 That's what I would have said. Yeah, exactly. Okay, where we get started? Let's get started. Good afternoon everybody. How was lunch? You're retired? Taking a nap? No. Right. I'm Ben Johnson and hopefully we'll all come here to talk about cars for about 45 minutes. Yeah? All right. I like it. How many people are car guys? Collect cars, work on cars, play with cars? Tough crowd. Tough crowd for me, Jeff. Well, that's a start. Do you know where the hood releases? That's a myth. We are a company owned by Snap-on Tools, yes. Our company is Mitchell Run and we are a major provider of repair and diagnostics information to the automotive repair industry. And so what we're going to chat about today is kind of our journey from being a traditional publisher up to having a mark logic data store that's serving up a whole bunch of information. So in there, we kind of talk about cars a little bit. Because we want to talk about cars from the context of our users who are the professional mechanics that hopefully you guys are taking your cars to to give them service. And you want to make sure that those guys have the best information so that they're servicing your car properly. So without any introduction myself, I'm the director of product management at Mitchell Run. So I have the dream job for a car guy is that I have owned a shop. I have worked in lots of shops in my younger years. Product manager for a number of companies. I've developed a lot of products all with the eye towards how that technician is going to use that product and how it's going to make his life better. And if his life is better, then my life gets better because he wants to buy more things from me. And it's a lot of fun. So if I was known for anything, it's not just cars, it's anything with an engine. So I designed the diagnostic system for Harley-Davidson Motor Company before I came to Mitchell and a few other things, but kind of a car guy at the root. I remember those guys coming out of high school just was, when you asked me what my favorite subject was, it was auto mechanics class. My whole life has been around training and developing product for that industry. I'm joined with a guy that's not a traditional car guy, so a little bit of a slack. But he buys a little club because he actually owns a nice classic firebird convertible. So we let him in the club just introduce yourself, Jeff. It's tough to be an employee of Snap on or Mitchell without being a car guy. It's not good. So let's talk about Mitchell. Mitchell is one of these old companies that has managed to reinvent themselves. They were started back in 1918 with the simple idea of it just like fixing anything else. It's a lot easier to fix cars if you had information about the cars. In 1918, cars were kind of set out of a factory, burnt it together, and when they broke, literally people were building parts for them because there wasn't a lot of availability and people were just kind of second guessing the, you know, what, eight or ten wires that might have been on the car from the starter to the switch to the white lights. And there was no consistency of information. There was no consistency of understanding how long it should take to do a particular job on a car or what that effort might, how much you might should charge somebody for it. So it was a very real industry at that time. And Glenn Mitchell is the guy that kind of came up with the idea that said, well, if we could start standardizing and presenting information to these garages, it could make their lives a lot easier and it could help their customers because they could get some idea of what it might take to get some of these vehicles repaired. I'm sure Glenn had no idea what his company was going to become in later years. But they started with their first real product, which was an electrical manual. And I can only imagine that had to be like the size of a brochure today because we really didn't have a lot of electrical things back then. But you had generators and distributor plates and a lot of things that we don't have today because they've been replaced with simpler electronic components. The first actual Mitchell-Branded manuals came out in 1946 when we introduced a consolidated parts manual. Because at that time again, and this was kind of focused at the collision industry, car came in wrecked. You had no clue of what it was going to take to fix that car either from the sheet model to the body parts and things like that. So he introduced the first consolidated parts manual and then had a labor time guide that followed shortly thereafter. So you could look at the estimate and this really helped the insurance companies because they were going to fix this. No matter what my labor rate is, it's going to take about two and a half hours. And by knowing that, then me as a customer, if I know that it's two and a half hours, then I can shop and look for the guy with the lowest rate or the guy with the best reputation or what have you. But it kind of helped a lot of the playing field in the automotive industry so that mechanics could charge fair prices and actually really for the shop it was them not getting burned. So they knew kind of a guide as to what they might charge to do a service. So Mitchell kept growing and growing their library of products until about 2002 the printed manual was still available. So it's fairly recent that we actually got rid of the printing presses and moved fully to online publishing. Now when you think about that, I want you to start thinking about the way we had to do that because even with the more simplistic manuals that we had back then and Jeff will show an example or two when we get to him, we had to take the, and we built everything with the professional shop in mind. So let's just take a 1966 Chevy Impala. We said well, you know, a tech by this time was expected to know how an engine ran so we didn't go into the details of how an engine was put together. We gave him the details of how the torque specs for the particular 283 small black Chevy that was in that Impala, what the torque specs were, what the water pump specs were, things like that. But we didn't go into the details of the basic, you know, step-by-step, here's how you would take that engine apart and out of the car because if we did, you take that Impala and then add a Nova and add a Corvair and add a Caprice and then add Fords, you'd want to build a whole bunch of books even in the days of simple data. So when you look at these old Mitchameras, you would see the things that we felt were most important for that professional mechanic to understand that might be different than their basic core of knowledge. You know, we know how the engine works, but how does this engine differ from the standard engine? Well, our carburation works, but how does this Rochester quadrijet differ from the thermo quad? So we had specific information and it's kind of key as we grow ourselves here. In 1989, we actually introduced our first electronic, electronically published product. And so this was a CD-based product, DOS-based, you put it in your computer and we had a variety, I think we started with like seven or eight discs or something, it grew a lot more than that. And it really took the same data that we had been publishing and made it electronically available. And that's really what started the revolution in the automotive industry of moving to online publishing of information because, guess what, most of our garages have pretty limited space and those mechanics like to fill that space with tools. Where the books are an important tool, but as these vehicles became more and more complex, you started needing more and more shelf space to fill with these books, right? So then in 2001, we actually had been acquired by Snap-on Tools and we took our electronic information online with what we call Michelin Demand, which is still alive today. So it started in 2001 as of 2013, you can still get that product. We still have about 50,000 plus users that subscribe to that product even today. And I guess I just talked about that. So over 200,000 technicians utilize our stuff to help fix your cars since nobody fests up that they would work on their own. And to that, just as a history of Michelin, we've expanded our offer. So we actually offer end-to-end solutions of shop management systems. These are the systems where you call and schedule the appointment. They plug that in when you actually sell something in your shop. It's what records the sale. It's what measures the technician productivity. It's what measures your profitability of the shop. It does all these things. So we have the service write-up and scheduling software. We have the actual information to repair the vehicle software. We actually have CRM software where we go back and let that shop stay connected to their customers and help them maintain that vehicle run term. So we do a number of things now that are focused at this automotive industry. Well, I should say at the vehicle industry because we've also expanded into the medium and heavy-duty truck offerings with our information. So we are, this is our, what we do. We don't, it's not a sideline. It's not a product lying within a bigger company. This is what Michelin is all about. Now, vehicles, we believe in our evolution, the vehicles we're changing too. So largely in response to government pressures, better emissions, better fuel economy, things like that, we started adding technology to these vehicles. And until about 1979, 1980, vehicles are a pretty simple thing from a mechanics perspective. But then the computer-controlled ignition came along. Actually, it started about 1975. Moved into the onboard computer. It was adopted pretty much across the board in domestic cars for 1981. And these started, the computer started controlling the carburetion back then and then fuel injection today. And, you know, transmissions, the braking systems, the ignition systems, everything else. Until we got a pretty complex car today. In fact, even in the last 10 or 12 years, we've exponentially increased the amount of technology on this vehicle. As this slide shows, we used to average one to three control modules on a car. We had the engine controller. And then we had the anti-lock brake controller. And then we recently started with the transmission controllers, the HVAC controllers, the vehicle stability control systems, upwards to where the 2008 F-150. This isn't the fancy Escalade, which is up here. But just your F-150, you know what, some of us still think of the road form work truck, had 20 electronic control modules on it, with over 50 sensors, 40 actuators, things that the computer actually controls, and three different networks, vehicle networks on them. And so, pretty complex vehicles, even just your basic work truck. When we're looking at something like this, this is actually a, I think this was a 2011 Escalade block diagram I got from Delphi Corporation. Just talks about all of the technology that's on here. So I was actually in a meeting yesterday, where one of our guys actually has a Cadillac, not the Escalade, but the SRX, I think, David. And he was describing how, you know, this is the fun of it, right? He comes in with his keyless entry system and pushes a button just wrong, and suddenly the car just goes dead. The dash won't respond, nothing works. 20 minutes later, it resets and it starts and runs just fine. Well, guess what? Somewhere he confused some network of systems in there and there you go. And we're seeing more and more problems like that on these newer vehicles, because the vehicle is actually an assembly of parts from different manufacturers. So Delphi might manufacture the air conditioning system, Vistia might do the original electronics, Bosch might do the ABS systems, and, you know, all these things have to kind of interconnect and inter-talk. So it's pretty complex, and in fact, as I show here, the really high-lying cars, like your BMWs and things, have upwards of 70 electronic controllers on them. In fact, BMW is fun because one of the things that's happened with all this technology, and all these other problems that I'm talking about, is this thing called reprogramming, where just like our computers here, where we see the little flash that says you have updates available, where there are updates that are engineered in after the car is put into production, where they realize that, oh, if a guy pushes this, does that, winks once, some of the car won't start, they engineer a fix for that, and they'll allow the shop them to download a program and install that program under the onboard computer systems, where the BMWs in Austrian, you know, smarts, if you want to call it that, decided that all these things have to be reprogrammed at the same time to make sure that everything's kept in sync. So if you had a BMW dealership and your BMW happens to need a reprogramming event, they will tell you to bring it in in the afternoon and leave it overnight, and they will start the reprogram before they go home, and it'll be just about wrapping up when they show back up the next morning. The average is seven and a half to eight hours to reprogram a BMW right now. And that's a more proprietary high speed network. Anyway, so when you look at our product, this is our legacy on the M5 product, so this, you had to imagine coming from a print world where the consolidated information editorialized it, the way I talked about, so we didn't have these big huge books, we only had what was needed. Well, guess what happened when Elvis Electronics started getting on the cars? Couldn't do it anymore. You need it all. So what we do at the core of it is we license the information from the people that build the cars. General Motors, Ford, Subaru, Toyota, Honda, Hyundai, Kia. And we take that information and whatever format they distributed in, we go through a process where we commodize that content so that we can get it into our databases and then we produce it back to our customers that subscribe to it. So this is an example of that Armaman 5 product that's been around for a decade now, and kind of the way it looks, and it hasn't changed much from a 1990 Caprice to that 2012 Escalade hybrid, because as we have things very grouped, we are starting with 1940s, 1950s. We had fuel systems, electrical, engine. There wasn't a lot else and we tried to figure out how to stuff all these modules into these categories and keep it as true as possible. But what happens? When you're in a publishing business, especially when it's a reference tool, and let's just pick on General Motors for that, I already picked on BMW, General Motors says, let's see what's on the on-star. We've all heard about that. You push the little blue button and the magic woman comes on and helps you with whatever your problem is. Well, if that on-star system goes down, what do I look for that in these categories? Is it an accessory? Is it electrical? Is it general information? Is it maintenance? I don't know. And I pulled my product team one day and we came up with three or four different answers to that. Everyone was wrong. And they said, well, then Mr. Smarty, who's right? The technician, because he's going to be thinking about all these different things too. So this technician over here may think it's electrical, this guy over here may think it's an accessory, this guy over here may think it's an engine problem. And so our challenge in life is to make that information findable. Because while all this has been going on, we used to process thousands of pages per year and we were proud of that. We were from thousands of pages to tens of thousands of pages to hundreds of thousands of pages. Last year we processed over 1.8 million pages of only content, normalized that, and got it into our databases. And so, you know, I use the analogy that your technician, you've got the information. Here's the A-Stack. It's in there. Have fun with that. Not the right answer, not the right response. So our challenge was to reevaluate what we had done, how we had evolved from this print world to the online world, and think about how are we going to not only thrive, but survive in this new world. Because there are new players coming along that didn't start with our legacy and baggage. They started with newer technologies, newer ways of thinking about it. And we had to really, if we were going to survive and grow in this business, we had to think about it and, you know, while we talk about it, we think we were in the big data business, at least in the reference of the automotive world, before anybody was even talking about big data. And so we did that. We stepped back, we thought it through, we thought about our life as car guys and how we should see the information and how we could find the information and how technicians would relate to what we wanted to do. And we envisioned the way that we would want to support those customers. And we called into MarkLogic as one of the big data providers. We researched a lot of organizations. I'm not going to say anybody was worse or better than anybody else, but however we came to it, MarkLogic were the people we chose to help to partner with to get us through this journey. And I've got to say that we're really, really good about helping us understand big data in some ways that we hadn't even thought about it. So, you know, when you think about how we had done things, how we had grown, how we had started with SQL and, you know, taking all these data from all these different sources and categorizing it and bucketizing it and putting it into common formats, and you take that world and you say, hey, we just adopted a not only SQL data store that really thrives on unstructured data, get over it. I think we'll turn over to Jeff who will tell us that that's an interesting journey. Did you have a question before we got into that? And I think it is a good time to segue to Jeff because the decision was actually, honestly, made before I showed up at the company. But I think the answer is yes to everything you just said. I'm not sure. Turn to Michael. So, having said that, I showed up right after the decision was made. But I understand why it was made. So, the first thing you're going to notice is, jeez, all these Mitchell guys have colds. That's because we care about the consistency of the audience experience. The moment I knew Ben had a cold, I ran out and got one just so you guys didn't have to adjust your listening techniques to the guy with the cold and the guy without the cold that you know of. So, as Ben said, lots of information out there. So, the volume of content is escalating. These numbers are a couple of months old now, so I apologize for that. We're actually at about 14.2 million pages in the general vicinity of 6.4 billion words. And every time I ask my guys, how many words do we have? Now, they give me this really disparaging look. Trudge off and come back a day later with a number. The types of content are going crazy. You know, it used to be, you'd get some text and you'd get, you know, a couple of drawings. And these days, it's multimedia. It's images with hotspots. It's SGML. It's 23 flavors of XML because, you know, God forbid that any two OEMs should deliver things to us in the same fashion. The sources are multiplying. There's new OEMs every year. People are coming in from the Chinese markets. Technicians, we've got now, I want to say, on the order of 250 technicians that every day sit down and kind of document, here's the new or interesting or strange things I saw while I was under the hood today so that we can take that, capture it, turn it into experience-based information and get it out there so that when other technicians see the same thing for the first time, you know, we can give them a clue about what's going on. Analytical data from repair shops. We have now, from our shop management systems, about 290 million repair orders. And you can think of a repair order as, you know, the intersection of a person and a car going to a shop and having something done to the car. It could be an oil change. We don't pay too much attention to those. It could be, you know, they saw these codes on the car with the engine light was on and you ended up replacing these parts. So when you have enough of that information, you can start to do some interesting things with statistics to say, here's what's going on with that car out in the real world today. But it is a lot of information and the volume is growing. So Ben mentioned I'm an honorary car guy. That is this. This is the repair manual for my personal car, 1967 Pontiac Firebird Convertible. I live in San Diego now, so it's great fun to have this. Everything you needed to know. 400. Everything you need to know. It's a Holley Street Avenger, a recent four barrel upgrade. It's three on the tree. Or automatic. I'm sorry, automatic, yeah. Yeah, not the shift on the tree, but just a three speed automatic. So everything you needed to know in 1967 to fix that car. Yes, of course. It's Southern California. How could it not be a convertible? Come on, man. Everything you need to know to fix that car is in here as evidenced by some of these. When I get brave, you know, there's actual like greasy thumb prints in here because I can get under the hood of that with a fair degree of confidence. So that was sort of the state of the art in 1967. Around 1994 it was starting to get interesting. That is a Lincoln Mercury slash Ford Thunderbird repair manual. Technically it covers two cars, but really they're basically the same car with a badge swap. So you can think of that as one car's worth of data. This is around the time that all the new systems were starting to come in. Engine controllers, ABS, very early days of that sort of thing. That is the last set of physical manuals ever delivered to Mitchell. That is a 2009 Camry. That's one car's worth of information. Right around this point, if you wanted to get the printed manuals for every car that might come into your shop, you're not going to have any room in your shop for anything but the manuals. They would fill the shop. It is a lot of information. That is a 2011 Mazda Tribute. We printed that one out for our own purposes. If you printed it out in the same format as the Camry, the stack would be about twice as tall as the Camry. And good luck finding anything in that. If you wanted to look up the heating and air conditioning system, it kind of evenly distributed through all three of those big blue binders. So you're going to spend a lot of time flipping pages and going through tables of contents and indexes, and you're not going to have a very happy result. At this point, the amount of space that physically printed material would occupy is almost irrelevant. It's a warehouse. It's not going to fit in your shop at all. So it's not a needle in a haystack, as Ben said. Now it's kind of a needle in a giant haystack of needles, and you'll have all these people searching for things, and you've got like this one guy over here who's actually happy, because he stumbled over the right needle pretty much by luck. So what's in that data store and sort of how have we come to deliver this online? An article you could think of as a chapter in a book. The OEMs give it to us and it'll be this is sort of the chapter on how the water pump works, or this is the chapter on how to remove something or how to install something. They're not particularly coherent pieces of information, as much as we'd love to think so. So we've got, now the numbers are about 214,000 narratives, and that's anywhere from a couple of pages to a couple of hundred pages. We've got about 107,000 technical service bulletins and recalls, and that's that little thing when you go into the shop, and they go, oh, you know, they pull up the little three-page thing and go, oh yeah, I've got a warning about that because, you know, your airbag things wrong, and so I'm going to replace that part for you free. And then 15,000 and some odd maintenance schedules, which are sort of dull, but you know, it's the time to replace your oil, time to look at the timing chain, time to inspect this, time to bend that. Images, we've got about 6.2 million. Most of those, again, sort of associated with the articles, and that's anything from a picture that says here's the tire rotation pattern, all the way up to an exploded view of a transmission, you know, so there's a pretty big spectrum of complexity in those things, a wiring diagram, schematics, all kinds of things, and they all get managed slightly differently. When it's all broken down, not counting images, it's about 16 million mark logic documents, so it's what we refer to as a document fragment, and what we've tried to do is break those articles down into the smallest possible meaningful piece. So it might be two paragraphs about the location of an O2 sensor and the picture that goes with that. So kind of an atomic unit of information, if you will, that we can then rebuild on the fly. So how do we describe that? There is a very tightly controlled taxonomy that we manage. We have a taxonomy manager, actually we have two. There's preferred terms and those tend to be the ones from the Automotive Society of Engineers. That reminds me, anybody that didn't know all the car guide terms from Ben, I've double checked on Google Translate, they are available there, so if you didn't get what he was saying, you can find it there. So they used to describe components, which is about, you know, now it's up around 13,000 diagnostic trouble codes, which is pushing 68,000 and information types, which is, we've got about 98. And information type, think of those as verbs, right? So it's going to be inspect, adjust, remove, install, replace. So we've got kind of a whole bunch of nouns and then a handful of verbs to describe things you might do to the nouns. And we've got non-preferred terms and those tend to be OEM specific because we just can't have Chrysler and Chevy both calling an identical part and alternator, just won't do. They've all got to have a special word for it. And the technicians don't really always know the special word. So in our taxonomy, we have to manage that so when they say alternator and we go always looking for a Ford's, which is a really generator, okay, we'll go find it. A lot of management around there. Codes don't have preferred terms. They are very, very, very specific. It's a letter and then four digit number and they are extremely explicit. So there's no alternatives to what those are. And then spatial references because, you know, change the shock isn't specific enough. Change the window motor isn't specific enough. So you have to have references like left, right, front, rear, top, bottom to identify what point in the car you're talking about. So what do we do with all that? We take those and we ingest them into our editorial data store. So to the point that Gary was asking earlier that's where we do a lot of our semantic tagging and it's one of the reasons we work with MarkLogic is as we're bringing these things in we can run them through a series of algorithms and say, okay, you know, based on sort of looking at this article we know it's about removing. We know it's about this year make model engine. We figure out some other things so we can sort of categorize that for retrieval later. Yeah, yeah. And one of the things they do very, very well with MarkLogic is they deal with kind of unstructured data. And if you've ever had to sort of mess around with taking data from a bunch of sources and getting into one, most of the time you actually end up investing to take all the data and massage it into a common format and then stick it into your data store with MarkLogic and it could be with others. I mean, I've just got experience with MarkLogic quite frankly. I'm an old school DBA, you know, DB2 Oracle back in the day. I didn't have any code in 10 years. But the thing that took me the longest to get my head around with MarkLogic was they don't care about the structure of the XML as long as it's a valid structure, a valid form of XML. You have to do some basic declarations around it but it's pretty flexible in terms of how you manage those sorts of things. We then use a flexible replication. We send it out to the data centers. There's two, there's a primary and a backup. Both brilliantly located on opposite ends of California on a fault line and we pull off at the same time. And then we use that enrichment, that semantic tagging to reassemble larger documents on the fly. So to go to the air conditioning and heating analysis I was talking about earlier where it's spread through all those manuals. What we do now is somebody types an HVAC and we go out and we gather everything that's semantically tagged as belonging to the HVAC system and we assemble that into a set of documents for the user. We're looking at the use of triple stores to start to bridge some of the gaps and types of content because we've got historically like parts and labor, right? So there was a room full of guys working on parts information. There was a different room full of guys working on labor. There was a whole different room full of guys working on taking the OEM repair information. And so we've got sort of 35 years worth of that and we're now starting to use triple stores to bridge the gaps between them because we're not going to go back and update 30 years worth of content. It's just not a runner. So we use enrichment to group similar items. So descriptions, locations, procedures, diagrams, specifications we categorize those things so that they're easy to find for our users. And we also use it for sequencing because if you think about a technician going on to a job well if I got to do an O2 sensor replacement the first thing I'm going to do is take it off then I'm going to put the new one on then I'm going to test the new one to make sure that it worked and that the systems can all talk to it. So we use that enrichment for sequencing the display of information to the users. And did this meet our expectation? Kind of. You know we had a building full of people very traditional data people, right? I mean it was rigid processes very formal. We used XQuery to replace SQL with something of a learning curve everybody kind of went, yeah MarkLogic is kind of an interesting place to put our content. That'll work. We'll move from Oracle to MarkLogic and it'll be great. And we continued our transition when I first joined SnapOn and I came in sort of through the diagnostics and tools guys and so the SnapOn culture to this day is still we cut metal with fire and then I got transferred to Mitchell and I'm like man now I get to go work at the information side of the place that's going to be fantastic and I get to Mitchell and they go we put ink on paper. Now it's electronic ink and electronic paper but we still put ink on paper. So I'm working through that transition with them but you know yes, yeah, XML data store and the product got out there right but we learned some lessons that we're now sort of having to work back through. The MarkLogic professional services guys gave us loads of advice. They said look you guys really need to think differently about your data you need to think differently about your content you need to think differently about your delivery systems and how you build things and how you get them out there and we went yes, yes, yes of course but we've been working with this data for 95 years and obviously we know our customers and we know our content better and drove on and we built our new flagship product based on our history and our knowledge and we got a better customer experience than a legacy product a lot better experience in fact but at the end of the day we didn't get it right here's why we didn't get it right this is service manual module inside of ProDemand it is the only piece of ProDemand that uses the native MarkLogic search features to hold our own because we knew our stuff better than they did and sort of one guy in about a month put this together using the native MarkLogic features and so it has free tech search so they can type in anything they want we go we find it everywhere else it's type ahead you've got to kind of know the terms for our taxonomy or pick them from a list and here we did some things with information facets you know we said well we can filter it's a relatively meaningful term unless you're one of us but it means you know sort of what information type is it's a DTC is a diagnostic trouble code it's the thing that makes a little engine light turn on on the dashboard and then you can kind of click on what the code was and very quickly you get down to two articles we kind of just a couple of mouse clicks and typing a couple of words and by the way this is the highest performance part of our application and so that led to some interesting conversations it is the most commonly used section of ProDemand by far it's not even close even though the others are much more they're visually a much better experience this actually you know to a utilitarian bunch of guys trying to fix a car this gets the job done faster it's a go to option for the sales and support team member asking them a difficult question this is where they go and this is the part they show off and it leads to one of the most commonly asked questions from our customer base is why are you making me mess around with all those other searches so what we're in the process of doing now is again partnering with the MarkLogic professional services people to redo our newly redone flagship product to actually take advantage of search of those of that now and it's starting to look like it's going to be pretty interesting but with that I'm going to turn it back over to Ben with the accent from the cold to take okay thanks Jeff nice job so we're only now as Jeff said we've got a MarkLogic data center we've got a commercially available product we're very proud we actually had a celebratory lunch yesterday in 14 months so it's the strongest growth we've ever had in a product so and yet with that as Jeff said we really need our customers needs and expectations we do have a much better product than we had this is an example of our own product this is an example of I'm trying to search for how to replace the compressor I think it's a Chevy Cobalt and clicking around through the systems there's an HVAC system then I got to click HVAC system so that you can scroll for miles through and ultimately we'll get down to where there's stuff or you do what I did here and type air-conditioned compressor that seems straightforward and the first thing that's shown to me is common specs no cylinder should read less than 100 BSI oh yeah compression not compressor a lot of problems here's the haystack there's a needle in there somewhere and ultimately you get to the article which even then so now we've got our procedures which are pretty well laid out and then we've got figures which are referenced so if I click this I'm going to picture of whatever it's telling me this drive belt tensioner picture of this picture of that but technicians are very tactile people if they don't see that picture see kind of what it represents they don't want to click on this experience and guess what this is why we're not number one we're number two in this industry right now not by much we're going to catch them again but this is one of the reasons why so take a look at portamand portamand component search same kind of thing same car much cleaner interface typed compressor and it's starting presenting to me what the options were that I could select from and here as Jeff talked about I'm looking at the procedures for this compressor right now here's the removal procedure if I scroll down I'll get the installation procedure as I scroll down here I didn't really get this screenshot enough but you can see that I don't have references to figures I've got thumbnails of enough detail of a figure that I can see what it is so as I'm scrolling down if I want to look at that figure and click on it and it'll expand out which one is known for the best wiring diagrams in the industry and here we've got all kinds of tools about paneling and searching and zooming and all the things that the technician needs to do so they know that there should be a tan and black wire on whatever that pin number is on the serial data line and they can measure that value and back to the specifications so the point of it is is that everything I know about that subject compressor is within one click away from them and I really need a diagram so back out of the article that was in the R&R procedure we're into the electrical and we'll search for it and back and forth very nice much better experience information is now findable it does rely on a very complicated and sophisticated enrichment process which is further complicated because as Jeff said we get this data in so many different formats I drive the Volkswagen I can tell you it's not easy to find information about that car even in the Volkswagen website so as Jeff said we're continuing to work with Mark Logic we are overhauling it yet again we have even a better user experience within just a couple of months from now and that's one of the nice things about this architecture too is we've been able to evaluate, produce obtain feedback and regenerate and enhance in very short time cycles and we'll keep doing that until we get it just where we want it right now we're looking at as Jeff kind of alluded to consolidating all of our search experiences down to one I want to be able to say I'm searching for a trouble code P0301 or I'm searching for a compressor in the same search I should be spawned off as a solutions provider to know what they're looking for and deliver it and we're looking at alternate content types of information we're also going to be serving up experience data about the vehicle and I'm making this quick because I know we're out of time but we'll actually be blending information out of as millions of hours that Jeff talked about into that search experience and blending it with the OE data and I think that pretty much wraps it up this is just ancillary stuff talking about some of the other information that's in there that the technicians find useful that is trying to try and like how to survive in the 21st century and we've done pretty well with it I think that's a root problem I wasn't there for that but my sense is that we made an assumption that people were going to drink the Kool-Aid a lot faster than they did but you're exactly right I mean if you were going to take on a project like this converting from what I'll just call a legacy shop to a no sequel type shop I think it's worth the investment enough knowledge and understanding about what this type of an environment can bring to them what they did frankly we took a great technology and we forced it into trying to act like our own technology did and that wasn't really the best approach but we got it now hey thank you guys