If our clients were a hundred people
One slide, several toggles — age, gender, reach. Normalising to a hundred makes the shape comparable week to week without publishing raw counts, which also happens to be the safer disclosure.
DCHS Insights CX Human-Centered Design @ DCHS PROTOTYPE no real data DCHS Insights CX How We Spotlight — Who we served
The week, as a hundred people
Age, language, and reach in aggregate. The group with the most obvious appeal and the sharpest small-numbers and purpose questions attached to it.
What this is
This group turns the week's volume into a picture of who was actually in the building. “If our clients were 100 people” is the most immediately engaging idea in the deck, and language access is the measure most directly connected to whether somebody got served well.
It is also the group where publishing carries the most weight. Every field on these slides is a question somebody had to answer at a desk, and a small enough group is an identifiable one. Both facts should shape what runs, not just how it looks.
What could go in it
Options to react to, not a recommendation. Strike, keep, and add — reacting to a concrete list is far more productive than starting from a blank page.
Language access
Share by language, and separately whether the request was met. The second half is the one that says something about service quality rather than demographics.
First visit versus returning
How much of the week was new people. Arguably the single most useful client-side number for anyone planning outreach, and not currently in the deck.
Geography, against Census context
Which parts of the city walked in, compared with where need is concentrated. This is where the community-fact slide and the client profile could reinforce each other instead of sitting apart.
Households rather than individuals
How many of the week's visits were the same family arriving separately. Changes the story about volume considerably, and is likely the hardest of these to produce.
Questions to settle
These are the decisions that have to be made before anything here can run every Monday. Most of them are editorial rather than technical.
What is this slide for?
- To check whether who we serve resembles the city? To prepare staff for who is walking in? To surface a gap?
- A demographic slide either changes something or it is decoration. Which is it here?
- If the answer is “it is interesting”, that may still be enough — but say so out loud.
Where does this data come from, and what did it cost the client?
- Every field here is a question somebody was asked at a desk, often at a bad moment.
- Are we publishing what we already collect, or creating a reason to start collecting more?
- Optional fields produce a distribution of who chose to answer, not a distribution of who came.
What is the smallest number we will print?
- A language with four speakers in one week is a small enough group to identify.
- Do we set a floor and aggregate below it — and does landing in “Other” read as not mattering?
- Does the answer change if this is ever forwarded outside the department?
Compared with what?
- Twenty-one percent Spanish means nothing on its own.
- Against the city's share, against last year, against the population eligible for the service, or against nothing?
- Each of those baselines tells a different story and invites a different reaction.
Requested, or met?
- Share of clients by language, share of interpretation requests, and share of requests that went unmet are three different slides.
- The third is the honest one and the hardest to publish.
- Is there an appetite for a measure that can make us look bad in a week where we were short-staffed?
What it would take
What each slide needs, and the question that has to be answered before we know whether we can produce it. The systems are named loosely on purpose — which system holds what is exactly what the working group should nail down.
| Metric | Where it would come from | What we would need | The open question |
|---|---|---|---|
| Client demographics in aggregate | The case system, or whatever records intake | Weekly aggregate counts by band — never records, never anything row-level | Which fields are reliably completed at intake, and which are optional enough that the resulting picture is really a picture of who answered? Who can answer this: ______________ |
| Languages served | Intake, plus whatever logs interpretation requests | Weekly counts by language, and separately requests made and requests met | Is language recorded as the client's preference, or as the language the encounter actually happened in? Those are different slides with the same title. Who can answer this: ______________ |
| First visit versus returning | Whatever assigns a client identifier | A reliable way to recognise the same person across weeks and across programs | Does that identifier hold across centers, and is matching good enough that “new” really means new? Who can answer this: ______________ |
| Households rather than individuals | The case system | A household relationship that is maintained, not inferred from an address | Does a household identifier exist, and is it reliable in the programs where it matters most? Who can answer this: ______________ |
| Census context | Public ACS tables | The comparable share for the city, with the vintage stated on the slide | Which geography and which release — and are we comparing our clients with the whole city, or with the population actually eligible for the service? Who can answer this: ______________ |