Ask a firm who owns records retrieval and you will usually get a name, followed immediately by a qualification. She owns it, along with calendaring, client calls, demand drafting, and whatever came in that morning.
That qualification is the whole problem. Retrieval is not unowned in the sense that nobody knows it needs doing. It is unowned in the sense that it is the residual — the thing that happens in whatever time is left after the things with deadlines. Requests go out when someone has a spare hour, which means they go out inconsistently, which means the follow-up is inconsistent too, which means nobody can tell the difference between a custodian who is slow and a request that was never sent.
Why the clock is worse than firms assume
The HIPAA Right of Access rule at 45 CFR 164.524 gives covered entities 30 calendar days to respond, with one possible 30-day extension on written notice. Firms quote that framework a lot, and it is worth being precise about what it covers: it governs a patient's own request for their records. It does not apply uniformly to every attorney authorization, subpoena, or third-party disclosure.
So the real-world range runs from a few days to many weeks per custodian, with no single statutory clock to plan against. Multiply that by forty-five files with several providers each and retrieval is a full-time function. In most firms it is nobody's full-time function.
The failure is silent, which is what makes it expensive
A missed deadline announces itself. A stalled records request does not. It sits in exactly the same state as a request that is progressing normally — sent, no response yet — and the two are indistinguishable until someone goes looking.
Firms usually discover the difference at a file review, or at a demand, or at a deposition. By then the cost is not the delay. It is that the delay has already propagated into every downstream step that was waiting on the record.
What actually fixes it
Nothing about this is a diligence problem, so nothing about it is fixed by asking people to try harder. Four things change the behaviour:
Give it a single owner and a queue. Not a person who also does five other things — a defined function, with every open request in one place: provider, request date, authorization status, escalation history. If completeness is not visible at a glance, it is not being managed.
Standardize the request itself. Itemized, provider-specific requests naming date ranges and document types do measurably better than "any and all records," because most preventable delay is a clarification cycle that a vague request invited.
Put the follow-up on a cadence, not on a memory. Day 7, day 14, day 21. The cadence is less important than the fact that it is fixed and does not depend on anyone remembering.
Escalate by exception. Follow-up call, then written notice, then a flag for a subpoena decision. The person who owns the file should receive a short list of what is stuck, daily. The attorney should receive nothing at all unless a decision is required from them.
The part that matters more than the retrieval
The reason to fix this is not the records. It is what the fix does to the person who was absorbing it.
In every firm we have watched go through this, the same thing happens: the long-tenured paralegal who had been functioning as a records clerk goes back to case strategy, demand review, and client relationships — which is what she was hired for and what she is expensive for. She does not lose the function. She stops executing it and starts supervising it, which means her judgment is applied to exceptions instead of being spent on provider hold music.
That is the shape of the whole problem, and retrieval is only the clearest instance of it. The work that makes a case reviewable — retrieving, splitting, normalising, sequencing — is enormous, gates everything, and requires almost none of the judgment that the firm is actually short of.
Separate those two and the constraint moves. Leave them fused and you are paying senior review rates for document assembly, indefinitely.
Field notes
Four of thirty scenarios in this series. All are composites, built from patterns we see repeatedly rather than from single named engagements. Firm details are illustrative.
04 · Records requests with no owner
Personal injury · solo practitioner, 45 active files · review owned by the senior paralegal of nine years
One long-tenured paralegal knew every file in the office. She was also doing calendaring, client calls, demand drafting and record chasing at the same time, so requests went out whenever she had a spare hour — which meant they went out inconsistently.
What changed. Retrieval became a pipeline stage instead of residual time. Every open request landed in one queue with provider, request date, authorization status and escalation history. Requests were itemised and provider-specific, naming date ranges and document types, rather than "any and all records" — which is what invites the clarification cycles that cause most preventable delay. Follow-up ran on a fixed cadence at day 7, 14 and 21, and a stalled request surfaced to her for a subpoena decision rather than sitting silently.
Outcome. She stopped functioning as a records clerk and went back to case strategy, demand review and client relationships. Stalled files surfaced within days rather than at the next quarterly file review. Attorney contact: a Monday priority call and a Thursday sign-off queue.
05 · The calls that came in during depositions
Personal injury and auto · 2 partners, 1 case manager · review owned by the case manager · deployed with Specter review
Two partners split intake between themselves and their case manager. Calls arriving during depositions, independent medical exams, or after 5 p.m. went to a voicemail box nobody owned.
What changed. Clio puts the law firm missed-call rate at roughly 36 percent, with about a third of those callers never trying the firm again — and injury claimants typically contact several firms the same day. Every hour of the working week was mapped against the partners' actual court and deposition calendars to find the windows where nobody could answer, and those specific windows were covered. The screening script was built from the firm's own case acceptance criteria rather than a generic template, and every call landed in one log the case manager owned.
Outcome. Intake stopped depending on who happened to be free, and the case manager moved from being the intake backstop to owning the function.
06 · The comp solo who was his own intake department
Workers' compensation · solo practitioner · review owned by his paralegal of six years
He fielded every new client call himself, often from his car between hearings, because he did not trust anyone else to screen for statute of limitations problems and compensability red flags. His paralegal agreed with him, which is why it had lasted six years.
What changed. Nothing about his screening standard. His mental checklist — injury date, employer size, reporting timeline, prior claims, treating providers — was written down as firm-defined criteria and became the thing the system evaluated against. His paralegal owned it. Anything that did not fit cleanly was flagged to her, and only she escalated to him. Hearing hours got covered first, because that was the window costing him the most claimants.
Outcome. He kept complete control over case acceptance while closing the window where he had been losing claimants. Attorney contact: three a week, all on borderline acceptance decisions.
07 · Capacity, but no conversion
Premises liability and slip-and-fall · solo transitioning to small firm · review owned by her paralegal
She had automated much of her administrative work and genuinely had room to take more cases. She could not fill it. Inquiries arrived and did not convert, and her instinct was to spend more on marketing.
What changed. The funnel was audited before anything was spent. The constraint was response speed and follow-through, not lead volume — and with referrals cited by roughly 59 percent of solo and small firms as their leading acquisition channel, each individual inquiry carries more weight than it would in a paid-volume model. Buying more leads would have leaked the same percentage out of a bigger funnel. First contact got a single owner and a defined response window, and non-converting leads got a multi-touch sequence instead of one pass-or-fail call.
Outcome. Signed cases from existing lead volume improved with no increase in marketing spend. Evidence preservation also started days earlier on every matter, because the preservation letter and scene checklist were triggered at inquiry rather than at signing.