Some loss is priced into every service business, so a portfolio total does not tell you much on its own. The spread does. In the demonstration data above, the worst location misses about four times more per work order than the median one. In practice a gap like that usually comes down to paperwork habits at one branch, which makes it fixable.
When a branch manager sees their own number next to the portfolio median, the number starts to move. Most customers end up folding it into their monthly operating review.
IF EVERY LOCATION MATCHED YOUR MEDIAN
Bringing every location to the median would be worth about $96k a year in the demonstration portfolio. Half of that sits in nine locations.
WHAT WE REPORT
Recovery per work order, by location and technician. Every figure carries a confidence interval and the sample size behind it.
WHAT WE DO NOT REPORT
Work that was never written down, and collections outcomes. The audit ends at the re-bill packet; your AR team owns the follow-through.
In most roll-ups the management company pays the bills and the operating entities do the work. The account model keeps those two separate, the same way your accountants do.
Every operating entity is its own tenant with its own intake address and price book. The parent org reads across all of them; children never read each other. Row-level security enforces this inside the database, and the cross-tenant test attempts the forbidden read in CI.
The management company holds the Stripe relationship and the annual invoice, paid by ACH. Locations onboard and upload without ever touching billing. Adding an acquisition is a new child org, not a new contract.
Per-location findings aggregate to portfolio views: recovery per WO, acceptance rate, exception depth. Exports are plain CSV with stable column names, ready for whatever your analysts already use.
Locations are compared within trade and ticket band, never across them. A septic branch and a garage-door branch miss different things, and comparing them directly would produce rankings that do not mean much.
Benchmarks publish with a confidence interval and the sample size behind them. A location with three weeks of history shows wide intervals and a "young data" flag rather than a false rank. Intervals narrow as history accrues; typically four to six weeks of paper before quartile assignments stabilize.
RECOVERY PER WORK ORDER, HVAC, $300 TO $700 TICKET BAND
Demonstration distribution. 90% CI on the median: $26.10 to $33.00, n = 14 locations.
The details your security review will ask for, in one place.
Per-tenant row-level security in Postgres (Supabase), US region. Parent orgs hold read-across scope; sibling entities are invisible to each other. The forbidden cross-tenant read is a standing automated test.
Sign-in is by emailed link only; there are no passwords to steal. Roles: portfolio reader, entity operator, admin. Admin surfaces sit behind Cloudflare Access with hardware keys. Every consequential action lands in an append-only audit log.
TLS in transit, AES-256 at rest. Source documents delete after 90 days by default, configurable per entity. Deletion requests complete inside 30 days and cascade through storage. We are not yet SOC 2 certified and will not imply otherwise; the audit is scheduled.
Subprocessors: Cloudflare, Supabase, Anthropic, Stripe, Resend. The list, plus sample MSA and DPA, downloads from the diligence page without a sales conversation.
Annual agreement, invoiced to the management company, paid by ACH. Pricing scales with locations and work-order volume rather than seats, so you can give read access to anyone who needs it.
The assessment produces a portfolio-specific estimate and a scope worksheet. The figure below is the actual starting price.
Four steps, about two minutes. You get an estimate with a range and a link you can send to your partners. No email required to see the number.
Operating entities under the parent. Estimates only; nothing here commits you.