About this talk
This talk discusses the challenges of increasing enrollment in higher education amid changing student demographics. The speakers emphasize the importance of optimizing the digital enrollment journey to reduce friction for prospective students. They highlight trends such as the reliance on AI tools and the expectation of quick responses from institutions. The session explores strategies for managing content structure, user experience, and governance to enhance discoverability and clarity in enrollment processes. The speakers advocate for adopting a product management approach, focusing on continuous improvement and user-centered design to adapt to the evolving needs of students.
Full transcript
All right. Well, welcome everybody. Um, we're going to talk this morning about increasing enrollment amidst uh changing student demographics. And I hope this is a little bit of a downer, but we're not we can't control the enrollment clip. We're not going to solve that problem today. Uh, we can't really control every shift in the market, every behavioral change, any of those kind of conditions. But we're what
we're going to talk through is some of the things that we've seen, CJ and I have seen um and our client base. We'll get into who we are in a second here and some practical examples of how we work with our clients to help improve that student journey, reduce any friction once they land on your site and also room to to get to their goals. So, my
name is CJ. I'm a digital experience solutions architect at Image X with over two decades of experience in the digital world. You know, I work primarily over the years uh in the marketing and communications department. So, financial services companies and also in the higher education uh sector. Uh in particular, while I was in higher ed, I was the product owner at the University of Manitoba up in
Canada. It's about 25,000 students university. Uh and I managed the website as the product owner and the Drupal instance and where I really was just focused on shaping the strategy, shipping out features and getting everybody on the same page uh of the road map that we developed there. >> And I'm Chris. I'm the senior portfolio director at Imagax and I oversee um all of our higher education
clients. Um I went to university. I didn't work at one but I've been in the digital space for almost 20 years. Um and just look forward to speaking to everyone today. So so we're both from Image X. Uh just a quick overview. Our team spans the globe. There's about 19 of us here at Tupacon this year. coming from across Canada, as far away as Brazil, and as
close as about 50 minutes from here here in Chicago. So, and a little bit about image X as a makeup, we have three verticals. Um, we government and NFP is one, commercial is another, and obviously the coolest vertical is the higher education vertical. That's mine. >> Uh, we have about 30 clients on the portfolio working with them in anything from major redesigns to continuous maintenance work. So
uh yeah it's a little bit about us a little bit about the company. >> Yeah. So let's get into it though. So let's get into the one of the core challenges that higher education faces always is enrollment. Right. The reality is that today we're operating in a shrinking enrollment environment. We're engaging with different generation of students. This means we got to address the challenge of enrollment. Not
it's not just about the marketing side of things and just shifting marketing messages. It's really fundamentally about the digital architecture that journey as well. So today's session is more about a conversation about digital systems and how institutions like yourselves can reduce the friction in the student enrollment journey. So in other words, how can we adapt to how students are changing? So let's set the stage understand where
we are today. So this data from the western inst interstate commission for higher education projects the total number of US high schools graduates expected peak last year and then a steady decline until 2041 at a 13% decline from that peak. So if you think about it, if all things remain equal, this environment will become more competitive amongst yourselves in in in the sector and it also means
that the student pipeline is actively shrinking, [snorts] right? So which effectively raises the stakes for everyone here. It's why we have to talk about reducing friction as we compete for a smaller pool of prospective Now according to another research from UPCA which is an education association they talk about 77% of respondents high school that students that is consider university websites the most reliable. So it's still key
to the journey for their journey while AI tools becoming more credible and they say 32% of prospects trust AI tools for program search now. So not totally unexpected obviously, but I believe the trust in AI will just continue to grow as that gets better. So we have to optimize [clears throat] for AIdriven search and there's still a need to connect students from that AI search to your
content. And just one more piece of research I found interesting. Uh 54% of high school students expect response rate within 24 hours or less. like that. That's pretty damn quick in my opinion. And if you think that's quick, consider that another step. 58% of students said that a slow response rate uh will decrease their interest in the institution. So they want instant, you know, communication with you
and that that's a tough situation. And with this recruiting environment just being more competitive and AIdriven and with an expectation to get clarity uh much more quicker than usual, it just makes it more important to get rid of that friction. So because you know digital friction is easier to lose to and it's something we can control, right? A confusing journey through your site to find you know
for for students to find the information they seek. you know, you need to ask ask a question wait and they need to ask a question or you know wait for a response could really you know shape uh whether the momentum in their journey with your site either builds it builds up and get momentum or breaks. So we don't want that to break. Now we know that there
is a shift in the students discovery process. You know, we touched on the AI piece and we know that they're looking at the Google AI generated overviews and as quickly as the student students journeys have changed with the rise of AI, most higher ed websites or many maybe it's most have not kept up. So students, you know, do not discover the pro programs the way they used
to, right? They still tend to uh they start with more questions now, right? And intent and uncertainty and often experience friction and journey. So going way back just to give a show my age a little bit and going way back websites started with text links and browsing you know for me personally it was geio cities back in 1999 um you know menu and text links on the
page were basically the primary navigation for many sites then site search came along and introduced more discoverability including faceted search and then there's also like comparisons you know and it reminds me at one point in the early 90s or not early 90s early 2000 thousands there was this debate about you know menus and and search and what was the primary navigation on a website and you know
in the end both menus and search are needed right you go to a menu if you go to a website you need the menu to get yourself going and then if the menu fails you got search to help you out so search today you know sorry search back then changed the way people would find programs to study program to study AI is doing the same thing and
the reality is that there is a need to provide direct answers to the mail. So quick response time maybe that's you know implementing AI in your site uh to get students um responded to as quick as possible or at least feel like it right so as mentioned many higher higher ed websites not have the ability to keep up sometime it's a budget or resource challenge and often
it's technical debt right so we've heard from multiple clients where you know it becomes an anchor for them this technical debt and they can't move forward You might be frustrated with your web well this is the web team here but I've heard stories where you know the your internal clients your you know recruitment they're struggling to they want to do stuff they start hacking the site you
know you've all experienced stuff like that so you know technical debt becomes more of an anchor it overwhelms slows teams down and really just becomes a roadblock so other friction points we commonly hear about uh the enrollment jour earning usually includes weak search results, you know, inconsistent program pages where and you also got competing like different types of program pages. One on the college site, one on
your main institutional site, uh in your academic calendar in some cases there's confusing pathways to content. We've seen some very interesting navigations. We got tabs within tabs like horizontal tabs within vertical tabs and an apply button somewhere else. like people are trying to innovate and and they're just working with their systems and it's you know they're doing what they're doing their best right there's also unclear next
steps you know that might be an unclear call to action or and also sometimes there's a lot of institutional jargon kind of mixed into the pages which doesn't help students who just need plain language right so there are multiple pieces to enrollment you know institutions think about enrollment only through campaigns and communications. They miss a big part of what really shapes the outcomes. Outcomes come from a
combination of your content. Absolutely. For sure. But it's also the IIA and UX accessibility governance. All those things are part of the picture and there's more to it. But that is why this whole conversation is about architecture and the digital systems being built because there's all these different pieces to the pie essentially. Now, so one way to address en rolling challenge is to treat the enrollment funnel
and digital system behind it as a product and manage it. And that's kind of my background is more product ownership and product management and strategy. So in other words, and this is how I sum [clears throat] up product management and I got this from Lenny's podcast and I I've listened to it a lot and and it's great for product management and it's really if we use that
product management as an approach to evolve your journey, the student journey. We really focus on shaping the strategy, shipping features and syncing people. You do those three things, it could lead to the growth enrollment or growth of your site and and changes to that can adapt to the journey. So I'll get into that more because we [clears throat] want to contribute and and really focus on uh
managing your enrollment journey over the long term because this is not just a onetime project, right? So what do I mean by shaping it? Well, shaping it has multiple steps. You got and it really it comes down to design thinking. If you're familiar with design thinking, this is kind of how to apply it as part of the product management approach. So first you got to gather insights
from your users and stakeholders. Um, you know, that could be from the surveys, that could be from personal stories from recruiters talking to students. You know, you got to ask the recruiters, what stories are they hearing? What, you know, a student might be complaining about our website? Like, it's terrible. I hear it from my own my I have two kids in university right now. So, I kind
of hear it sometimes about like, yeah, this site's not great. And, you know, that was part of the problem back in the day. But, we, you know, we we definitely had to improve it. So you learn what the students are saying to them, uncover those pain points, right? And then you got to define and understand the problems that you're you're trying to solve, which means you got
to listen to their stories and look for this. You got to look for the symptoms, but you really got to identify the root problems you're saying. So it's that five W's. Ask why why why why? And you kind of drill through the symptoms and eventually get that underlying cause of a problem. It's like visiting a doctor, right? ask your symptoms and they figure out what your in
most cases hopefully they'll figure out what your what's your real problem. You got to really start figuring out how to solve that problem and then that brings you to the northstar metric, right? What are you ultimately trying to do with your product, your your enrollment product? Well, it's about something hopefully measurable like increasing applications. Maybe it's also about just conversion on a certain CTA that might be
like a kind of a KPI type of uh metric like just getting them to slate CRM right that's probably a common case here as well like and then slate's a whole other piece and that's understandable so for your website it's up to there then you have all these problems you're trying to solve and then you start building a road mapap based on some ideation right how to
implement what you got to do next or do now, next and later. So that's what the whole shipping part is, right? Like you got to ideulate on features. So uh again it's more design thinking ideulate, test it and then implement it, right? In general this shipping part like being a product manager has multiple roles, multiple hats, right? Shipping is more about being a project manager and a
or a product owner, web manager of some sort. these ideas of of capturing um how to capture all these ideas is like putting it into a product road map or in your product uh backlog and then you want to take them test them build them do concepts validate them with users your stakeholders uh it could be like a new CTA button it could also be content right
I see buttons uh before where it just says how to apply but you know shorten that to just apply right um it could also be changes to your Ein patterns, right? Program pages can always get better, especially if you think back 10 years ago to now, like it's changed significantly with more storytelling aspects to it, right? Testimonials, um, career opportunities, those like guided pathway stuff do out
in California. So, all those things contribute to and those are the ways you've got to ship it, right? We want to change make changes from your product road map because it always ties back to your original northstar metric. So everything becomes this north star. Now the last part is probably the hardest part is just sinking people. Um you know it's a tough it's a tough job. You
got to talk to people. Some are more natural at it. Some aren't. Um but you can't just do this in a silo. I've I've been known to do that sometimes and not collaborating enough just going on my own. But it doesn't really it doesn't work right. It go only goes so far. It really does take a team. Whether it's your whoever you report into, maybe it's stakeholders,
maybe it's a web community of practice which you had to work with or it's the project steering committee. Um you have often products really fail when the teams including stakeholders just aren't on the same page. Right? There's a a real exercise of change and management that has to happen. You're familiar with that at being at universities. And I I always use the adcar model when I apply
change, right? Um you know in awareness you want to educate uh people on your problems that you're finding and then sync them on the strategy. Sync them on this [snorts] the the north stars and your vision for your journey and that product product road map. And then as you build awareness, you still need desire, right? you still need to somehow connect to people and what and and
relate to them somehow with your problems. So, and that that depends on everybody, right? [clears throat] Whether it's developer, you want to be more technical, the marketer, obviously you want to focus more on the messaging, the positioning um and you know designers, right? Or UX problems. So, those are ways you can connect on the desire depending on audience. And then comes knowledge and that's about just sharing
your road map, right? share your strategies, talk to people, build that knowledge, give them that knowledge, and then the ability part is a little bit trickier with this stuff, but it's more about uh essentially giving a tool to people. So, your student recruiters all of a sudden, you know, they they go to high schools, they'll talk to the students and they'll be like, "Okay, this is our
website and hopefully you can figure it out." Well, that's not the best, right? So, and they might complain about it or whatever. They shouldn't, but they might. Um, but now it becomes more of like, hey, now this this tool, what are you interested in? Like, is it math? Is it hands-on work? You can kind of build out ways to figure out what their uncertainties are and their
questions and, you know, maybe direct them to a semantic search of some sort to figure out what they're looking for. Or maybe they just know the program to study that they want to take and focus on. Okay, that's undergraduate. Or maybe it's micro credentials. Like that's a whole other piece, right? And I've seen in one of our uh clients suman they mix their program finder with both
uh undergraduate programs graduate programs and micro credentials basically showing a complete picture of their offering. So that's different and that's a choice you have to make right I see a lot of people do undergraduate graduate separate some international students get confused what undergraduate and graduate is depending where they're from in the world. So, but those are your choices. Those are the things you have to work through.
So, that's the ability aspect. And the last part is just the reinforcement. And it's a little about adoption, right? And that's a long-term process, right? You got to continue to meet up with the people that you worked with. Um, discuss their new stories that come in and then also continue is basically set up that feedback loop that you need and then involve them in in the future
changes. you know get them involved in seeing the prototype for for example or an upcoming feature just gather them to get into the design process. So when you look at the well journey more as a product now right shaping shaping and syncing people you start to ask better and if you have that northstar metrics you start kind of asking better questions and and looking at that product
experience and you kind of work backwards from that northstar like get into slate. Okay. Well, you know, for example, we can ask, why does someone want to apply? So, the past three years, I've been to two graduations for my kids. And you can kind of hear it across when they cross that stage, at least in my graduation seminar that I went to. You know, for example, my
daughter Emily like who's my daughter? They they say like wants to be a plastic surgeon and plans to go to the faculty of sciences at the University of Manitoba, right? That's a good story to hear because from a I want to know what they're thinking like what are their dreams, what their aspirations are and then you can start working backwards. Okay. Well, she wants to be a
doctor. Okay, great. So, next question. You know, how does a high school student discover what the need to follow that occupation, that career opportunity? Well, then you have to find way connect the programs to them and lead them into that profession. Like it's hard sometimes to be like my oldest son didn't have a clue didn't have a clue what he wanted to do. So what do you
tell him? Like there's a smorgish board of programs to take. Follow basically just follow what you like and we'll figure it out. Well, you need some guidance just like many others. And so you just start helping them compare programs. uh you kind of figure out what's kind of the confusion that you had in your in your journey or what are the steps in the journey where you've
built enough confidence knowing what you're going to take for a program. So this whole mindset helps us move from a focus on content only to more of that journey right where coming of all their interests and really address all those uncertainties and then just kind of a remake of the same slide there's a lot of touch points and activity in that journey as you're aware of in
awareness you know trying to build up awareness of your site so you got SEO to worry about and trying to make that good first impression on the students of what it's like to be at your first year Next is that interest step, right? Students need to be uh need to have an easy and delightful experience as you go through the programs. They want to get excited about
your you want them to get excited to be be there, right? The student experience, learning experience and all that. They're looking for clarity. And then it's the consideration aspect. Comparing programs and schools and costs. I know some schools have tuition calculator. Some don't want to disclose that stuff. Whatever your choice is, you have to know that that's a potential feature to add to your road map if
you don't have it today. And you have to have those discussions, right? And then of course the plying then often brings students to Slate, right? Or which is totally different experience and whole other slew of uh of friction there. And then the friction points also continue with students evaluating their offers. Like once you get it, like you still got to enroll them, right? What's the average of
of applications? Like it's just under seven. It's like 6.8, I think, applications per student. That's a lot of offers that they'll probably get, especially the students that you want, right? So, you have to kind of [snorts] continue that journey and and explain the current student experience and bring them along uh until they enroll, figure out how to enroll because a lot of these students as well are
just doing it on themselves. There are helicopter parents like me sometimes for one son and not for my daughter. Um but that's my story. Uh so next we're going to get into more of the Drupal enrollment levers. >> Yeah. Thanks CJ. So we'll shift now a little bit from market context to more the practical part. Not to say it wasn't practical CJ. Stuff was really good. The
theory is good. The journeys were good. But uh we're going to talk a little more about what we can do dayto-day using the tools we have such as Drupal. Uh so we'll talk through things like content structure, discoverability, um UX governance and then where AI can help responsibly. So the first lever of foundation under everything and what is probably the most important is structured content. Um so
this is one of the call this the power move uh for Drupal. This is so if you're really going to focus on one thing um rebuild the content model around core university entities like programs, departments, people, events, articles, things like that. standardizing these content types, required fields, texonomies, and entry relationships, all those Drupal terms we're familiar with to make them consistent. So this this matters because it's
it's not only human readable, but it's machine readable. Um so this first the point is that modeling program information consistently so that people, machines, and everybody can understand it. So if you if you have one piece that says it's online program or one that's distance or one that's something else. Um yes you have content but it's not consistent content and it's not structured. And then there's a
second layer that I won't get too deep into now but it's making that structure easier to search easier for search engines to find and AI tools to understand. So this is how Drupal can publish it to um JSON LD which is kind of like schema. [snorts] And if anybody wants to know more about schema schema gentleman right back in the room raise your hand. Richard he's our
SEO guy he can help that's a separate talk but the point is structured data uh can help with that publishing to to JSON which can help AI and Google read it better and the point here is that it's not just um backend hygiene cleanup it's um it's the foundation for other things moving forward in terms of search like CJ talked about so that that's transitions to the
next piece which I'll talk about um how students actually experience it so this is where the structure content becomes more visible to perspectives or parents, doesn't matter really. So when students arrive, they're they're not they're not arriving ready to select a program immediately. They're they're trying to narrow options, compare possibilities, and and understand what makes one path different from another. So the the shift here is from
browsing to discovery. So So browsing means the student has to figure the site out, not what we want. Discovery means the site helps the student figure out their options. That's what we want. So the screenshot here is uh from northwest north northwestern polytechnics we're say up in Alberta. Um so the value here was not just adding a a search tool or a one click to get to
the programs page. It was creating more guided exploration experience. So instead of ask asking students to decode the site uh we made it more more intuitive with with structure that helps them move from broad interest to to shorter more relevant set of options. This this is my favorite UX term. Reduces cognitive load. Um overuse but important and it's true. It supports comparison and helps student build confidence
and momentum in their journey. So again it's not a it's not a convenient feature. It's it's practical enrollment support pattern made possible by by structuring your your content. And the next one here we'll see it's actually visible. So on the left um the old side you could browse programs on NWP. NWP is acronym for Northwest employee technique. Um but there was no clear path to get to
a broader discovery experience. Um so it meant more manual scanning. Um students were just getting frustrated because a bunch of clicks and they couldn't get back and forth. So they didn't and if they didn't know what they wanted that page was not helping. So in the in the rebuilt experience on the right um you see this lefth hand navigation. It creates a much clearer path entry into
all the programs. So we have this pretty robust filtering system. So you can use that as your navigation. On the right you have the the course descriptions and short descriptions before you get into the catalog. Um and it's according and expands accordingly. So you can stay in that one page. So one click from the from the navigation. Just making sure this um one click from the navigation.
Um and then you're actually on one page when you can consume all the course data there rather than clicking more and getting deeper and deeper and getting potentially lost. Um so that's not only applicable so structured content is only applicable to program discovery program finders. Um the same logic applies to IIA and UX. So here at the the University of Arizona College of Medicine Phoenix, um their
issue wasn't lack of content. There was plenty of content. It was that students had to work really hard to find key information that they wanted. So we did a bunch of discovering interviews with both current and prospective students and found a familiar pattern. Um the content was spread across way too many pages and the path to program and admissions information wasn't clear as it could be. It
wasn't clear at all actually. So we did some discovery activities like card sorts and tree tests, wireframe testing um and then looked at some analytics data to kind of simplify that journey. Um so the result as you can see on the bottom is really this kind of messy navigation with embedded navigation within a navigation kind of sideways horizontal. At the top is the new experience where it's
just one pull down, one click and then you get to that landing page and then that left nav kind of force your journey on that page. So much cleaner to to rather just kind of trying to navigate the the top part of the page. And this can also help once we did this or completed this exercise that also helps um from the measurement standpoint. So once we
did this redesign of the information architecture and UX and the NAV um it introduced clear roles, role-based pathways, streamlined sub menus and and a more intuitive structure for for program admissions content. Um sorry did you like behind? >> Yeah my phone over there. Uh so anyway this kind of IIA work also it also improved our SEO in practical ways. Um and and the data speaks for itself.
Um what it increased traffic 200% year-over-year. So it's a lot of clicks. 498,000 of 457. It's pretty what we wanted in impressions almost three times. Um, so those are good stats that that proved our discovery is worth it in talking to current students and prospective students was worth it and untangling that messy information architecture and navigation was was super helpful in and the numbers show So all
those stuff is great but um one of the biggest differences between turn I mean the one university one experience I know is is what we all strive for for sure but it I'm not sure how realistic that is I I understand it but the point here is consistency. So perspectives are current. They don't think of your site in terms of um central marketing, admissions, faculty departments, content,
whatever. They're coming to your your site because it says University Arizona College of Medicine, not because of any of those pieces. So if you're if a student is navigating throughout the site and it's a different menu on the profiles page versus a different menu or different uh set of blocks or features on the admissions landing page, um it's going to be confusing and trust is going to
be eroded. So this is where Drupal becomes valable. Shared components, reusable layouts, all those matter. Um, so standardization is key. What you see is that screenshot here of the right of what we did for Trinity University in Texas. We helped them build a governance model around how to use the web components and how they should be placed and best practices. It's a pretty exhaustive document, just one
screenshot from it. Um but but the key here is that standards matter and they need to be in uh maintained and enforced which is next year on on ungoverned everybody's favor. Um like a better experience does not say better on its own. If if no one owns a page it drifts and anyone create content without clear criteria for all returns if teams are not reviewing performance that
friction stays and it's and it's hidden. So this is where governance matters and in practice that means assigning content owners, defining when a new page should or should not be created using share templates, um routing content through clear Drupal workflows and approval systems. Uh and reviewing this highv value content on a regular basis to make sure it's still getting good SEO value and make sure it's still
resonating. Um so good good digital experience is isn't an operating discipl it's it's it's not a launch event. Um so that brings us to the next part of the story where AI fits in and where where it does not to be to be to be quite quite frank. Um so far we've talked about some some patterns around it's been clear like structure content structure reduces friction discoverability
improves clarity um and governance helps the experience stay together over time. So AI can strengthen that work um if it's if it's applied properly with with guard rails in place. It also can expose very weak systems and expose bad content. Um just just fair warning, we'll get into that here in a bit. So the first stage is that shift um because as CJ mentioned um AI has
changed expectations amongst users. Um students are are getting used to direct answers very quickly like CJ has said. conversational guidance and systems that help them get oriented without doing a lot of manual work. Um that doesn't mean that we all need to rush out now and do this big AI strategy. I think we just talked earlier that there's there's there's things that lead up to this, but
it does mean um the baseline expectation is is changing. uh students if the students are un uncertain um if they're comparing options or or trying to understand their next steps they they they expect this digital experience to help them move forward quickly. Um this is also why a lot of credible AI work is practical user center the opportunity is not AI for the sake of AI is
using AI carefully where it reduces friction um improves student way finding and helps people find the information faster. So this is where we want to talk about um retrieval. So so getting the information back and semantic searches as CJ mentioned earlier and this is probably one of the most practical ways to start with with AI. So a lot of your sites already contain the right information but
again struggling people are struggling to find it because they search in natural language um and keyword doesn't support that. So so it often leads to um undesirable results because again those terms are not exact match. And this is where a semantic search becomes useful. Instead of looking for matching terms, it interprets intent and and returns the most relevant answer even when the wording is different or misspelled
or whatever. I'm a terrible speller, but I can use perplexity for the best one because it doesn't matter how bad I spell things. Um, and this screenshot is um California State. Um, it's we're building a semantic search implementation for them. It's a screenshot of the actual implementation. It's in a we're in the UAT part of the staging environment. I'm not public yet. Um, but anyway, we're we're
in we're in deep testing with them and and it's going it's going really well. So, it's it's something and it's and we'll get to this on the next slide, but AI that the semantic search is one of those ones I like to call dipping your toe into the water with it because it's not an invasive change to your search and it's it's search bound by the domain
of your site. So, it's not going out to the internet to get things. So just just to drive home the differences here between keyword search and and [clears throat] semantic. So traditional keyword search is like think of the the index at the back of the book. It looks for matching words and then the user has to do the filtering to find it. Semantic search changes that it's
the system does more the interpretive work for you. It understands what you need, not just the exact text you type. So again, this this example of a nursing degree versus, you know, can I study nursing while working part-time with no experience um semantic understands all the nuances and all the all the intent of that phrase and gives you a few answers that that responds and if nursing
degree might come back with whatever nursing and degree hits in that context brings up a pretty big mess of results. So that brings us to um a bigger one and the next step we'll call it the second lever. We're taking AI to that next level. um assuming your your data is good, your content model is good, and your your house is clean, if you will. Um so
this can become more helpful than search alone. So a lot of perspectives are are looking not just for one answer. They're trying to make sense of a lot of options that you offer. They may be comparing programs, wondering whether they are eligible, um asking if certain pathways exist or just trying to understand what to do next. And this is where a guidance assistant um can be useful.
Um it helps them with this early stage orientation. um can respond conversationally, narrows options, explains requirements and points student towards the next relevant relevant steps. Um and and we have a talk coming up and this we'll get to this end that that dives deeper into this demo um we're doing some pretty good um discovery with the University of Wateroo and understanding these user journeys and we're building
a prototype. I'm training the model about how undergrads and grads um act. But uh there's a full demo. We'll get to that by the end of the show, so you can see that tomorrow. And of course, we want to talk about guardrails. Um this one's key. Um AI will not solve messy content. Weak governance or poor architecture. It's just not going to make it's it might actually
make it worse. Um these will actually become more visible. So if the underlying content or data is inconsistent, source material is is not trustworthy or if no one has defined where AI should should stop, meaning hand over to a human or just not give a good answer, not give an answer at all. Um the experience breaks down quickly. So that's [clears throat] that's why it's the starting
point is not a big AI roll out. It's it's readiness, structured content, good source material, um defined use cases, doing proper discovery, uh [snorts] human handoff when that's needed because there's only so much the robot can take you to and then you definitely need to talk to to a human at some point. Um privacy, bias, awareness, all those things need to be need to be considered. So
AI can be useful, but only when it's grounded in strong digital foundations and clear operational boundaries and governance. So if if we want to move on, what's what are some next next points? Um so what can we do in 90 days or 120, 180, 200, whatever. Um forget about the amount of days, but it's more about having a plan. Um so we have an bring up all
three. So yeah, through so in the first 30 days, it's it's more about that discovery or diagnosis. So if you want to do some audits like what's what's not performing how messy is my content model what can I do to improve that um and how can that feed and help our our search and navigation and then in the next 30 days improve the highest value parts of
the experience. So standardize key pages and the patterns the layouts um tighten up the navigation address accessibility um or address any things that make that page um or session hard to follow. And then in the final 30, maybe do a pilot based on some kind of AI implementation. Again, I would highly recommend something like a semantic search or smart search where it it's it's it's very safe
because it's only going to read the content within the domain of your site. Doesn't require a full rebuild. It doesn't require a whole lot of um just, you know, take a year to redo the entire site. You can kind of pick away at certain pieces of it and get it done. And so in closing, um, like I had mentioned at the start, we're we're not here we
we weren't here to solve the the enrollment cliff today. Um, what we we can't control is the demographic reality shaped enrollment um cake market shift, competitor behavioral change, all those kind of things. But what we can control is our content and our data. Um we can we can we can control whether programs are easy to understand, whether discovery uh feels guided or or overwhelming. Uh whether our
systems are structured enough to support uh measurement and and eventually artificial intelligence implementation. So the real opportunity is not necessarily in the work. It's a massive overhaul overnight, but but structured iterations, improve what students experience next, and measure what changes and keep keep building from And that concludes the talk. Like I said, the third the third one on the on the on the agenda here is is
where our colleagues are getting into a deeper dive of semantic search. We're doing a live demo of that. Um and we're also going to get into the program assistance. Um they give you a live demo of that with the University of Wateroo and the other ones with California Um so yeah we'll uh take any questions, comments, feedback, discussions um show. >> You want to see you want
to see the semantic search? >> All right, let's go check that out. >> Yeah. So this is again this is what we're doing for California State Stannis loss. They're um it's in the staging environment. So we're we're in testing now with them. So you can see in the we've got the the standard icon that's hopefully recognizable as it's um AI. So a single click there will get
you to the to the page and and just so we can toggle like not everybody wants to use AI or doesn't want to feel forced to use it. So we have this toggle here where we can turn it off to get to um you just go to your standard keyword search with you can do search sort filtering all that kind of stuff. So yeah, if we just
want to ask the conversation. So how about um tell me about your music program. This is my kind of tellt music program. >> Okay. >> And then it'll do some retrieval. It'll do some streaming here. And then it'll sit back and answer kind of like in a perplexity type of fashion. And it'll also give citations to go links to the to the to the site to get
more information about each citation. So yeah, it gives you human readable um search rather than just links. Um and it gives you it it's just more personal in terms of how the content comes back. So as a followup CJ, ask um ask uh how much is tuition. So and you'll see when we put this in, it's going to understand he's he's already searched for music. So there's
going to be context that is and it's going to frame that context around tuition for the music program. I hope I'm saying this now and it hasn't come up yet. Uh, did it? >> It didn't. U, but anyway, you can see it gave the fees. Um, and you'll notice that California State Sus calls it cost of admissions or cost of attendance, I'm sorry. And we put in
tuition. So, it knows that it's it knows what it's referencing by intent. And then those >> Yeah. So, now we're talking like a human. I meant tuition for music program. Yeah, we'll see. And the point of this testing that we like to do is we want to try to break it because we want to give it all the use cases that um we can help, you know,
train the model in the back, give it better prompts, make it smarter. >> Yeah. So, there we go. They don't have broken. >> No. Well, there we go. We're broken. But the point here is that the conversation content issue. >> Yeah, it is. Um and again, this is another thing we're working with our clients on to clean up their content. Um, it's been in UAT for a
while because it's exposed some content nuances and and kind of discrepancies that we want to fix. So, um, yeah, it's taken a little longer than normal as you can plainly see. Um, but the point is that it's getting back um, good results in terms of how it's processed, how it's laid out, and the citations here that goes to the more information on the within the site. Are
you able to uh modify or change the layout of this response section? >> Yes. So the question is can we modify or change the layout? Absolutely we can. Yes. Um we're using a front end framework that we have all the control that we would want to do. This is the way it was designed. This is the way we work with the client for theirs. Um but yeah
we can change the UI >> and if we decide to use conditional can >> y >> is it like just like what is the back end behind I mean is it using just the content it's just using the content from the site right >> yeah the the question was what about what is the back end um and yes it's just searching the content in the Drupal database
on the site Yeah. So all that anything that's indexed, anything Google can index, that's what it's getting. And it and it won't go. So if if CJ asks what the weather was, um it'll give you it won't know because that's not in the database. >> I hope. Yeah. They can't weather because >> how legal team has identified some sort of questions that cannot allow generative AI to
respond to limit those questions from being asked particular around financial services. >> So the question was how do we have guardrails around legal >> or just if you wanted to say black a set of questions. >> Yeah. Yes. So absolutely we can we can add those guardrails to to not include that data or do do not respond to that data or give an answer like that. So
yeah there's all kinds of things you can do to if there's our talk will talk about prompt engineering the other one and that's how you kind of train it to when that comes up not to respond or give that generic answer I can't help Any other questions about uh product management design thinking? Oh, okay. I'll leave you off with one quick story then. So, just from my
product, I talked to you about this. Not this example, but we're talking about management. And so when I was on campus, I would, you know, walk around campus trying to work somewhere else other than the office, trying to find a nice little place, you know, toward the end of the day. But, you know, I ran into a student, we're in the same hall, ran into a student
and student asks me, do you know where the John Allen building is? I'm like, I have no clue where the John Allen there's 100 buildings on this. Looking it up and realize, oh, it's 20 minutes, 20-minute walk from here. You're 10 minutes late. You're not going to make class. So from my perspective as a product manager, that's a way finding problem, right? Whether it's physical digital signage,
maybe if you're into that space, it's also how do the maps on your website work, right? So those types of stories when you go out there and you listen to people, you got to think through a little bit more and trying to translate it. That's kind of the key to the product management side. I just wanted to share that little story. Um, but I think we'll wrap
here today with so we thank you for your time and and your attention today. Uh, we really really appreciate it. If you have uh any questions to follow up, you can always look us up on LinkedIn or uh IEX website to to reach out to us. Okay, everyone. >> Thank you everyone.
