 Okay. Hello. My name is Michael Lascaris. I am from the National Library, as Andy mentioned. I work on Andy's team as part of the Digital New Zealand team, and currently my role is user experience lead. I'm going to be talking a bit about Papers Past today and the recent redesign we did. This won't be a deep dive on Papers Past with this crowd. I'm going to take it as a bit of a given that most people know what Papers Past is. If you don't, go look it up afterwards, and you're welcome. I also need to acknowledge right up front that we originally built this as maybe a pitch for a longer, sort of one hour deep dive on the whole project. We're asked to sort of do a 15 minute, so I'm just focusing on the user experience. It was a very large, very long, very very very challenging project. Some of my project folks are in the audience, and they're going to look at these things and go, man, you're glossing over a whole bunch of stuff and making it sound really easy because I'm compressing two years into a 15 minute presentation. But for those of you who are not familiar with Papers Past, it is the New Zealand's National Historical Digital Newspaper Archive, which is searchable, full text, scanned OCRed newspapers, dating from approximately 1830 through to 1945, which is about as far forward as we're comfortable with pushing the copyright. These numbers are where it was when we launched, so this was the service that we were moving. 130 newspapers, 3.5 million pages more or less, more being added all the time and about 40 million articles. It's also very popular. I think we're one of the most heavily trafficked sites within the DIA portfolio at about 3 million page views a month. That number of 3.5 million pages now that we've moved everything in is about up to 4.7 million. In a big project like this, before you go into a redesign, it's really important to know why you're doing this. It's a question we ask ourselves and it's a question a lot of people ask us like, no, seriously, why are you doing this? Because the Papers Past audience is very, they like the service. It was very popular and people were very, very nervous about change with good reason because it's working. However, it's important to know that the business driver was not the user experience for this project. The business driver was behind the scenes. Two main things I would boil it down to, technical debt and content workflow. The business driver was this. Papers Past was newspapers and we were going to expand it to full text. We wanted to bring in documents of other types, some of which were very newspaper-like, some of which not so newspaper-like. So the challenge was bringing in these four other services that already existed into the Papers Past world. So that's newspapers, but we're also bringing in the manuscript and pictorials site from Alexander Turnbull Library. The Teahoe and the Royal Society were periodicals, journals, and the parliamentary papers in the A to J's. The reason for this is largely technical in that a lot of these other services were different and built in an older era where you had a really big collection, you built a website around it. We had inherited them. They were out of support and possibly not going to make it very far. So what we were doing behind the scenes was the main driver of bringing all of these services historically into a single platform and a single technology stack so that going forward we could iterate, we could build on the same tech, and we wouldn't have all these other things to support and these sites wouldn't fall over in the meantime. The project offered some constraints. One, as I mentioned, very popular. People don't want it to change. We didn't want to mess with what was working. We wanted to understand what was working so that when we brought new things in we didn't unnecessarily impede the current service. The different formats presented a big part of the issue and I'll talk more about that in a minute. So if we're bringing in manuscripts and we're bringing in periodicals into a newspaper world, how do we not overwhelm the user with new information even though everybody tells us they want more stuff on papers past? Smallish budget, it's a lot of work with a certain amount of resource. But also presented us opportunities. We realized that to bring in all these new resources we're going to have to rethink the user experience. So on the one hand not messing with things but also bringing in a whole lot of new stuff. There's a balancing act there and we really had to do some deep thinking on the UX. We needed to future proof so getting rid of that technical debt and having one stack to maintain would serve us well for the future. And this turned out to be a big one as time went on. Device friendliness which three or four years ago was kind of a nice to have, making sure it works well on mobile and tablets and the multi-device world. There's a slide I forgot to put in here which is that papers past in a given month gets visits by Google analytics from about 1200 different screen sizes. So it's the multi-device future has been here for some time. And it turns out Google is actually punishing sites that are not mobile friendly. So if you have an older mobile friendly site you may have seen in the last year your Google organic search numbers going down. So we were starting to see that with papers past and we needed to sort of get onto a responsive version and build back up. So our design approach we did a lot of discussion. We did a lot of surveys. We did a lot of talking to our audience. We were going to prototype and put prototypes in front of people as soon as possible. We're going to do a lot of user testing early on that was largely one-on-one sessions. We would put the prototypes in front of people and ask them to use them while looking over their shoulder and giving them as few prompts as possible. And we did a lot of those sessions probably about 40 hours all told. And we're going to use an agile development process of two-week sprints kind of working on things once we got into production. So iterating and being able to iterate and taking the iterations into the future. So I'm going to look quickly spend a little bit of time of just giving you a taste of what those iterations look like for a particular page. So this is taking the newspapers front page of papers past looking at the site where it started and where we wound up. So this is what papers past looked like when we started. It hadn't changed since about 2007-2008. Fairly simple, fairly straightforward, works pretty well. Not responsive. So it looked exactly like this on every screen on every device. If you looked on this at your phone it would just be this shrunk down to the width of your phone. But we've got basically a very small number of things going on. So we've got the search box there, which is where you can search. Or if you just want to browse, dig in, pick a date, pick a region, pick a title, start looking at newspapers, you click on these buttons here. So our very first prototype was this. We just sort of started hacking in the Foundation CSS framework and HTML, put together some HTML pages and just said, okay, this is essentially taking the front page of the newspapers and translating it into a new mobile friendly chunky site. You'll also notice there's colors in here and the colors are going to jump around. The fonts are going to jump around. It's just an artifact of the fact that I believed in full, they call it high fidelity wire frames. These are essentially, we thought of these as wire frames, but we tried out fonts. We tried out colors. We tried out a few things to see how it goes. But this is essentially the same stuff that was on the papers past home page. And this is where we started. So we started testing with people. And we got this great comment from one particular gentleman who was one of our patrons who was user testing with us. And he talked about, he said this sentence somewhere along the lines, sometimes I start browsing and then I search and sometimes I search and then I start browsing. And we just started thinking like, oh, these are not really two particular routes. It's like, you might want to browse for a while. And then when you browse to a particularly interesting area, you might want to search within those results. Or you might go the other way. You might search for a word and then facet down and filter. But we found it was almost 5050 with the approaches. People sort of, when they're starting from the home page, won a couple of different ways. The vast majority of our audience comes in through Google. They Google something and they go right to the page they want. But from the home page, it was kind of an even split. So we moved to this, which is the idea that over on the side, you'd have some facets where you could pick an area or you had a nice big search box at the top where you could search. And we started this split for, which we had for a long, long time, of if you search for a word, you would get article level results from the full text. So if I search for Wellington, I find articles containing Wellington in my search results. But if I just used the facets, if I just said all of New Zealand or just, you know, Otago in 1940 to 1945, I would get the front pages of newspapers. And we were enamored of this idea for a long time of like, if the search was blank, you get front pages. If the search was not blank, you get articles. Turns out it didn't work so well. So what did we learn here? Well, we learned, first of all, in user testing when we tested this, nobody knew that those things on the right were interactive. So we added little change buttons and bigger change buttons and bigger change buttons. Fit's law, user experience law says the bigger something is as a target, the easier it is to interact with. So our buttons just got bigger iteratively over a period of about two months. But we still had everything else more or less placed in the same place. And you can see the colors are jumping around a little bit and changing. But we had a little bit of a disconnection going on because the next thing that was happening was when you interacted with these boxes over on the left, they would drop down little interactive panels coming out of the search box. So what we found were people were going left and right. I'm interacting with this over here and then it's presenting these little boxes on the side. So we had done all of this prototyping up to this point before our design vendor got involved. It was ClickSuite. So Emily just spoke earlier. She worked on Papers Past with us. And ClickSuite came in, helped us out, did a bit more user research, did a bit more surveys, did some design iterations. And the next couple of slides here are still the prototypes, the screenshots of the prototypes that we were building internally but with some of ClickSuite's ideas in the background. One of the first things they did was really sold us on the idea that the search box needed to be on the side. I love big search boxes. And I really liked the big search box across the top but they said with the faceting put all the interactions in the same place on the left and we tested it and that turned out to be right in the end. So here this is getting a little bit more, a little bit more simplifying up at the top. The other thing that ClickSuite did was move the facets to interact with the sidebar. So when you click they sort of fly out from the side so that it was very obvious of the thing you're interacting with is related to the panel a bit more directly. So we put the title picker in here and then this is actually the very first design draft of the sort of the new design which ultimately rolled into what you'll see as the final. And we got rid of those front pages in the beginning and you want to know what happened as it turns out historical newspapers when you shrink the front pages down to thumbnail size, they all look alike. It wasn't visually very interesting because nobody could learn anything from the homepage. So even though we really like that as a pathway into the site, newspapers from the 1890s were they packed text onto every given service and the front pages were just not interesting. So what we have instead is a featured newspaper and the browse button. So in this iteration here there's a browse newspapers button which went to that sort of browse by date, browse by region and those wound up migrating down into the visual front end. So here's a quick run through the redesign and probably the biggest question we had was the thing I alluded to earlier is like how do we bring all this new stuff into newspapers and not mess with the existing newspaper experience. So before we even get into the redesign the very first question we wanted to solve was one of two approaches. And I've got, I put this here, I don't want this to be seen like I'm in any way messing with Trove. It has more to do with our audience. There's two approaches and one is what Trove does which is probably one of our closest peers where they have newspapers and also a bunch of other stuff. And this is sort of the very rich search result. I search for Wellington, I get books, I get photos, I get newspapers all about Wellington in the same search result and if I want to filter down I can filter from facets on the sidebar. We didn't go with this approach and had we started from scratch if this was a completely new service I think we probably would have gone this approach and we would have made it work fine. Yikes. Andy's flashing me too much. I missed a five. The other one is the Google approach. And the Google approach is really horses for courses. So Google you do web search, if you switch to image search you actually get a completely different interface. And that wound up being where we were going because we found out that when we did the analysis of the metadata as we went from pictures to manuscripts to newspapers to periodicals the only two facets of metadata that were consistent throughout were date and publication name. So they're actually tend to be very different searches. So this is our homepage. This is our old homepage. This is the new homepage. Anybody notice anything? What's missing? There's no search. And we did that intentionally. Lots of Glam people have commented on it. Hardly any patrons. We made it really simple. There's only four buttons. Pick the one you want. You'll start there. So we take you from here to newspapers. There's the browse buttons as I mentioned. We have a nice big feature in the middle. We keep the format and the layout consistent across the other formats. Clicksuite also helped us bring in this nav bar across the top. So instead of doing a little drop down you have a nice always visible tab. So you see where you are and it kind of guides you. Search results. We started using a lot of the extra space here. These were nice and simple. But we brought this in. You can sort of sort by newspaper by date. It helps you scan by newspaper, scan by date, see the patterns and bring in all the faceting right into your line of sight. So even though this feels a bit more condensed there's a lot more information and more search results actually fit on the screen than previous. This is the most popular article on the site last year thanks to Reddit. One of the other things we learned is newspaper articles historically are almost all very tall and thin. So rather than stack up everything on the top we made more use of the right hand side of the page. We moved all this metadata over here. Kept a lot of chunky hierarchical information here and the distance from the top to bottom here even though we moved all this other stuff in is almost exactly the same as the old site. So we didn't push down the article any further. Searching just really quickly. The advanced search worked great when we had eight newspapers but we were up to 130. So this was the title picker in the advanced search and that needed a lot of help. And also you can see in there it says Canterbury 1870 1877 to 1839. So if you wanted only Canterbury newspapers it was shift clicking through a tiny little box all the way down. When you had eight that was fine. When you had 130 it was maddening. So we now give a lot of real estate to the title picker and you can also just check on if I want just Taranaki papers I check on the Taranaki box and I get just Taranaki papers. We also did the mobile redesign. I haven't spent as much time as we wanted on the mobile. There's still a lot of reification we could do there but we did make it nice and chunky and simplified and bring everything together. I'm going to spend 30 seconds on URLs and then I don't want to be that guy who keeps you from lunch because all the URLs if you remember ever interacting with papers past and citing it anywhere we had very 1997 URLs with CGI bin papers past question mark and then all of this stuff. So all the new URLs do you look at the URLs because we're quite proud of them. They are hierarchical and they have information in there. So this is format title date. This is the hierarchy that the URLs follow all the way down until you get to an article and then with an art with an article if we did Newspapers Mount Baker Mail 1883 to page five and the name of article it would have been way too long. So when you're on an article we've already got these nice IDs that contain a shorthand of the publication and the day so we went with them. So that's easy to understand easy to parse. We're seeing some bandwidth savings once we've had some peeping issues with Google. I would have gone into a bit more but suffice it to say we're spending a lot of time in webmaster tools just to tune the new site because a 40 million URL crawl is pretty big. Result feedback initially very positive. We went live immediately after launch it dropped to about 60% as all the people who now had to use the new site came in. People don't like change. Most of that feed most of that negative feedback on day one was sort of of the unhelpful variety like you've changed things on me so and we did get a lot more later that said you know actually I'm getting used to the new one it's okay I spoke too soon. So signs are pointing to more engagement we're seeing more searching we're seeing more pages per visit on mobile and there's more to come and in the interest of time I'm going to stop there. We're available if you have any questions come find me at lunchtime and thank you all. Thank you.