Quick answer. A retrieval vendor absorbs the requesting and chasing and hands you PDFs. A dedicated hire, roughly $15 to $20 an hour or $12,000 to $24,000 a year offshore, does the same work plus everything the vendor will not touch. Automation covers the third thing neither of them does: knowing, at any moment, which of your active cases is missing which document and how close it is to a deadline. Most firms need a vendor or a person and the tracking layer. Buying one and calling it solved is how cases go quiet.
If you want to know what records work really involves, do not read a vendor's brochure. Read a job posting written by a firm that is drowning in it.
A New York plaintiff firm recently advertised for a paralegal to run pre-litigation records acquisition. They were explicit that this was not a general legal assistant role but a specific, repeatable records-chasing function. Their published checklist:
- DMV crash report searches by county and date, using the state's crash report request form
- Police collision reports through the online retrieval portal, with direct calls to precincts when the report has not been posted
- Freedom of information requests for police radio-run records and fire-department EMS patient care reports
- Chasing the client's own no-fault carrier claim file for defendant identification when no police report exists at all
- Preservation letters to businesses and homeowners for third-party camera footage, sent within days of signing
- Identifying municipal vehicle exposure and calendaring the 90-day notice of claim deadline that comes with it
- Maintaining a per-case records checklist in the case management system with hard service levels, requests submitted within 24 hours of signing
- Flagging cases where defendant or location data is simply unrecoverable, as triage rather than abandonment
Medical records and billing from every treating provider run alongside all of that. The firm added one line that tells you everything about how they had been burned before: they wanted someone who could work a high-volume queue with service-level accountability, "not billing hours to look into it."
That is the actual job. Now here are the three ways firms buy their way out of it.
Option one: a retrieval vendor
You send a signed authorization and a provider list. The vendor submits requests, follows up, advances copying fees, and returns records as PDFs with an invoice. Pricing is normally per request plus per page and varies widely by state, provider, and whether the record is electronic. Get quotes against your own case mix rather than a headline rate, because the variance is the whole story.
What you are actually buying: the phone calls. That is a real product and for many firms it is worth the money on its own.
What you are not buying:
- Status inside your own system. The vendor knows what it has requested. Your matter in Clio or Filevine usually does not, unless someone re-keys it.
- Coverage of the non-medical half. Crash reports, FOIL requests, radio runs, preservation letters, and carrier claim files typically fall outside a medical retrieval scope. Those come back to you.
- Deadline awareness. No vendor is watching your notice-of-claim clock.
- A view across cases. You can ask about one case. Asking which of your 200 open files is missing a bill is a question the vendor cannot answer and your system usually cannot either.
Option two: a dedicated hire
Firms do this constantly, and the market rates are public. Records and intake staff are hired at roughly $15 to $20 an hour. Full-time offshore legal coordinators are advertised at $1,000 to $1,050 a month, and paralegals with case management system experience at $1,400 to $2,000 a month. Call it $12,000 to $24,000 a year for one person, indefinitely.
What you get: flexibility. A person handles the precinct that only answers on Tuesdays and the provider that still wants a fax, and they do not care whether the record is medical.
What you get in the bargain:
- The knowledge lives in one head. Which counties are slow, which portal is down, which provider needs a specific authorization wording. When that person leaves, it leaves.
- No visibility. A person doing this well looks identical to a person doing it badly until a case blows up. There is no dashboard.
- It does not scale linearly. Doubling caseload does not mean hiring 1.0 more people. It means hiring 1.4 more, because coordination cost grows too.
Option three: automate the loop
This is the option most firms have never been shown, so it gets oversold and then mistrusted. Being precise about it matters.
What genuinely automates:
- Request generation from matter data. The matter already holds the client, date of loss, providers, and carrier. The request letter, the authorization, and the cover sheet are assembled from those fields rather than typed. Same input, same output, every time.
- A request register. Every request logged with what was sent, to whom, on what date, and what came back. This is the single highest-value piece and it is barely software, it is discipline made structural.
- Escalation on a schedule. Fourteen days with no response produces a follow-up. Thirty days produces a task for a human with the file already assembled.
- Deadline calculation. A case with a municipal defendant gets its notice-of-claim date computed at signing and written to the calendar. Rules-based date math is exactly what computers are for, and it is the piece that saves cases rather than saving minutes.
- The cross-case view. One screen: every open matter, what is outstanding, how old it is, what deadline it sits against. This is the question no vendor and no individual can answer, and it is answerable the moment the register exists.
What does not automate, and anyone telling you otherwise is selling:
- A precinct that has not posted the report. Someone calls.
- A provider with no electronic intake. Someone faxes.
- A case where the defendant is genuinely unidentifiable. Someone decides.
- Judgment about whether an incomplete record is worth chasing further.
Automation shrinks the job. It does not delete it. A firm that automates the register, the escalation, and the deadline math typically keeps its vendor and keeps a person, and that person stops spending their week on status spreadsheets.
Once the records arrive, the second bottleneck starts
Getting records is half the work. Turning them into something usable is the other half, and it is worse. A single medical chronology commonly runs 15 to 80 hours by hand depending on file thickness, and a firm carrying around 40 active cases can spend roughly $17,600 a month on manual records review. Outside chronology services charge on the order of $300 to $800 per package.
The same records also feed deposition summaries and paid-versus-incurred bill analyses, all built the same manual way from documents the firm already possesses. We wrote that side up in detail in building medical chronologies from records already in Clio. If you are evaluating records workflow end to end, read that alongside this one, because fixing retrieval and leaving the review manual just moves the queue.
How to decide, in about twenty minutes
Pull three numbers off your last quarter.
- Requests per month. Not cases. Requests. A single auto case can generate eight.
- Average days from signing to complete file. If nobody knows, that is the finding.
- Cases where a record arrived late enough to matter. Late demand, missed deadline, a settlement discussion held without a bill in hand.
If number three is zero and number two is under 60 days, you do not have a records problem and should spend the money elsewhere. If nobody can produce number one without a week of digging, the tracking layer is your first purchase regardless of which of the other two options you choose, because you are currently managing this on faith.
Where Oktopeak fits, and where we do not
We do not retrieve records. We are not a retrieval vendor and hiring us as one would be a mistake. Keep your vendor.
What we build is the layer around it: request generation from matter data, a register that lives inside Clio or your case management system, escalation rules, deadline calculation for the dates that end cases, and the pipeline that turns delivered records into chronologies and damages summaries with attorney review built in. It runs on the Clio API, and we shipped the open-source Clio connector so you can read our integration code before you call us. For personal injury specifically, the PI case management page covers the wider build.
The first step is free and deliberately narrow: a 30-minute review of your records week where we tell you which part is worth building, which part your vendor already covers, and which part should stay a phone call.
