Disaster Case Management Software: Verified Loss, Duplication of Benefits and the Committee That Releases the Money

Why recovery cases break the tools built for programme case management

It is a Tuesday evening seven months after the storm. Nine people from six organisations are sitting in a church hall as the unmet needs committee. Case 412 is a widowed homeowner, uninsured for flood, whose roof and subfloor need $41,000 of work. Her case manager presents: federal individual assistance paid an amount, a small business loan was approved and not accepted, her homeowner policy paid something for wind but not for water, one faith based partner already bought her appliances, and the state programme has her application pending. The committee has to decide how much to release tonight, and the arithmetic that determines the legal answer is being read aloud from a folder.

Get it wrong and this is not a bookkeeping error. The Stafford Act prohibits duplication of benefits, federal grantees are required to recover duplicated assistance, and a subrecipient that awarded twice for the same repair can be required to pay it back long after the money is in a contractor's account and the roof is on. In practice the recovery group either freezes, sending cases back for more verification while a family lives in a damaged house, or moves and hopes. Both outcomes are common and both are avoidable.

The tools in the room are real. The Coordinated Assistance Network exists precisely so relief agencies can see that a household is already known to someone else, and during the response phase it earns its place. Apricot by Bonterra and CaseWorthy are capable configurable case management platforms used across human services. What none of them model natively is the thing this work turns on: verified loss broken down by damage component, an assistance ledger categorised the way federal duplication rules categorise, and the arithmetic between them. So the recovery group keeps a workbook of awards beside the case system, and the workbook is the real system.

Problem 1: duplication of benefits is category arithmetic, not a yes or no

The question is never whether a household received assistance. It is whether they received assistance for the same need. A federal award for personal property does not offset a roof. A wind settlement does not offset flood damage to a subfloor. An approved loan that the household declined is treated differently from one they accepted, and the treatment varies by programme and has been the subject of specific federal guidance for community development block grant disaster recovery funds. A volunteer built ramp has a value that may or may not count depending on your funder.

A case note that says received FEMA assistance carries none of that. Neither does a single total assistance field, which is what generic case management gives you.

What a custom build does: assistance is recorded as line items with source, programme, category, amount, date, and status such as approved, accepted, declined or pending. Verified loss is recorded by component. The duplication engine matches category to category and produces an unmet need per component with the working shown. When the committee sees $18,400 available for structural repair, they also see exactly which awards were offset against which losses, and so does the auditor two years later. That transparency is the product.

Problem 2: verified loss is the number nobody actually stores

Most recovery groups have plenty of documentation of what a household received and thin documentation of what they lost. Damage assessments arrive as a federal inspection summary, a contractor estimate, a photo set from a volunteer team leader, and a case manager's narrative. They disagree with each other, they cover different scopes, and none of them is stored as a structured number tied to a component of the home.

What a custom build does: damage is captured by component, meaning roof, envelope, structure, electrical, plumbing, mechanical, flooring, contents, with an amount, a source, a date and the document attached. Where sources disagree, the record shows all of them and the verification standard your programme adopted. Contractor estimates that arrive later update the component rather than replacing the file. Because the loss side is structured, the unmet need calculation is reproducible instead of being a case manager's judgement recorded in prose.

Problem 3: the committee is the workflow and it runs on paper

The unmet needs committee is where the money actually moves, and in most recovery groups it is supported by a case manager printing packets on Monday night. Cases are deferred for missing documents, which means a family waits a fortnight for a piece of paper somebody could have requested three weeks earlier. Decisions are recorded in minutes that are not attached to the case. Conditional approvals, the ones that say yes if the state programme declines, are tracked in someone's memory.

What a custom build does: a case becomes committee ready only when its required evidence is complete, so deferrals for missing documents stop being the main reason cases stall. The agenda assembles itself with the duplication calculation attached to each case. Decisions are recorded against the case with the conditions expressed as rules the system can watch, so when the state programme finally denies the application in April, the conditional award activates and notifies the case manager rather than waiting to be remembered. Disbursement is tracked through to the contractor with lien waivers and completion evidence, because paying for work that was never finished is the second most common finding after duplication.

Problem 4: several agencies share one survivor and none share a database

A household may be worked by a national voluntary organisation, a faith based partner, a legal aid provider and a state funded case manager at the same time. Each holds part of the story, each operates under its own privacy commitments, and some of them cannot show you their notes at all. The result is duplicated intake interviews with a traumatised family, and a committee that does not know an award was made across town last week.

What a custom build does: partner organisations get scoped access rather than a login to everything. A partner can see that a household is active, which components have been addressed, and the assistance ledger entries relevant to duplication, without seeing counselling notes or immigration status. Consent is recorded per organisation, per category, with an expiry, and enforced when the record is read. Contribution goes both ways, so a partner who records an award makes the ledger accurate for everyone. This is the design decision that determines whether other agencies actually participate, and building it as an afterthought never works.

Problem 5: recovery outlasts the staff and the grant

A major disaster produces cases that stay open for three to five years. Case management grants are funded in shorter cycles. Staff turn over completely, sometimes more than once. A case handed to a new manager in year three consists of whatever the previous manager wrote down, and if the record is prose, the new manager restarts the family's story from the beginning.

What a custom build does: the case carries structured state that survives handover. What is verified, what is outstanding, what was awarded and by whom, which conditions are live, what the household said about their own priorities. Reassignment is an operation with a handover summary generated from the record rather than a folder handed across a desk. And when a second disaster hits the same county, which happens, the system knows which households are still open from the first one, which is information nobody has today.

What this costs and how long it takes

Across the 2,000-plus projects Digital Heroes has delivered, a focused first release, meaning the household and damage component model, the assistance ledger, the duplication of benefits engine and case management with document capture, runs $70,000 to $140,000 and ships in 10 to 16 weeks. Adding the committee workflow with conditional awards, partner organisation scoped access, contractor disbursement tracking, volunteer work order integration and funder reporting brings the total to $180,000 to $400,000 across 6 to 12 months.

What drives cost up in this category: the number of funding sources with distinct duplication rules, since a programme carrying community development block grant disaster recovery money has stricter and more specific requirements than one distributing only philanthropic funds. Partner count, because every additional organisation is a consent design conversation before it is an integration. Construction management scope, if the recovery group is acting as the builder rather than referring out. And urgency, because standing a system up in the weeks after an event costs more than building it in a calm period.

That last point deserves emphasis. The best time to build this is when nothing is happening. Recovery organisations that pre position a system, even a lean one, are taking intake in structured form on day four instead of collecting a thousand paper forms they will spend six months typing up. If you are reading this during a response, build the intake and the ledger now and add the committee workflow in month three.

When buying is the right answer

Buy if you are a single agency handling under roughly 200 households, distributing your own funds, with no committee spanning multiple organisations. Configure CaseWorthy or Apricot, keep a disciplined ledger, and put the money into case managers. Participate in the Coordinated Assistance Network regardless of what else you do, because knowing that another agency is already working with a family is worth having and you should not rebuild it.

Build when two or more of these are true. You are moving over roughly $3M of unmet needs funding. Your committee draws from several organisations who need different views of the same household. You administer or expect to administer federal disaster recovery funds with formal duplication requirements. Your cases will run for years across staff turnover. Or you have already been asked by a funder to demonstrate how a specific award was calculated and the answer took a week to assemble.

How to choose a developer for disaster recovery systems

Ask them to model duplication of benefits on a whiteboard. If they draw a total assistance field, they will build you a case tracker and you will keep the workbook. The answer you want has categorised line items, statuses including declined and pending, verified loss by component, and a calculation that shows its working.

Ask how a partner organisation sees part of a case without seeing all of it, and how consent is captured and expired. If the answer is separate logins with different roles, ask what happens when a partner needs to see the ledger but not the notes on the same household.

Ask what happens to a conditional award when the condition resolves eight months later. A developer who has done this builds a watcher. One who has not will suggest a reminder field.

Ask about offline and low connectivity operation for intake, because early intake happens in shelters, parking lots and neighbourhoods without power. Then settle ownership before kickoff: repository, cloud accounts, data export and the right to hire anyone else. At Digital Heroes the organisation owns the code from the first commit, which matters in a sector where the entity that built the system may not exist in five years and the cases will.