All Episodes

August 15, 2024 • 38 mins

In this episode, we sit down with guest Sam Elliott - Director, Data and Analytics at Calgary Homeless Foundation

You will gain insights into:

  • Innovating with data to improve real people's lives
  • Navigating the challenges of data collection and data management for various programs and services
  • How to measure progress and KPIs for serving people experiencing homelessness.
  • How to increase the maturity of data in an organization that's ready to be data-driven

and more.

Sam Elliott | LinkedIn

Calgary Homeless

Mark as Played
Transcript

Episode Transcript

Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Welcome to Making Data Matter.

(00:02):
I'm your host Sawyer Nyquist.
And I'm your co-host Troy Dueck.
And today we're joined by guest Sam Elliott.
Sam, welcome to the show.
Thanks for having me.
And for folks meeting you for the first time,
haven't heard your name or run across you before,
give us a little bit of background on Sam.
Who are you and what do you do?
Yeah. So I am currently the Director of Data Analytics

(00:23):
at the Cowherty Homeless Foundation.
I've been there for about four years now.
And we are a funder of programs that support those
that are experiencing homelessness.
The role for data for us is we operate a record-keeping software,
so the HMIS software that works for shelters,
housing programs, outreach teams, you name it,

(00:45):
as well as we have a business intelligence team
that's building your classic sort of data browsing techniques
and all that sort of descriptive analytics.
And so leading those teams to help support
our community at large fight against homelessness.
I'd love to start the conversation here
because I've never been to Calgary.

(01:06):
Tell me just a little bit about, like,
what's the nature of homelessness in Calgary?
Like, why does your organization exist?
What's the nature of the need there?
Yeah. So I think we've been around for...
I think we're on our 26th year.
And, you know, it's...
I don't know if we necessarily have
the highest population per capita, but there is a need here.
You know, any major city across the Western world these days,

(01:26):
you're seeing that there is a challenge
with dealing with homelessness right now.
So we came around, yeah, 26 years ago,
mainly focused on housing programs.
So for those that don't know,
we offer sort of recovery-oriented
supportive housing programs.
The idea is that if you experience homelessness,
we get you into a housing unit.

(01:48):
It's either a specific built unit
or a landlord-run unit on the market
and provide that case manager, social worker-led support
to try and increase your capacities for...
and life skills and try and really support you
as you deal with some pretty serious, you know,
typical physical or mental health,

(02:08):
any of those sorts of morbidities,
and get you back on your feet.
Now, to support that and sort of ramp up
beyond just the housing programs,
we also offer prevention diversion programs,
which is if it's your first day of homelessness
or if you're showing up to a shelter,
if we can just simply work with you
to make sure you never experience homelessness,
that ends up being a very cost-effective solution.

(02:30):
So we offer things like that.
And then for the more complex situations,
so for those living in the streets or in encampments,
we look to support programs that will go out in the streets
and provide the services there
with ideas better connecting them and trying to, you know,
at all steps, make the individual's life better

(02:51):
and get them better.
And that's sort of where we play.
So in Calgary, I think, going around, you know,
our last formal number count was around 2,700 people
were experiencing homelessness on any given night.
Now that was in 2022.
We're doing another formal count come October.
We're likely seeing that number increase as well,

(03:12):
especially in the more complex situations,
which is a challenge,
but that's sort of what we're trying to work on here in general.
And so that would be, I'd say, the situation here in Calgary.
You know, you see down in the States,
some on the West Coast, for example,
some pretty extreme situations.
I wouldn't say we're into that,
but it's not a light situation in the least.

(03:35):
One thing you mentioned, Sam,
that I'm particularly curious about,
and I think it's something nonprofits can struggle with
in general, is sometimes they just have difficulty
collecting data.
For whatever reason, it seems like there's these
phantom metrics that people struggle to get their arms around.
And so I'm curious, even just counting homeless individuals,

(03:59):
I bet that can feel elusive at times.
Well, how do you count them and how do you know where they are
and how do you get those numbers and metrics about them?
So I'd love to hear a little more about some of the strategies
that you've employed to be able to just simply collect the data
when it can be difficult to capture at times.
Yeah, so we have two formal ways of addressing

(04:22):
sort of the whole population number.
The first is, and this is a Canada standard,
it's now becoming an annual point in time count.
So that is on one day, we coordinate a bunch of volunteers,
typically social workers that work in the sector,
as well as work with shelter providers
and do a mass street count, all that sort of stuff.
And you can imagine it's a very cost-heavy process,

(04:46):
as well as, you know, as a human-to-human process,
it's not necessarily the greatest feeling to just be counting
an individual just for the sake of counting
without providing them any sort of services.
And so the other approach is this term by name list
or, you know, those experiencing homelessness.
And so with this approach is we actually use our record-keeping software

(05:07):
across all of our programs, across the shelters that are using our software,
as well as shelters will provide us their data directly on, you know,
an automatic basis.
And we'll be trying to use unique identifiers from the person's name
to match them across systems.
There's two challenges to this.
The first one is we operate under client-centric data capture method,

(05:28):
which the idea is that if you tell me your name,
that is who I'm recording you as, you know.
And so that could be a challenge if I'm Sam or if I'm Steve
to do different providers, but also it can be if I'm Sam or Samuel.
So what we're actually working there,
and we've had some pro bono professor over at the University of Calgary here
from the Electro Engineering School, is that deterministic matching.

(05:50):
How likely is it that people that have given slightly different info
are actually the same person?
And so that's been a really interesting pathway.
Even just by simply doing some quick distance matching,
we've been able to reduce the duplicates by 35%.
And so that's one pathway that we're pursuing for that piece.
The other challenge is that when it comes to street outreach,
there's only actually so much capacity with this approach.

(06:11):
Our teams aren't able to go out into the city
and support every individual that's in the encampment on a daily basis.
And so then you're kind of doing some extrapital of pieces
where you say, oh, and that's how 14, 30 or 90 days,
how many different people were supported by these encampment teams?
And so that's where it can become a bit tricky to say on any point in time
how many people are experiencing homelessness,

(06:33):
is you have those two factors that kind of play into the game.
And so it's not a perfect solution,
but it does give us some pretty good and deep insights
into what's going on on a regular basis.
How does even things like seasonality affect those populations?
I think about like winters in Calgary
being quite a bit different than summer experience
and the impact that would have on homelessness

(06:55):
and how homelessness is experienced.
How does that weigh when you're taking population counts
and the experiences there?
Yeah, so we definitely, even on our program basis,
ramp up during the winter.
So we have our coordinated, oh, I've forgotten the acronym,
CCEWR, I'm not sure what it stands for,
but effectively it's our coordinated winter response.

(07:18):
And the idea is that we actually ramp up
and run warming centers during the winter here.
As you can imagine, I think last winter we hit like a minus 39 Celsius day,
which inherently you can opt out on the streets in that temperature.
And so you don't actually see, in our seasonality,
we don't see too much fluctuation in that total population number,

(07:41):
but we do see the location shift,
where shelter usage will be higher in winter,
and especially during those cold snaps,
and we have programs to encourage and get you into the shelter,
versus when you're necessarily,
if you want to be in an encampment type situation,
it's a bit more appropriate in the summer,

(08:02):
where it can get hot here, but it's very lovable.
Yeah, you mentioned your foundation provides services to centers
or homeless services throughout the city.
Is that just in Calgary?
Does that span beyond Calgary?
And how many different organizations are you partnering with across your footprint?

(08:23):
Yeah, so we fund all the folks that are providing the services ourselves.
We don't actually run the services ourselves.
On an annualized funding basis,
these are agencies that we will continuously fund,
because programs need sustainability,
is about 23 different agencies.
So there's quite a few players,
and that's not 23 different programs in those agencies,

(08:45):
that's just 23 different agencies.
And then on a fluctuating basis, we support quite a few others as well,
but those are sort of one-off grants, let's say.
Largely focused in Calgary,
we, you know, there's surrounding areas,
and there is, you know, migration with the population.
But up here in Canada, it's sort of each city is sort of,
has either a community entity like ourselves,

(09:06):
or is run by the municipality,
as sort of leveraging their own approach.
Pros and cons to that, you can localize the support in that setting.
But then from a data perspective, as you can imagine,
you're going to building a lot of, you know, either duplicate infrastructure,
or it'd be great to unify some data sets and all that sort of thing.
And so you do see some challenges that emerge from that data lens.

(09:29):
One interesting thing, and one thing that I take as a point of pride
for what we've built from both our record-keeping software
as well as our data software,
is that both our funded agencies that are mandated to use our systems use it,
but as well as unfunded agencies as well.
And so there's clearly to these frontline service providers

(09:50):
a high value in having this data at the ready for them,
and having the record software.
So in addition to that direct funding that we provide as a service,
I view our sort of data in our platforms as a service as well.
Yeah, so from a perspective of the record-keeping as well as the data platform,
what does it offer?
Like what kind of information does this data collection offer these,

(10:13):
these are for organizations and service providers.
They're getting like names of people and records.
What other types of data points are useful for them
in terms of as they're offering their services?
Yeah, so I look at it at three levels of that sort of that micro,
mezzo and macro level.
So at that micro level, if I'm a caseworker, I can use this as your classic,
here's my case management software.

(10:34):
Here are goals that I have for the individual.
Here are the services that they need.
Here are the referrals I need to make to other agencies or other platforms.
And so that's where for that front level,
if I'm a frontline staff, that's the value there.
If I'm managing my shelter, I can see my shelter occupancy,
who's in which bed, all that sort of stuff that direct,
I'm providing the services, this is what I need to know.

(10:56):
Then we go to that mezzo level at sort of that program manager,
director level, if I'm at the agency,
I can now actually get that broader picture of what's going on
from my frontline staff, you know, your classic business intelligence.
What is the volume of people that we're supporting?
What are the services that we're providing?
Are we seeing any spikes?
All that sort of stuff that the program managers need to plan with all that.

(11:18):
And then at that macro level, we get that broader picture.
And that broader picture really helps our own internal system planning team
when it comes to that funding decisions.
We only have so many resources,
and so we need to allocate them for the highest impact.
And so that's where that data really comes into play.
As well as we can then provide that our government funders
with data on an immediate basis.
And so that's one thing, you know, every nonprofit has the challenges

(11:41):
of reporting to funders and all that jazz.
The nice thing is that we effectively just roll up that micro level
through that mezzo level up to the macro
and just have automated that reporting to our funders,
which is primarily government of Alberta, the government of Canada,
as well as we receive some dollars from the city here in Calgary.

(12:02):
And so we can automate there.
And so that's sort of that value add across the board.
And so, and I think that's the way you have to go with data,
as much as there's always going to be the stick of,
you need to report what you're doing in order for us to validate.
There's also that carrot of making, you know, folks at all three levels,
life's easier.
And that's, and provide that information they need to make valuable decisions.

(12:26):
That's where we play.
That's such a neat tiered approach and where, you know,
you're at that more really fine detailed grain data at that micro level,
but then that can roll up easily to the macro level.
And I think we all sort of aspire to that as we're architecting
and building systems in the data space and easier said than done usually.

(12:48):
So what enabled you to get to that level of maturity in the way that you've
developed your data system there?
Yeah, I'd say the first thing that to be honest would be
we had executive level buy-in with this.
Our previous VP of Community Impact came in and said,

(13:10):
we need to do data right that he had come from, you know, a banking environment
where data was ready at the fingertips when they needed to make a decision.
They had it there.
And so he said, let's get there.
Let's build out order analytics looks like.
Let's build out this modern BI.
So I think that's the, that was the first step there.
And it's a lot easier to over invest in the beginning and build out your governance

(13:33):
and all of your architecture when you have executive level sponsorship
from that standpoint versus, you know, sometimes you'll be cast adrift as a data person trying to
and the executive will say prove that there's value here, which is really tough.
We don't have any infrastructure.
So it'd be that and then honestly, it's just your classic data warehouse
and business intelligence and data governance and just constantly building on yourself,

(13:58):
building on itself.
And so the approach that we typically took was with every data product that we needed to create
for an emerging need then is sort of what of these principles can we develop
and really sort of horizontally scale across all of our different future data products
one step at a time and just keep growing our capabilities and maturity in that fashion.

(14:20):
So we had a great launch launching pad.
And then from there, we had just kept approaching just, you know,
nothing's reinventing the wheel here.
And I don't think that I don't know of many nonprofits that actually do need to reinvent the wheel
when it comes to this.
You know, we're not tech companies trying to change the road through technology.
We're trying to make data in a fashion that actually allows you to make informed decisions

(14:44):
and, you know, make a greater impact.
Yeah. So it was on that point that I wanted to talk about how do you define success?
I think, of course, getting as many people out of that state of homelessness
is going to be at the core of those metrics.
But when you're defining, say, you know, those key performance indicators
and you're sending those up to the core leadership to display,

(15:09):
here's how we're doing.
What are some of those metrics and measurements that you're doing in that homelessness space?
Yeah, it's a good question because we're actually sort of in the process of reinventing
what those look like.
But, you know, right now it's your top level descriptive analytics.
Here's the number of people that are experiencing homelessness.

(15:30):
Here's the number of people that are in our housing programs.
Here's the amount of people that are leaving our housing programs
or leaving our system altogether.
You know, that's sort of that home run.
And all of that sort of jazz.
Program utilization becomes a key thing.
Those sort of efficiency metrics that lead up are sort of a bit more influenced
and a bit more achievable by programs themselves.

(15:52):
So how long is it taking you to work with a person or get them in your program
or any of that sort of stuff?
Now, because, and this is where I like to think of sort of,
we're on the edge cases of edge cases when it comes to trying to measure this with folks.
You know, we have extremely complex folks for a variety of reasons
and you're trying to get them better.
And up until now, we have effectively tried to measure, are you getting a home run?

(16:16):
Well, what about like the singles or doubles or triples or any of that?
And so our evaluation team is actually working with the programs themselves right now
to sort of look and define principles, this principles based evaluation of our individuals getting better.
And this is across four domains.
So right now we've said, is someone getting housed?
Which is a great improvement if you're experiencing homelessness,

(16:38):
but we're not actually measuring is their financial situation getting any better.
So is this going to be a sustainable improvement
or are we going to need to support them very long term?
Is their health getting better?
You know, it's for folks that are complex in our system,
it is not just a financial driven situation.
Now the finances are typically aggravating their situation.
You know, for example, your mental health and physical health will be worse

(17:00):
the longer you are homelessness or experiencing homelessness.
But when we support you, are those indicators getting better?
And the last one is that community connection piece.
Are you actually getting more connected with your community,
either volunteer or events or any of those sort of things are really going to bolster a person's strengths.
And that's sort of that framework of that recovery oriented system.

(17:21):
And I'd love to say that we've cracked the code and have our KPIs rip in right now,
but that's something that we're working in progress
and really working with the programs that we serve,
as well as those frontline staff to really define what those look like.
I think, you know, trying to in our, you know, we don't provide those direct services.
And if we were to try and define those KPIs entirely by ourselves,

(17:42):
it'd be a sort of an ivory tower type situation.
And we really look to sort of bring our community and those that are actually providing the services
to really help refine how we do our approach as well.
It's a non-answer to your question, but it's in the progress.
You gave me lots of flavor text there.
Thank you. That's great.
Yeah, it's just a fascinating way you outline the problem

(18:05):
because for a lot of organizations, it's like the more people in our programs is a good thing.
And to some extent, that's true for you. You want people in the programs.
But the goal is also for people to leave the program and to like be stable and successful and healthy on their own.
The part I'm curious about is even thinking about the evidence of homelessness is one clear evidence.

(18:25):
But the sources of that are so varied in terms of like all the different factors that could go into why.
So it's like we're tracking like one end goal.
We talk about just like the variety of things that could cause someone to be experiencing homelessness.
How do you, how is that assessed?
Or how do you track like the types of variety of reasons or ways someone might be experiencing homelessness?
Or is that something you try to, I feel like you're trying to get your hands around it somehow.

(18:46):
Like thinking about these are the different factors we're trying to influence that help somebody.
But even not collecting that data or thinking through how that is assessed.
Yeah, so in our assessments, when you know an individual is either being intaked into a program,
like I remember apprenticeship diversion programs or being triaged into housing, we ask these sort of questions.
So we act as sort of if you want to enter a housing, a supportive housing program here in Calgary,

(19:12):
we first, we work through, you know, on the ground folks called housing strategists.
They'll first determine if you're appropriate for a supportive housing.
But then they'll ask some pretty detailed assessment questions to try and understand a bit more about you and which programs are effective.
And so we really have these client centric ways of capturing this data.
But that's, you know, it's effectively at that point, by the time we're asking the individuals already actually experiencing homelessness,

(19:37):
you know, it's, you know, typically interact with our system until they're experiencing, which isn't, you know,
and as I alluded to before, if we can prevent it from ever happening, that's better for the individual.
It's better for the system in general, as it's way more cost effective.
And so we're actually working and we've had some academic research partners try and point us into potential indicators.

(19:59):
So obviously you have housing market indicators.
You know, I know in the States it's pretty expensive, but in Canada, housing is really seen some major challenges.
I think in Calgary, we saw year over year rent increases by 20 percent.
And so we look at rent increases and then also the idea of a low income cutoff.

(20:20):
So you have your poverty line, but folks are experiencing homelessness are typically even below the poverty line.
So what is that cutoff point that if we see a large portion of people under that number,
we're like they see an influx into homelessness.
Other indicators, and I know this is an ongoing academic research project, so it's not done yet,
but there's some promising results, is that we've actually combined anonymity,

(20:43):
well, anonymized our data with the Calgary Food Bank data to see if we see a spike in food bank usage.
Can we predict if that individual is likely to fall into homelessness?
And so we're seeing some very interesting results there where they, you know, as most research shows,
the answer is obvious. Yes, if an individual spikes up their food bank usage, they're more likely to experience homelessness.

(21:06):
That will allow us to actually intervene with the program.
And so that's sort of where a lot of additional data efforts are going right now,
is sort of that connections beyond just our sector with the broader sectors.
We're involved in the Community Information Exchange pilot project,
which I ideally hopes to bring together sort of your 2-1-1 with your health, with your R data,

(21:29):
and really sort of connect and also create those cross-system referrals and so on and so forth,
where it sort of can we intervene before they even hit our system,
or if they do hit our system, we have the programs in place to get them quickly out of it.
For all the buzz of AI and LLMs, it's not like you can easily throw your data at one of those

(21:49):
and get these predictive, you know, solutions to just come right out of them and say,
oh yeah, this is exactly what you need to do to solve this problem.
You have to do a lot more building and maybe even training of systems.
I'm curious, is that a space that with all the hype that's out there around what AI can do for you,

(22:10):
what's it looking like in your particular space?
Yeah, I think the main premise of you have to build your data in a great spot before you can actually
drop any sort of AI on top of it, it's kind of a range of really true for ourselves.
But there is some interesting use cases.
So, for example, before when I alluded to that client-centric piece and deterministic matching,

(22:33):
can we find if two individuals are the same?
If we now, you know, we've built a data model that has the person's longitudinal journey through homelessness,
we combine that with this lots of our predictive or lots of our assessment data,
can we predict their next step?
If so, can we see where gaps are?
Can we create an intervene?
We don't want to get into ever a situation where we're trying to tell a caseworker what to do with

(22:56):
an individual through AI just because they probably know way more than the AI system.
And that's an area right for bias.
But then it's sort of the can we predict what our system is going to look like in a year from now or any of that sort of stuff?
And the last thing would actually be, and this is another place where we have to be very cautious about sensitivity or data,
is with all the case notes that these caseworkers are writing or anything like that,

(23:21):
can we plug a potential solution or even just like a word tokenization to be like,
these are the flags of this case note or these are the actual like getting sort of more of those quantitative
indicators out of that qualitative data that's there on the individuals could be another use case.
Now, we haven't done that done any of these or operationalized any of these,

(23:42):
but we've all done some exploratory efforts.
And we actually had a company that's much more talented than myself in the data science world
actually sort of validate and provide a feasibility study and guidance to see could we even use our data for forecasting purposes?
And the answer was yes, but it's tricky.
And the one interesting thing that we never even thought of when it comes to all these forecasts is that most of these forecasts are

(24:06):
built off of the idea of demand generated.
This is what your demand will look like there or any of that sort of stuff.
And we're really a capacity limited system.
We don't have we can't flux our resources to match demand.
And so that's been one interesting approach when it comes to that prick demand analytics is that a lot of the literature
and all that's built out of demand based modeling when we're actually capacity limited.

(24:30):
And so that's something that we've had to incorporate as well with our approach.
One other question I wanted to kind of circle back to is something,
sorry, you mentioned about seasonality and Sam,
you mentioned that you don't necessarily see that total population flux,
but you see their location shift and that that got me thinking about,

(24:50):
you know, as we're talking about defining success,
as we're talking about predictive indicators, how does location and geography play into these factors?
I don't know that we've touched on it much,
but are you mapping these folks or certain neighborhoods and locations?
And do you find that there are certain areas that you watch more closely for one reason or another

(25:16):
because it's location driven rather than other indicators that are spiking.
So any thoughts on that right there?
Yeah, so actually recently we launched a sort of live mapping tool for our encampment teams here in the city.
And so this is, you know, your up to a minute map of where the encampments are located in the city,

(25:38):
when they've been last supported, and any sort of notes that these teams are providing to one another.
A really cool thing that we'd like to highlight here is it's not just one organization that's using this map.
It's a multi-organization tool.
And so right now it's aimed at just simply here is the map of where folks are,
and here's when they were last served, here's those types of services, here are their needs,

(26:01):
here are some potential safety pieces as well to think of.
But then we're trying to take a step further and working with those programs on self-coordination of movement.
So right now, if I'm an encampment team, I'm going to say I'm going to go to XYZ neighborhood today to provide services.
But that's actually, they might not know what the other organizations have been up to.

(26:21):
And so you might have these four different teams going to the same encampments,
providing similar services or the same folks and actually be missing an entire part of the city.
And so we're working with them to really create the platform for them to self-coordinate their services.
Versus, and we don't want to get into a space where we're telling these different organizations,
we provide funding, but we don't boss them around or anything.

(26:46):
We want to provide them the resources for them to determine what makes the most sense for them
from both that caseworker perspective, once again, that micro, mezzo, macro level.
And then we're looking to expand it for all outreach teams in the city.
And so this is a sort of our three phased approach going on right now.
We're done phase one into phase two, and then phase three would be that all outreach teams
and really just providing them a good tool that allows their lives easier on the front end

(27:11):
and then provides us a better understanding of what's occurring on the streets of Calgary is kind of our goal.
OK, so from a technology perspective, how does that work?
Like, how are you live mapping up to a minute, like knowing where people are at and what's going on?
Where's that data coming from? How are you collecting that?
So we found, I found an open source tool called Mage.

(27:32):
Interestingly enough, so, you know, when you think geographic information data, you think Esri.
But after talking with our sales rep, there's some weird data residency requirements in Canada.
And some of the data might have been in transit in the States and there might be some health information,
which is a no go. And so I had to figure out an alternative solution and found this Mage open source app.

(27:55):
And so we run it on our own VMs and it actually has just a phone app right there that the caseworkers can take out,
either a company provided phone or their own phone, have their security.
We also have, you know, API connections into our broader HMIS world.
So that way, individuals names can be mapped and we can once again map it to that longitudinal journey.

(28:18):
And then track sort of consent. We operate under consent based privacy model here.
So then we can make sure that we have a single source of truth for the consent while still allowing those field teams that
optimum access. And then further on, we actually integrate that with our BI tool to then sort of see what's going on
at that macro level when folks were last served and sort of do that time decaying of service provision and all that.

(28:44):
So, yeah, it's been a great tool so far. And I like the fact that it's no licensing fees for us, which is as a nonprofit key.
Yeah. But having the I guess you have the technical skills to be able to stand up and like host your VMs and to run an open source project,
but not every organization can muster the skills to be able to pull that off.
Yeah. Well, it was something new to me. We have an IT team as well that is familiar with standing up VMs and all that jazz.

(29:13):
And one interesting just small thing was everything was written in Node.js for this package.
And our IT team is not going to be out supporting Linux because they don't know how it works.
And so we learned a lot about running Node.js in a Windows environment just to stand it up.
But, you know, you got to work with the resources you got and make it happen in that result.

(29:34):
Yeah, I love that. Just think about the technical capacities your team as developers ambitious enough to tackle.
And earlier in the conversation, you talked about a new leadership person came in and said, hey, we need to take data seriously.
And I want to touch on that a little bit more. Like what did and I don't know how long ago that was or where that fit with your tenure there.
But like, what did things look like before and how what were the what were the baby steps looking like to go from whatever data collection look like at the Caviar Foundation before to some of those early steps to move?

(30:02):
Because a lot of organizations we talk with and work with are on the very early end of that data is not taken seriously. They don't have the infrastructure.
You've built it into a very mature environment at your organization.
So I'm just curious a little bit about that journey, what it looked like before and how do you take to start to take steps there?
Yeah, so it was an interesting place when I started.
So we knew that we wanted to invest in the data analytics world.

(30:23):
We were starting a proof of concept of data warehousing.
But before that, we were doing sort of a record keeping software had its own in-app querying tool that would be used.
We dump it all and then we'd use these massive R scripts and run them on some person's local environment and come up with some data.

(30:45):
And then obviously, you know, you're rocking a lot of Excel manipulations too, in addition to those R scripts.
And, you know, it's always Excel.
You can't get away from it. And it does have some value.
But you need to make sure that your governance is in place that way.
The data that's going to Excel is actually what you think it is, which I think is the number one problem.

(31:05):
And so I think there are replication problems quickly emerged from this situation.
And if, you know, a key individual was off or anything, the whole process would fall apart.
And if you're running all these in-app querying tools and all that, unless you're successfully storing these in some sort of database,
you will almost always lose that historical context and trend analysis and all that.

(31:28):
So effectively, our data before was reporting to government or just providing data maybe to researchers.
And we'd have some operational usage, but the operational usage was so capacity driven that it would simply take weeks for us to discover
who is like what trends have been emerging on our triage list.
We'd spend, I'd lose, I think when I joined the organization, the first two weeks of an analyst,

(31:52):
the first two days of an analyst week were just spent creating our triage list for the week
to see who would be triaged into housing. It was, yeah, it was something.
It was something emerging coming into this world and seeing that.
So that's where we started.
And so personally, from like your experience, like where did you come from and what drew you to want to work in this organization and work with homelessness?

(32:17):
Yeah, like why is this meaningful to you and why did you direct your career this way?
Yeah, so before I was in a venture capital, venture builder organization here in Calgary
and sort of played a lot in that startup world sort of throughout my career,
and we're supporting small businesses and finances and all that sort of stuff.

(32:42):
So I've always liked the idea of making an impact, even if it's a person small business, seeing it grow or any of that sort of stuff.
But really, the main thing that drew me here was I can make an impact.
I could do work with some really cool data without also moving into, and this is a bit of a selfish piece, government or academia.
I never worked in those two, but I don't feel like my personality and the way I like to approach work would necessarily mesh.

(33:07):
And so that was kind of the main thing was that I'd be working with numbers that, you know,
a 1% gain when you're working for, let's say a series B startup is exciting, but it really wasn't necessarily getting me fired up
versus a 1% gain when you work in the nonprofit world, that 1% is an individual's life.

(33:28):
And so if we can figure out with the data how to keep making those 1% gains, that's a lot of lives that can add up.
And that makes it really cool.
And it makes it, the other selfish reason I like to say is I have to spend less energy hyping myself up to do the work.
It's a simple reality.
It's mentally easier for me to be like my work is going to matter today.

(33:51):
And that makes it a lot easier to come in in the morning.
That's wonderful.
I love the stories that I hear from people in the nonprofit world because what you just said,
like the people show up and they're there for a reason, like it matters in a different way than it does when you're going into a for-profit
W-2 job.
There's a different sort of mindset and different sort of impact that comes out of that.

(34:11):
And that affects more than just the type of work you do or the people that are impacted,
but affects like your psychology about showing up at the office and showing up to your work.
I'm curious now over the next six months, what are you most excited about from a data perspective or from your organization's perspective?
New initiatives, something on the horizon that you're looking forward to?
Yeah, I'd say the first one is, you know, the continuing phased approach and rollout of our mapping system.

(34:39):
I think that's going to be a real huge value add to our teams and level of understanding and really in meshing that level of support.
We're revamping some of our assessment tools.
It's going to be really exciting as well as that principles based evaluation that our evaluation team is running where we can now start to actually,
you know, next time Troy asked me the question of which KPIs are we tracking,

(35:03):
we'll be able to say here are the indicators that we're looking across the system.
Here's what we're seeing across the programs in a meaningful driven fashion.
So excited about that as well.
And then we just have, you know, ongoing quality continuous improvement with some of our pieces that gets me excited.
But, you know, that's your data governance type stuff and things like that that aren't the sexy pieces, but we'll make that long term impact.

(35:27):
You mentioned a couple of tech tools already.
I'm curious what other pieces overall make up your technology stack that you're working with?
So you mentioned BI and databases.
Like what does that look like for you?
Yeah, so we're mostly an Azure based organization.
So we're running Azure SQL.
We've got a data factory operating as our orchestration tool.

(35:49):
We're just actually just using stored procedures for all of our ETL.
Quick, easy, everyone can interpret SQL, plug in some Python and function apps where, you know,
where we need that additional bit more advanced than SQL or when we're working with our semi unstructured data is where we got Python playing in.
For the predictive analytics, we're integrating Databricks.

(36:09):
So that's based off of Azure.
And then for our BI tool, we do use Qlik.
It's been, yeah, I hadn't worked with Qlik up until this organization,
but it's a classic data visualization tool.
And yeah, and then we have our wrapper of sort of Azure DevOps on top for managing all these deployments and all that sort of stuff.

(36:31):
Well, I do have to ask you, Sam, as a Canadian, I stumbled across this at one point in time where two analysts were trying to troubleshoot the SQL together.
And the one analyst just couldn't figure out what the other one was referring to.
He said, yeah, it's the the table that I aliased a why couldn't he find the table?

(36:57):
Well, you know, sometimes you have to use different words up here.
No, yeah, it's there's actually Canadian English that has some different spellings sometimes.
So I will say I wish programmers allowed color to be the original spelling.
But I guess we have to sometimes switch to American English sometimes.

(37:20):
That was the trouble. You nailed it, Sam. Good job.
Great. Well, Sam, this has been great. Thanks so much for your time today.
Thanks so much for sharing about your organization, the work you're doing, the really advanced and exciting work with data that you're doing, the technologies you've been implementing.
It's been fun to hear about it for people who want to maybe reach out to you or find out more about your organization.

(37:42):
Where can people find you online?
Yeah. So our organization is CarryOwnless.com.
We are also, I think, active on a few of the social media platforms so you can find us there as well.
And for myself, I'm on LinkedIn.
So if you want to ever reach out and learn more, happy to answer any questions and connect as well.
All right. Thank you so much, Sam, for joining us today.

(38:04):
And folks, listeners, thanks so much for joining us as well.
That's it for today on Making Data Matters.
We'll see you again next time.
Advertise With Us

Popular Podcasts

Stuff You Should Know
Dateline NBC

Dateline NBC

Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.

On Purpose with Jay Shetty

On Purpose with Jay Shetty

I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!

Music, radio and podcasts, all free. Listen online or download the iHeart App.

Connect

© 2025 iHeartMedia, Inc.