all sorts of different ways and over long periods of time to see what's working and what's not so in business so so this is what makes using measurement and analysis in business different from things like science, which is in science you're usually trying to to prove something causal, right? If I do this than I do this and it happens the same way every time that's the scientific result that you're looking for, ideally, if we could get to that in business, that would be awesome that's kind of a gold standard, but it doesn't have to get anywhere near that solid for it to be useful and so correlations in business people who exhibit this particular characteristic tend to do things like this more of a correlation type of thing that's valuable enough to do something with like because you don't need a causal chain, all you need is enough information to do an experiment and it's the experiment that's going to give you the weather or not doing this improves whatever it is that you care about more customers, higher probability, whatever and so for business. We are really interested in behavior. What are the things that cause people or influence people to do something new and get just enough evidence that we can actually do an active experiment in the business to figure out whether or not it works now last concept in the structure, which is humanization and humanization is using your data to south stories about real people with real behavior, so one of the things that makes business unique is that as we're collecting information, were collecting information about people it's not just numbers on a spreadsheet it's not just numbers in a database it's not just ratios, we're measuring the behaviors of real people doing real things. And so this is one of kind of the quirks of the human brain when it comes to looking at numbers it's very easy to think about these things in the abstract without connecting that all the way back to the real behavior of a real human being in a real situation. So for example, let's imagine that you are responsible for measuring that calls for managing a call center. You have a bunch of customer support people who are supporting customers who are calling in to get help with their problems. And when you come in there's a big problem because average on hold time for a call is sixteen minutes terrible, terrible, terrible customers are ticked off. Let's say through your awesome management, you are able to reduce on hold time to eight minutes. Break out the champagne baby because you just produced that k p I by fifty percent awesome really great until you connect that metric of back to the reality of a real person waiting on a phone for help for eight solid minutes and still getting really ticked off right so the number says we reduced this critical thing by fifty percent but the reality is you still have a problem you need to reduce it even further and so that's really important when you're doing any of this stuff is when whether you're looking at a k p I or some other form of measurement is what does that mean and so imagine just a person that you know doing this thing is that a good thing or not? Is that useful or not? Is that an improvement or not and the more you can you can learn how to translate a number on a page toe unimagined person taking that action related to your business the more effectively you can use those numbers to actually influence your decision making and that's the purpose we're not collecting these numbers for the sake of collecting numbers were not generating reports because they look pretty the whole point of this exercise is collecting useful information and using that to make a decision that changes their behavior or changes our strategy if you don't do that situation you don't make sense ok? So general questions on analyzing systems measurement open anything before we go to a break um and this is kind of going back all general um back to earlier today when we're talking about making people feel important and you're helping paradise asks how do we differentiate between making people feel important and appeasing them? Yes all right, so making people feel important important is a general approach to the tone of a particular conversation right you're making them feel valued valued and respected and sometimes the only reason that you are uh you are talking to are communicating with the person is they have responsibilities they have jobs that are either being done or are not being done their part of the system for a reason and so if the part of the system they're responsible for is not being done you need to be able to communicate with them in a way that it's like ok, we have a problem there's something that needs to be fixed but the tone in which you do that can be very very different okay, so it's it's importance and appreciation is so much about tone and approach but when you have a problem you have a problem you need to fix the problem absolutely absolutely uh let's see another from the also wonderful sam cox measurements typically are free are they do you prefer a measure that's easy and cheap to obtain but inaccurate or um or expensive measure that's more accurate oh my gosh. Yeah. Um well first off if the measurement is inaccurate it's not going to be useful so you're wasting your time before you even begin um the real thing is figuring out what your key performance indicators actually are and sometimes the key performance indicators maybe really expensive to measure right if you're running a big manufacturing line for example, a key performance indicator is going to be error rate the number of things that come off that line that are broken or aren't well formed or non saleable or like that's bad that's really bad and in order to accurately measure that you may have to spend sometimes millions of dollars hiring employees and building systems and doing things to make sure that error rate is in stays low. But the spending millions of dollars is probably important because if you put defective merchandise into the market that's going to be an even worse problem so the amount of time and energy and money you spend accurately measuring something should be directly related to the importance of that particular measurement in figuring out whether this system was working great a stranger regarding segmentation what if your segmentation is causing you to take measurements that cause you to make decisions that repel other customers from your sails funnel so there's no way to measure them right? Okay, so the fix here is make sure that you first are are identifying your ideal customers and then the question is is this process that you're going through repelling your ideal customers or customers that are not so ideal if they're not ideal customers that's actually not that big of a deal um if you are if you're repelling your ideal customers that's a really big problem that you should probably text. So when we come back from break, we will have the last and most important part of this system's conversation, which is we understand how systems work. We know how to measure them. We know how to analyze them. We know how to get a whole bunch of information than okay. What do we do, do with it? How do we actively improve a system toe? Get a better result, whatever you say you're trying to do.