Your homelessness plan needs more than a forecast
A council asks what it will take to reduce homelessness over the next ten years.
At first, the request sounds straightforward. Then the practical questions start:
How quickly is homelessness growing?
Which investments would make the greatest difference?
How should limited resources be divided across prevention, housing and people with different levels of need?
Most teams have pieces of the answer spread across administrative systems, program reports, databases and spreadsheets. The hard part is bringing those pieces together in a way that lets staff test choices—not just describe the current situation.
That is what Karto’s ten-year homelessness modelling tool is built to do.
A forecast tells you what may happen. A model lets you test what could change it.
Traditional planning often produces one forecast based on one set of assumptions. That can be useful, but it can also create a false sense of certainty.
No one knows exactly how quickly homelessness will grow over the next decade or how many people will resolve their housing situation without entering the formal system. Those assumptions have a major effect on the result.
Karto makes that uncertainty visible. Using local data, it creates 25 ten-year estimates based on different combinations of two important assumptions:
how quickly homelessness is growing; and
how many people are able to self-resolve.
Staff can review the range and select the baseline they believe best reflects local conditions if current services and capacity remain the same. The point is not to pretend that one projection is guaranteed. It is to make the assumptions clear, choose a defensible starting point and understand how much the outlook changes when those assumptions move.
Start with the system you have
The model begins with aggregate, community-level information about the local homelessness response:
the number of people experiencing homelessness;
program and housing capacity;
program performance; and
program costs.
That information may come from HIFIS, another administrative system, a database or existing spreadsheets. Karto does not require a community to replace the systems it already uses.
Once the current system is represented, the team can see what may happen over ten years if services and capacity stay roughly the same. That becomes the baseline for testing possible changes.
Test the decisions before they become commitments
This is where modelling becomes more useful than a static forecast.
Staff can build different scenarios by changing the amount, timing or mix of interventions, including:
prevention and diversion;
program capacity;
supportive housing;
deeply affordable housing;
program performance;
available budget;
program and housing costs;
the people each intervention is intended to serve; and
how capacity is divided across different levels of acuity and housing risk.
Each scenario can represent a genuine policy or investment choice. One might put more resources into preventing people from entering homelessness. Another might increase supportive housing in stages. A third might combine prevention, added program capacity and new housing while staying within a fixed budget.
Teams can also test how the results change when costs rise, program performance improves, or limited housing and service capacity is directed toward different groups.
For example, a community could compare a scenario that focuses heavily on people with high acuity with one that serves a broader range of needs. This can show the effect on homelessness, waitlists and system pressure, including what happens to people whose needs are lower today but may become more serious while they wait.
The model can also test how available capacity is divided between people currently experiencing homelessness and people at immediate risk. A community might direct 80% of capacity toward people experiencing homelessness and 20% toward prevention, then compare the result with a different allocation.
Karto then shows how each option changes:
homelessness over ten years;
the number of people housed or prevented from entering homelessness;
pressure on the homelessness system;
total cost;
remaining need; and
outcomes for people with different levels of acuity and housing risk.
Instead of discussing interventions one at a time, teams can compare complete approaches side by side. That matters because homelessness systems do not operate as a collection of isolated programs. A change in one part of the system affects pressure, flow and cost elsewhere. Decisions about who is prioritized also affect who remains on the waitlist and how their needs may change over time.
An example: one question, several credible answers
Imagine a mid-sized municipality where homelessness has been rising, shelter use is consistently high and several housing projects are still years away.
Council asks staff to recommend the best use of a new ten-year investment.
The homelessness team could use Karto to compare three approaches:
expand prevention and diversion first, with modest additions to existing program capacity;
prioritize supportive housing, phased in as projects become available; or
combine prevention, housing and targeted capacity increases within the same overall budget.
Within each approach, the team could test different cost, performance and targeting assumptions. It could examine whether supportive housing is directed primarily toward people with high acuity, whether some capacity is available to people with lower or moderate needs, and how much investment is directed toward preventing homelessness.
For each option, staff could change the timing and scale of the interventions, rerun the scenario and compare the projected effect on homelessness, system pressure and cost. They could also test how the results change if costs increase, programs perform differently or the balance between prevention and homelessness response shifts.
The result would not be a generic recommendation produced somewhere outside the community. It would be a transparent comparison based on local data and assumptions the team can explain.
If council asks what happens when housing delivery is delayed, staff can rerun the model. If a new funding opportunity appears, they can test a larger intervention mix. If council asks about the consequences of focusing mainly on people with the highest acuity, staff can show the effect across the rest of the system. If local conditions change next year, they can update the inputs rather than commission a new study from the beginning.
A better conversation across departments
Homelessness decisions rarely sit with one team.
Program leads understand how services operate. Analysts know the data and its limits. Finance needs to understand cost. Senior leaders need to know which option is practical and what they can defend in a budget or council report.
A shared model gives those groups something concrete to work through together. The conversation can move from “Which program sounds most promising?” to more useful questions:
Which assumptions are driving this result?
What does this option change over the full ten years?
Where does system pressure remain?
What happens if implementation is delayed?
What can we achieve within the available budget?
Who is being prioritized, and what does that mean for others waiting for support?
How would different cost or performance assumptions change the result?
Archer, Karto’s AI planning assistant, can help staff review assumptions, explore plain-language “what if” questions and compare intervention mixes. This can include questions about costs, performance, acuity and who the system is designed to serve. The model—not Archer—performs the calculations, and staff remain in control of any changes.
Built for decisions that need to hold up
Homelessness plans influence budgets, funding requests and public commitments. The numbers behind them need to be understandable and repeatable.
Karto’s calculations are deterministic: the same inputs produce the same results. It works with aggregate planning data, not individual client records. That gives teams a consistent model they can review internally and explain to decision-makers.
HelpSeeker has completed housing and homelessness modelling with more than 100 communities across Canada. Karto brings that experience into a tool municipal and community teams can use themselves.
Keep the model alive after the plan is approved
A ten-year plan should not become a document that is opened once a year to see whether reality still resembles the original forecast.
Conditions change. Programs improve. Costs rise. Housing projects move. New funding becomes available. Council asks a different question. The needs of people on the waitlist may also change, requiring the community to reconsider how limited capacity is targeted.
Karto gives teams a way to return to the model, update what has changed and test the next decision. The value is not only in producing a projection. It is in keeping the reasoning behind the plan active, visible and useful.
Want to test a question your community is already working on? Book a Karto working session.