How to Compare Medical-Legal Vendors Without the Theater

The fairest way to compare medical-legal vendors is to send every candidate the same test file under the same written brief, then score what comes back against problems you already know are in that file. A demo built on the vendor's own sample shows you that sample. It tells you very little about what happens to your records.

I work on the record-processing side of medical-legal work at LezDo TechMed: OCR, sorting, page referencing, and the quality checks that sit between a raw production and a usable chronology. From that seat, the comparisons that go wrong look similar. A buyer sees three polished samples, picks the one with the cleanest formatting, and finds out on the first live matter that the cleanest format hid the weakest page handling.

This post is for the attorney, paralegal, claims lead, or life care planner who has to pick between vendors and wants the comparison to mean something. If you want the list of questions to ask before choosing, 12 Questions PI Firm Owners Should Ask Before Choosing Medical Record Review Support covers that. This post is about how to test.

Why does a polished sample tell you so little?

A polished sample tells you little because the vendor chose the file, the brief, and the edit. Every hard problem in medical records (bad scans, duplicate encounters, missing months) was either absent or already solved before you saw it.

That is the theater part. Nobody has to lie for it to happen. A sample is built to impress, and it is built from a good day.

What you need is the opposite: the vendor's normal day, on a file you control. If you are choosing a narrative summary provider specifically, how attorneys should choose a medical summary partner covers that pilot.

What should go into a comparison test file?

A good comparison test file is a real, closed matter you already know well, with its problems still in it. You know the answer key, so you can grade every return without guessing.

Clear the privacy side first. Use a set your firm is permitted to share for evaluation, de-identified where possible, and get your confidentiality terms in place (including a business associate agreement where HIPAA applies) before any page moves.

Then check that the file carries the problems you see on live matters. If it does not, pick a different file. Do not edit one to manufacture them. Useful ones to look for:

• A duplicate encounter. The same visit produced by two sources, such as a clinic and a hospital billing department.
• A quiet stretch. A few months with no notes in the returned set, where you know whether records exist and are outstanding.
• A hard scan. Faxed or handwritten pages where OCR struggles, so someone has to read by hand.
• A facility on the intake list that never responded. The correct return names it as outstanding instead of skipping it.
• Pages out of order. A batch where dates jump backward inside one PDF.
• For life care planning buyers: a recommendation with no documented frequency. The physician recommends a therapy, but nobody wrote how often or for how long. The correct return flags it. It does not fill it in.

Six traps is plenty. You are not trying to break the vendor. You are trying to see the vendor's normal habits.

How do you keep the brief identical across vendors?

Keep the brief identical by writing it once, sending it unchanged, and answering every clarifying question in writing to all candidates at the same time. If one vendor gets a phone call that explains the file, the comparison is already tilted.

The brief should name the deliverable, the citation format, the exact phrases you want for gaps ("not in returned set," "outstanding"), and the boundary: organize and flag what the records document, with no diagnosis, causation, or future-care conclusions. Give every candidate the same deadline. A reasonable one, too. A rushed test measures panic.

How to score the returns side by side

Score the returns against your answer key, one check at a time, with the vendor names hidden if someone else on your team can strip them. Formatting preferences come last.

That table takes an afternoon per vendor. It is the most useful afternoon in the whole buying process.

If a return fails the cites check, stop there. Everything downstream of a bad citation is something your team will have to rebuild.

What does a second wave of records reveal?

A second wave reveals whether the vendor can update a file without breaking it. Live matters almost never arrive in one production, so this is the test that predicts month three.

Hold back part of the test file, say one facility's records. After the first return, send it with a short note. Then look at four things. Did the new pages get sorted into date order with the rest? Did the old cites still work, or did a re-number break them? Did the vendor tell you what changed, in writing? Did any earlier gap label update correctly once the missing records arrived?

This is where sorting and indexing medical records stops being back-office work and becomes the thing you are paying for. A vendor with clean intake habits absorbs a second wave quietly. A vendor without them sends a new file that no longer matches the old one.

Where do AI-assisted workflows fit in the comparison?

AI-assisted workflows get scored the same way as everything else: on the return, not the description of the tooling. Your hard scan and your duplicate will show whether a human checked those pages. If every vendor on your list is pitching AI, how to buy medical-legal services when every vendor claims AI covers how to check those claims.

What a vendor's return should never include

A vendor's return should never include conclusions that belong to someone else. A medical record review organizes and flags what is documented. It should not diagnose, decide causation, or state what care a person will need.

For life care planning work this line matters even more. The plan rests on what the treating providers documented, and the future-care determinations stay with the planner of record and the treating physicians. A return that "helpfully" supplies a frequency the record never stated has failed your sixth trap, however polished it looks. If your work feeds plans, it is worth reading how LezDo scopes medical record reviews for life care planners against that boundary.

Security and privacy questions deserve their own written answers alongside the scorecard. Checking whether a record review vendor is compliant walks through what to ask.

After the scorecard

After the scorecard, you usually have one clear leader and one close second, and the difference shows up in a handful of specific rows. That is a decision you can explain to a partner or a claims manager, which a demo impression never is.

If you want a reference for what a careful reviewer does inside a chart, our guide to medical record review techniques gives your scorers a shared vocabulary. And if LezDo TechMed is on your list, our page on medical record review for attorneys explains how we scope a first file. Send us the same test you send everyone else.

When buyers compare medical-legal vendors on a file they already understand, the theater has nothing to perform for.