Intelligence
Your morning shouldn’t start with a query.
Reporting waits to be asked. By the time you have thought of the right question, built the right filter and found the right comparison, the window to do anything about it has usually closed.
Claritus.One does the asking overnight. What reaches you is a short, ordered list of things that need a decision, each one traceable back to the numbers that produced it.
The distinction
Dashboards tell you where to look.
Claritus tells you what deserves attention.
Those are not the same product. One assumes a person is watching. The other assumes nobody is.
The Daily Brief
This is what arrives at 6:14 AM.
Open any observation to see its evidence, including the checks that came back negative. Nothing here is asserted that you cannot inspect.
Good morning, Rachel.
Yesterday finished 0.0% above forecast on 0 visits. North Atlanta accounted for most of the variance and it was volume rather than acuity, so the revenue picture is unremarkable.
Two items need attention before today’s opening, both coverage. One item from Tuesday has closed. Wait time is the measure to watch this week: the network average has sat above target for six consecutive days, and four sites explain most of it.
Opening picture
- Centers opening today
- 64
- Need a decision before open
- 2
- Agents working now
- 4
- Waiting on your approval
- 1
- Closed since yesterday
- 1
What the narrative is built on
- Visits yesterday
- 0
- +4.8% vs forecast
- Network · 64 centers
- Door-to-door
- 0.0 min
- +0.8% wk/wk
- Against a 30 min target
- Provider coverage
- 0.00
- 3 sites below 1.00
- Weighted, today’s schedule
- Overtime hours
- 0
- −11.4% wk/wk
- Caught before 6.5 more
- Call abandonment
- 0.0%
- within standard
- 5% standard
- Eligibility exceptions
- 0
- 31 auto-resolved
- From yesterday’s visits
Needs a decision · in priority order
3 of 47 observations raised · 44 suppressed as routine
- Westlake coverageAgainst a 1.0 minimum standard, 4:00–7:00 PM
- 0.78
- Mountain Brook coverageAgainst a 1.0 minimum standard, 4:30–7:15 PM
- 0.82
- Forecast change since publishMountain Brook, driven by a competitor closure and a three-week arrival trend
- +14.2%
- Eligible providers with availabilityCredentialed for at least one of the two sites, inside weekly hour limits
- 4
- Overtime exposure if unaddressedAll four options land inside straight time
- $0
Recommended
Add 3.5 hours at Mountain Brook and 3.0 at Westlake from the eligible pool. No overtime required at either site.
Routes to · Staffing Agent
Site spotlights · last 30 days
Surfaced by variance, not alphabetically
- WestlakeWLKAttention · Atlanta West
- Visits
- 68
- vs fcst
- +9.4%
- Wait
- 47m
- Cover
- 0.78
- Mountain BrookMTBApproval needed · Birmingham
- Visits
- 74
- vs fcst
- +14.2%
- Wait
- 41m
- Cover
- 0.82
- North AtlantaNATWatching · Atlanta North
- Visits
- 88
- vs fcst
- +11.9%
- Wait
- 36m
- Cover
- 0.97
- BuckheadBKHAgent working · Atlanta Core
- Visits
- 91
- vs fcst
- +4.1%
- Wait
- 33m
- Cover
- 1.00
- MidtownMIDResolved · Atlanta Core
- Visits
- 82
- vs fcst
- +1.8%
- Wait
- 24m
- Cover
- 1.00
- BrookhavenBRKNormal · Atlanta Core
- Visits
- 55
- vs fcst
- +0.6%
- Wait
- 22m
- Cover
- 1.00
Illustrative demonstration data for a fictional 64-center network. Not customer results.
Under the hood
How an observation earns its place on the list.
Anomaly detection is easy to demo and hard to make useful. The difference is almost entirely in what the system compares against, and in how willing it is to stay quiet.
A baseline per site, per day-part, per season
A Saturday running at 60% of a Tuesday is normal. A Tuesday running at 60% of a Tuesday is not. Anomaly detection against a network average would raise the first and miss the second, which is why the baseline is built per site and per hour block rather than for the organization as a whole.
Elimination before assertion
Before Claritus tells you rooming time is the problem, it checks whether arrivals moved, whether provider pace changed, and whether the mix shifted. The checks that came back negative appear in the evidence, because knowing what was ruled out is most of what makes a finding trustworthy.
Forecasts that reach the published schedule
A demand forecast that nothing compares to your actual staffed hours is a chart. Every night the forecast is run against the schedule you published, per site and per block, and the difference is what becomes tomorrow’s short list.
Reasoning across systems, not inside one
Labor variance almost never explains itself from payroll. The answer usually sits between the schedule, the forecast and the visit log, which means it can only be found by something that holds all three in the same model.
Suppression as a first-class job
Forty-seven observations were formed this morning and three were raised. An intelligence layer that surfaces everything it notices is just a louder dashboard. Deciding what not to say is the harder half of the work.
Analytics
Ask in the language you already use with your team.
Self-service analytics quietly transfers the hard part to you. You still have to know which table, which join, which date logic, and whether the metric means what you think it means.
Here the question is the interface. What comes back is an answer, the evidence under it, and — the part that matters most — a plain statement of what the answer cannot support.
Ask the operation
Mountain Brook averaged 61.4 visits per open day over the last three weeks, against a forecast average of 57.8 — running 6.2% hot. The gap is not evenly spread. Weekdays are close to forecast; the overage is concentrated in the last four weekday evenings, which is when the competitor closure two miles away took effect.
- Actual
- Forecast
What stands out
- The five highest days are all Mondays or Thursdays. The pattern is a day-part effect, not a general lift.
- August 18 reached 74 visits against a forecast of 61, the largest single-day variance in the window.
- Weekend volumes track forecast almost exactly, which is why the weekly average understates the weekday problem.
- Provider hours were unchanged across the entire window, so coverage per visit fell as volume rose.
The forecast in this window was published before the nearby closure was known, so part of the variance is a stale baseline rather than a forecasting error. Rebaselining is scheduled for the next cycle.
Follow this further
- What did door-to-door time do on the four highest days?
- Compare Mountain Brook to the rest of the Birmingham market.
- What would coverage need to be if this volume holds?
Or hand it off
See your own brief
Bring last month’s hardest operating question. We’ll show you the brief that would have raised it on day two.
The most useful demo is not a feature tour. Tell us about a labor spike, a coverage miss or a market that went sideways, and we will walk through how Claritus.One would have surfaced it, explained it, and routed the work.