icon_podcast

69

Innovation and Exploration in Enterprise Search at Orrick

Scott Mortenson

Knowledge Architect Scott Mortenson shares his experience in building enterprise search applications for global law firm Orrick. Efficiency and effectiveness are everything for the firm and its clients, so creating a system that is both powerful and discoverable for end users means bringing all the right tools into play.

The team brings a powerful search solution for millions of documents and assets to users through an open source front end, backed by Microsoft’s SharePoint search. Today on the show, Mortenson shares the opportunities their latest search application offers the organization, along with some surprising challenges along the way.

Links & Notes

Scott-Mortenson

Scott Mortenson
Knowledge Architect for Orrick Herrington & Sutcliffe


Transcript

Pete: Welcome to “Shared Insights,” the podcast from BA Insight. My name is Pete Wright and I am thrilled today to be joined by Scott Mortenson. Now Scott serves as a knowledge architect at law firm, Orrick, Herrington & Sutcliffe. And we’re going to be talking about some of the decisions that go into implementing enterprise-wide search. Because of that, it serves us well that Scott’s responsibility includes all things intranet and search. Scott, welcome to “Shared Insights.”

Scott: Hi, Pete. Thank you so much for having me today. Looking forward to it.

Pete: I am looking forward to having you architect some knowledge today.

Scott: Thank you.

Pete: Does that…?

Scott: Thank you. Do you know what? You know, that title is a funny one. I don’t think I’ve even been able to really explain to my mom exactly, you know, what the job responsibilities of a knowledge architect are. But I do like to explain it as, really, you know, sharing knowledge throughout our company, helping people find knowledge. And a lot of this is done over the intranet and through enterprise search. And some of the things we’ll be talking about today.

Pete: Well, as many years as I have spent as a knowledge worker, I sort of feel a sense of natural subservience to you as a knowledge architect. If there’s a hierarchy in place, we are well in it.

Scott: There’s always ways to improve the sharing of knowledge. Is there not? I mean it’s a never-ending. I guess it’s a good job to be in because, you know, people are always adding new information and then they want to be able to access.

Pete: Yeah, absolutely. And so here we go. We’re talking about a law firm with a…and I would say, you know, I don’t want to speak too generally, but law firms tend to be document-heavy cases for search.

Scott: Absolutely. I mean, at least the bulk of information often on the work-product side, sits in documents. So there’s millions, right? You’re producing documents every day. But we also have a number of other things we need to search. And so, you know, ultimately, we wanted a solution that can handle the documents but also handle, you know, SQL databases and integrate with other systems as well where we need to pull that information out along with the documents.

Pete: Now this is not your first rodeo. I’m talking about, you know, implementing search and you have been at Orrick for some time now. I’ll let you talk a little bit about your background, hopefully, as you get into what exactly you’re looking for that sort of undergirds your decision-making process or how you approach enterprise-wide search. What exactly were you looking for, were you looking to get out of this for your users?

Scott: Yeah, luckily, I had done an implementation before so that helped quite a bit. Because the nuts and bolts take some time, and if you’ve already laid that groundwork, it certainly gives you an opportunity to improve the next time around. But if I could start sort of more of the 10,000-foot level, sort of at with a macro problem that we’re trying to solve, it’s, I’d say, probably, like most businesses, a top priority. It’s simply to save people time. And this is an especially critical driver in the delivery of legal services. Legal services are still often judged by the billable hour.

And so here’s a classic example. If I can save of a legal professional five minutes or even…you know, the goal is to save them five hours, obviously. But if I can help them find similar work we’ve done in the past that’s relevant to what they’re doing, or save them from having to do it from scratch or something, some little nuance, and if I can find where we’ve done that before, it helps us serve our clients more efficiently, you know, and it improves the quality of the work as well. And so this is why, to us, you know, search is so important. It can literally give professional services firm like ours a competitive advantage over other firms who may be less efficient in finding, you know, their past work product. So that’s strictly on the work-product side.

But if I get back up a couple of years, when all this sort of came about, in 2017, we were planning both a new enterprise search platform but then also a new intranet platform at the same time. And we’ve decided we were ready to move on from our previous search solution. We knew where we wanted to improve and we wanted to see what else was in the marketplace. You know, we had a unique opportunity, I would call it, because we’re making two platform decisions, which for many firms like ours, we want to work together, you know, in concert as much as possible. But these platforms of internet and search were coming up at the same time and we wanted to deliver these together, you know, with seamless user experience. So our users really couldn’t tell the difference between search and the rest of our internal-facing information, you know, that we try to share, be that intranet or otherwise.

You know, up until then, we had a completely, I would call it, a separate interface and navigation if the user really want to go to search or go look for something, you know, that they thought might be in a department page or that they were sharing as a team working for a client. But then on top of all that, we of course also wanted to find a search engine that could, you know, deliver, ultimately, a high quality and most relevant results possible. So we always compare there. And we also wanted to connect with all of our internal data sources as easily as possible. So it sounds like an easy problem to solve, right, Pete?

Pete: I can totally imagine you. What I’m picturing right now is you standing up in front of your team delivering the Kennedy “Man on The Moon Speech,” right? “We’re gonna launch this intranet and this search platform and do these other things.” And that is the scale of sort of the level of daunting that I’m feeling for you right now regarding this project.

Scott: Yeah. Right. Meanwhile, there’s nothing else going on, right? Nobody wants anybody else in the firm.

Pete: That’s a very sober response. Yeah

Scott: So one thing I think we did do as a good job at, not trying to bite off everything, you know so that we could get something released in a reasonable amount of time. You know, if we would have tried to build everything and then release, that would’ve been taken, like you said, putting a man on the moon. But in this scenario, we definitely, you know, said we’re going to deliver X, just develop that and get it out to our users.

Pete: My impression of the legal business is that there is a state of sort of required level of pace of innovation in order to keep up with the driving a billable-hour metric that you talked about. Where are you today? What is the scope and scale of your current search efforts and what are some lessons learned that you feel like you were able to address as you rolled out the latest generation of a search for the firm?

Scott: Yeah, I mean we have really been evolving. I’ve been with the Orrick quite a bit now, so I’ve been able to see a lot of it, but we’ve grown tremendously over the past 20 years. And that provides its own challenges. So, you know, on the basic side, you know, these are the good types of challenges that you want to have as your business grows, right? You know we expand globally and we’ve seen tremendous growth in the amount of information that we have internally. And then there’s also a bigger geographic spread of our employees, you know, and professionals, which means, you know, walking down the hall which was once the way to find information to just ask people questions. You know, that gets increasingly more difficult over time.

And, you know, everyone knows the problem that Google has caused now for all of us, which is the expectations and everyone wants to be able to find and share information across the enterprise within seconds. But as far as sort of what we learned from our earlier search efforts, we’ve had the concept of an enterprise search I think for 8, 9, 10 years. It was like an enterprise search 1.0, right? So, but it relied, what I thought, too much on the user knowing at least a little bit about where their information was.

Meaning, we had separate search repositories here for things like documents and work product or you separate search box for the intranet or another one for financial data and other smaller repositories. Of course, we need to find things, what we’ve done, you know, globally though. And we just knew there had to be a better way to deliver the search results from distinct areas, you know, to our users without making them bounce from one search box to the next search box, more of a sort of find everything in one place, kind of idea. And that was one of our biggest goals.

And with this implementation, you know, we can still drill down to search-specific repositories. But we always knew we wanted, really, a beautiful way to see results from multiple repositories on one page. And I found this to be, you know, sort of the most interesting challenge over the past year and a half of search, you know, both philosophically and technically. Because if you really start diving into it, even Google splits certain types of results into different tabs on their site, right? So you can search widely different results like web pages and then news and images separately. And so that’s natural for every search engine to have in place. But we also wanted to see more types of information in one place. And so we created that and we call that main page simpler, our everything page, right? So we don’t make up too fancy names. It is what it is. But it also does provide easy jumping points onto the sort of particular types of search results at the same time, you know.

Pete: Well, and that gets to this thing that you sort of opened with, which is this idea around that seamless user interface and addressing expectations of users in and around the Google problem. How did that ultimately play out? Did you measure it? Do you have any data to support that, you know, “Yeah, what we did impacts work product.”?

Scott: Yes and yes. Although I will say, you know, at the beginning, I don’t think we realized how important it was to change the user interface and what it is we wanted. We’re looking for something, but we didn’t know at first exactly what we wanted. What we did know is that sort of a SharePoint front end was not necessarily going to be optimal for us. We wanted a more flexible front end, let’s say, as our intranet was, you know, not going to be primarily based on SharePoint. We do leverage SharePoint for some needs and we’re very familiar with it. So that wasn’t an absolute requirement for us. And we do think of ourselves, you know, very aligned with Microsoft and all their tools.

So having a Microsoft search engine on the back end, running the crawlers and the indexes and the queries, was great. But the fact then that we could use SharePoint as the sort of the back-end search engine and then have an option for an open-source front end, what became extremely appealing to us. And I don’t think that was the very first thing that we look for, but it was the thing, ultimately, that made it, sort of, gave us, you know, the most leeway to improve on what we had before.

And I want to talk about sort of the implementation side of that. I mean, it’s some real effort to integrate an open-source front end directly into our intranet. And that was only, because each had their own sort of CSS-style classes and JavaScript classes, which they could conflict with each other. All right. So sorry for taking everyone fully into the weeds here. But I know some people you know are thinking about the technical implications of doing this, right?

Pete: Yeah.

Scott: But so, absolutely, it was worth the effort to sort of hunt down and fix any of those issues to create a more unified, you know, front-end experience which could really, you know, become an integral part of our users’ workflows. And you know, I continue to strive to make search as easy to access and use as say the internet behemoths of things, right, companies. So we look to them for, you know, we always look for people who put a lot R&D into something, you know. We look to them for hands-on, on how users are using search, and you can even go back in time and see what Google’s did three years ago or five years ago, or Facebook for search. And that’s really interesting because you can see their evolution pattern and then match that up to your evolution pattern in search.

And, of course, we’re going to be a little bit behind them, but, sort of, skate to where the puck is going better when you see how they’re evolving. And so those internet companies, you know, have forced us all to appreciate search more and more and more. And so what we have, numbers-wise, is we certainly have users using our search, let’s just say, twice as much as they have before. So you asked for some real numbers, let me just make it simple, right? The easier you put it into their current workflows and the easier they either perceive or make it actually easier to use, we’re certainly getting a return on investment there.

Pete: You call the main screen, the everything screen, and you’re setting the expectations fairly high.

Scott: Right. It’s true.

Pete: And so, you know, to see double the use factor. That’s a significant improvement. That has to feel good.

Scott: It does. I mean, we’re still new to the game of, you know, we just launched this year. So we’re really, you know, ramping up and seeing where we could continue to improve and add. And there’s always an art to keeping things simple but also giving people more flexibility and more power. So that’s the art. That’s the fun part, I believe, of the project.

Pete: I have let you go so far without mentioning some of the BA Insight tools by name, specifically, that you are actually using right now and that you’ve seen impact this project. What are you using right now and what are you most excited about?

Scott: I mean, we are certainly using a suite of tools. So the front end is the smart hub piece that, you know, ultimately, we’re very happy we chose. And then the back end is the SharePoint search. So we’re dabbling with things like AutoClassifier and those as well.

Pete: Well, that AutoClassifier, not to completely derail you, but that, that AutoClassifier is an interesting thing because you made a point earlier that, you know, when you’re looking at AI and you’re looking at the impact of machine learning on your data sets, like that’s an important thing on the horizon. You know, in the spirit of that question on innovation, where do you see some of these tools, these more sort of intelligent tools, making an impact at Orrick?

Scott: The more information that we start to create, you know, the bigger the challenge becomes. So it’s not all structured. And so that’s where we want to go next. It’s to the, sort of, the more unstructured. And what I’ve found is, is one of the challenges is really, sort of, enterprise search itself has so many nuances, but we’re attempting to build a single solution to meet everybody’s needs in the firm. But the challenge is that everyone in the firm does something slightly different. So each person has different responsibilities, like any company, and different goals. But the search implementation sort of has to service every internal department, you know, each one has its own separate needs. And then, of course, has to meet all the needs of our attorneys and every legal service provider, secretaries, you know, who are serving the external clients.

So you talked about sort of the art of being able to design a single solution. There is one of the biggest challenges. You know, our professionals often have much different needs from our admin and support sides who are, you know, trying to deal with internal clients. They’re serving the external clients. And what you’re trying to do is get everyone to have information, you know, at the tip of their fingers. So that was really one of the biggest challenges to develop, you know, a single interface that really helped with everyone. And we’re trying to support so many layers of information and then so many needs for finding that information at the same time.

Pete: What are some of the big challenges you ran into, things that you maybe didn’t expect in this last round of evolution of your search implementation?

Scott: I would say it was the matching of the front ends really to get it to work in the exact styles you wanted it to. So you know what we got sort of, if you want to call it out of the box, was great. And then we wanted to really wrap it in to sort of a core of our main website. And so there is the main challenge. The challenge was not getting the users to use this, you know, they were… Everyone is always chomping at the bit to find things easier. Right? That’s always the number one complaint of every internet, “How can I get things quicker and easier?” And so every minute you put it into search, I feel like has a big ROI on the back end of saving someone, you know, multiple minutes. You know, I don’t have that number, but you know, it depends on what you’re doing.

Pete: Well and helping them to invest in trusting the system. Right? That you come out of this being able to say, “Our users trust the system to deliver the information.” That’s a huge win.

Scott: Yes. And there’s nothing like really getting to start over from scratch and knowing where there wasn’t trust before and making sure you concentrate on those. Because when you come on, you have to say, you know, this is going to be better. It darn well should be better.

Pete: Okay. Last question. To those who are listening to this, who are in the same boat that you are in, who are looking to do a similar project, who have this sitting in front of them, what is the number one lesson that you learned on this project that you can recommend they take such that your life can serve as a warning to others?

Scott: I am happy to help. And one thing that’s nice is a lot of people helped me through this project and I always love, you know, the industries where people are willing to share, you know, lessons learned and things like that. But being that I’m pretty fresh into it, I’m happy to share sort of my wisdom on this. I would say two things actually. I’d have to split the number one into 1A and 1B.

So, 1A, I would say, be sure to put sort of a strong team on the project with a variety of skill sets. Because there’s always back-end, very data-heavy challenges, like to index all of the data, categorize it, figure out what skills you want to show people and filter by, and all of those things. But then you also have so many front-end considerations, right? So this is usually a different developer skill set here. So everything has to be customized…not everything, but you know, at least the information that displays on search results pages. You know, you always want to customize for your organization’s needs.

And so from the development perspective, I would say, you know, having a wide skill set, you know, of people working on the project versus one person who can do a lot of different things in the implementation, that’s sort of 1A, right? Because then, someone is not learning something brand new or, you know, we had some pretty talented people, but I was lucky also to have people with that skill set. And I think that’s one thing that made it possible.

But also, sort of, this is also a more general piece of advice that I always just sort of like to go back to myself and that’s, I would say, sort of always look at the user first. And so, you know, people always say they’re going to do that, but then sometimes it gets mixed up with some sort of initiative or something. So, users, you have to realize that they’re just gonna ignore anything they don’t need, right, and focus only on what they want, especially in search. So you want to be as empathetic to the user experience as you can when you’re creating this. And sort of, you know, besides having this being sort of separated from any other initiative, you know, you don’t want any competing goals. You really want this to be, “Okay, I need to get this information to this group of people or to everyone as fast as possible.”

So you just want to be as user folks as possible and whether that takes, you know, interviews or an agile approach where people are looking at it as you’re creating it, whatever you can do to do that, you know, speed and efficiency are the keys here. Definitely use sort of the 80/20 rule because you don’t want to overbuild this, right? You want it to handle most of the searching needs. But then the real art again is to sort of figure out where you can let users drill down deeper for more information, if they need it, but they might be sort of more of a special, you know, case scenario.

Pete: Can you talk just briefly, as we wrap up, about discoverability from a user perspective? I find that, you know, when you’re talking about these systems, the entire universe of your new application will consist of, you know, the five steps that any given user has to use every single day and they won’t necessarily expect that the application does anything beyond what they do with it every single day. How do you approach this idea of discoverability in your search efforts such that you can hopefully expand that world more seamlessly for your users?

Scott: One thing I’ve noticed that’s really cool and I’m sort of just probably putting this into words for the first time, but, you know, the more you make discoverable, the more the information that is being discovered is…then gets tracked. So let’s say it’s a knowledge base, right? Well, you could have these knowledge bases, but nobody ever use them. Or nobody knew that the piece of work product that they were creating would get reused. So as you make things discoverable, you sort of put into a place a circle now, which promotes people to keep the data better. I don’t know if keep the data better makes perfect sense. But you know, we’re always trying to get people to track information so that we have, you know, we can, in the future, go back and refer to that. So until you have the nice discoverability, people don’t see the actual benefit of keeping their data, you know, in working order, keep it up, you know, make sure stale stuff goes away. And that’s what I’ve noticed is it becomes a circle that builds on itself that improves many different areas of the organization.

Pete: Well, that is a terrific lesson learned I think to wrap us up, Scott. This has been a great conversation and I really appreciate your time. Anything else you want to tell us about Orrick and your work there or Orrick’s work, to give us a little plug, before we sign off?

Scott: Pete, I think that we’re all, you know, struggling with the same challenges. And I would just say that, you know, I really appreciate how Orrick had an open mind and sort of looking at different ways to solve problems because I think only then really do you have the breakthroughs that enable you to point back and say, “I’m glad we were brave and I’m glad we took that chance there versus just, you know, keeping something in play.” You know, I just think things are moving so fast and you have to have that as a culture, you know, within your organization to really make people free to feel like they can, you know, do whatever they need to do there to innovate.

Pete: Scott Mortenson, knowledge architect at Orrick, Herrington & Sutcliffe, thank you so much for your time today.

Scott: Great. Thank you very much.

Pete: And thank you everybody for downloading and listening to this show. We certainly appreciate your time and attention. On behalf of Scott and the entire team here at BA Insight, my name is Pete Wright. And we’ll catch you next time right here on “Shared Insights.”