OCR Subpoenaed Records: Making Third-Party Paper Usable

To make subpoenaed records usable, preserve the files exactly as the third party sent them, log and number a working copy, then run OCR so the scans become searchable text and the statements become tables. Keep the custodian's declaration with the set, and check your court's rules on sharing the records with the other parties.

What do subpoenaed records look like when they arrive?

Subpoenaed records arrive in whatever form the third party's records department chose: a box of photocopies, one long PDF by email or secure link, a disc, or a fax. Expect a cover letter, a declaration from a custodian, and pages with no numbers on them and no text behind them.

Federal Rule 45 says little about what a paper response must look like. Under Rule 45(e)(1)(A), "A person responding to a subpoena to produce documents must produce them as they are kept in the ordinary course of business or must organize and label them to correspond to the categories in the demand." Nothing in that sentence promises a searchable file, a file per document, or page numbers.

Look the delivery over before anyone starts reading it. The checks in this table tell you how much work the set needs.

Take a fictional wage case, Okafor v. Brightline Staffing. Three subpoenas went out. Marrow County Credit Union mailed a disc holding one 412-page PDF of statements. Pinecrest Wireless sent a link to a 96-page PDF of call records. Delmar Logistics, a former employer, delivered a banker's box of personnel and payroll paper. Three custodians, three forms, and none of it searchable.

Why do third parties produce records as unsearchable scans?

A third party can produce unsearchable scans because Rule 45 asks for a particular electronic form only when the subpoena names one, and a response put together from paper, printouts or faxes is a set of pictures. When the subpoena is silent on form, the responding person chooses within a broad default.

The two provisions are worth reading side by side. Rule 45(a)(1)(C) says "A subpoena may specify the form or forms in which electronically stored information is to be produced." Rule 45(e)(1)(B) supplies the default: "If a subpoena does not specify a form for producing electronically stored information, the person responding must produce it in a form or forms in which it is ordinarily maintained or in a reasonably usable form or forms." Whether a scan of a printout meets that default for a given set of records is a question for the lawyer and, if it comes to it, the court. State subpoena rules differ, so read the one that governs yours.

On your own set, look for which of these explains the images:

• The originals are paper. Signed applications, handwritten notes and old personnel files exist only on paper. A scan is the only electronic form they have.
• The system prints, and the department scans. A records clerk prints the responsive pages from an account system, adds a declaration, and feeds the stack through a copier that saves pictures.
• The pages were faxed. A fax is an image by nature, and a coarse one.
• Something was blacked out on paper. Account numbers or other customers' details marked through by hand and then rescanned leave no text behind.

The practical lesson is for the next subpoena, and it is the lawyer's call: if the records exist in a database, the form can be named in the subpoena itself. For the set already on your desk, a polite request to the custodian for an electronic export costs a phone call. If the answer is no, you convert what you have. The post on making a production searchable covers the same problem for party productions, in its section on why productions arrive as unsearchable images.

Log the subpoena response as it arrived. Write the custodian's name, the date, and the page count on the box or the download. These records are documentary evidence you did not create, so keep the cover letter with the scan. A short chain of custody note lists who handled the set. When you convert the pages, keep each custodian's records together so a search stays inside that set.

Converting subpoenaed bank, phone and employment records to text and tables

Convert subpoenaed records by type. Bank statements and phone logs are tables, so export them to a spreadsheet. Employment files are mixed paper, so export a searchable PDF for reading and Word or plain text for quoting. In every case, split the delivery into sensible files first and check figures against the page image.

In Schema Law the mechanics are the same for all three: upload a PDF, PNG or JPEG of up to 20 MB in the browser, look at the parsed pages, and choose an export. The Solo plan reads up to 250 pages in one document and the Team plan up to 1,000, so a long delivery may need splitting on page count alone. The pricing page lists the limits.

Bank records

A statement set is a run of monthly tables with a summary block on top of each. The work product you want is one spreadsheet of transactions that can be sorted and totalled.

1. Split the PDF at each statement's first page, so each file is one statement period.
2. Convert each file and export Excel or CSV. Schema Law's Excel export puts each table on its own sheet, with the rest of the text on another.
3. Add columns for the account, the statement period and the source page before you combine sheets.
4. Reconcile each statement: opening balance, plus deposits, minus withdrawals, should equal the closing balance printed on the page.

A statement that does not reconcile has a misread digit, a missing row or a missing page. The guide to converting a scanned PDF to Excel covers table extraction, shifted columns and total checks in full.

Cheque images and deposit slips come interleaved with statements. Treat them as pictures to read, not tables to export, and keep them in sequence with the statement they belong to.

Phone records

Call detail records are long, narrow tables with a row per call or message and the same header on every page. Three points need care.

• Read the key first. If the carrier included a legend or explanation page, it defines the columns, the time zone of the timestamps and the codes. Do not guess at a column's meaning from its heading.
• Protect the digits. Format phone-number columns as text before pasting, so a spreadsheet does not convert a long number into scientific notation or drop a leading zero.
• Count the rows. If a page footer or summary gives a record count, compare it with the number of rows you ended up with. A shortfall points to rows merged or dropped at a page break.

For Pinecrest Wireless's 96 pages, CSV is the natural export, because a flat file of rows is what the next person will filter by number and date. Check where one page's table ends and the next begins, and remove any repeated header rows.

Employment records

A personnel file is the least uniform of the three. One folder can hold a typed offer letter, a handwritten application, a benefits form with ticked boxes, performance reviews and a payroll register.

• Typed correspondence and policies convert to text you can search and quote. Export a searchable PDF for the working set and Word when you need to lift a passage.
• Payroll registers and time records are tables. Handle them as you would a bank statement, and check period totals against the printed totals.
• Handwritten pages and filled-in forms need a person. Schema Law's own site says recognition is strongest on clean print and weaker on handwriting, and that nothing marks a passage as uncertain. Read those pages from the image and type what matters into your notes.

Whatever the type, compare names, dates and amounts with the original image before a figure goes into a declaration or a damages model. Converted text is a finding aid. The scan is the record.

Custodian declarations and business records certifications in the stack

The declaration or certification that came with subpoenaed records is part of the set, not packaging. Keep it attached to the pages it describes, number it with them, and never separate, edit or re-sign it. A lawyer may need that exact page to get the records into evidence.

In federal court the reason is in the evidence rules. Federal Rule of Evidence 902(11) treats as self-authenticating "The original or a copy of a domestic record that meets the requirements of Rule 803(6)(A)-(C), as shown by a certification of the custodian or another qualified person that complies with a federal statute or a rule prescribed by the Supreme Court." Rule 803(6)(D) in turn allows the business records conditions to be shown "by the testimony of the custodian or another qualified witness, or by a certification that complies with Rule 902(11) or (12) or with a statute permitting certification".

States have their own versions. New York's CPLR 3122-a, for one, provides for a sworn certification by the custodian or another qualified witness to accompany business records produced under a subpoena duces tecum. Whether a particular certification is sufficient is a legal question for the lawyer on the file. Your job with the paper is narrower.

Handling checklist for the certification pages

• Find every one. A custodian may put it first, last, or once per account inside a long file.
• Record in the intake log which pages each certification covers. If it states a page count or date range, write that down and compare it with what you received.
• Number the certification in the same sequence as the records, so one range identifies the whole response.
• Convert it with the rest. A searchable certification lets you find the custodian's name, title and the date signed.
• Keep the wet-ink original if one came in the box, in the original folder, and note where it is stored.
• Tell the lawyer about any deadline you see. Rule 902(11) also says that "Before the trial or hearing, the proponent must give an adverse party reasonable written notice of the intent to offer the record". That date belongs on the case calendar, and the lawyer decides what it requires.

Also keep the cover letter, any invoice for copying costs, and the envelope or transmittal email. Together they show when the records arrived, from whom, and in answer to which subpoena.

Do you have to share subpoenaed records with the other parties, and in what form?

Federal Rule 45 requires notice to the other parties before a subpoena for records is served. The text of the rule stops there on sharing subpoenaed records, and the advisory committee note says the serving party should give access to what comes back. State rules and case orders can require more, so check your own court.

The notice provision is Rule 45(a)(4): "If the subpoena commands the production of documents, electronically stored information, or tangible things or the inspection of premises before trial, then before it is served on the person to whom it is directed, a notice and a copy of the subpoena must be served on each party."

On the records themselves, the committee note to the 2013 amendment, printed with the rule, says three things:

• "Parties desiring access to information produced in response to the subpoena will need to follow up with the party serving it or the person served to obtain such access."
• "The rule does not limit the court's authority to order notice of receipt of produced materials or access to them."
• "The party serving the subpoena should in any event make reasonable provision for prompt access."

A state can put the duty in the rule. New York's CPLR 3120, in subdivision 3, has the party who issued the subpoena serve a copy on all other parties and, within five days of compliance, notify each party that the items produced are available for inspection and copying.

So the places to look, in order, are the rule that governs your subpoena, the court's local rules, the scheduling or case management order, and any ESI protocol or agreement between counsel. The lawyer decides what must go out and when. This post describes how to handle the documents once that is settled.

The form to share them in

None of the provisions quoted here prescribes a file format for passing third-party records on. If no order or agreement in your case does either, a sound default is to give the other parties what you would want to receive:

• The records as the third party delivered them, unaltered, so nobody has to take your copy on trust.
• A numbered copy, so everyone in the case cites the same page by the same label.
• The certification and cover letter, in the same range.
• A short transmittal that states the source, the date received, the number range, and anything you did to the copy, such as numbering and adding a text layer.

If you share a searchable version, say so, and say that the text was produced by OCR and has not been proofread. That keeps anyone from treating a search with no hits as proof that a name is absent.

Numbering and logging subpoenaed records before anyone works on them

Before anyone reads, highlights or converts subpoenaed records, copy the delivery to a read-only folder, write one line in an intake log, and number a working copy. Those three steps mean every later cite points to a fixed page, and you can always show what the third party sent.

Preserve

Copy the files from the disc, link or email exactly as they came, with their original names, into a folder named for the custodian and date. Do not open and resave them. For a box of paper, scan it once as a preservation copy before the sheets are handled for review, and keep the box.

Log

One row per delivery is enough at intake:

Number

Third-party pages need a label that says where they came from. Give each custodian its own prefix so a page from the credit union can never be confused with a page from a party. Stamp the working copy only, and leave the preserved original alone. The guide on how to Bates stamp a PDF covers the prefix, digit count, placement and the stamping run.

Number before you split or convert. A page count taken from the file as received, matched to the last number stamped, is your proof that no page was lost on the way.

Then work

With numbers on the pages, split the working copy into documents or statement periods, run OCR, and export what each record type needs. Schema Law states on its security section that documents stay inside the account that uploaded them and that you can delete any document whenever you choose. Check any upload against the protective order and the firm's policy first, since third-party records can carry account numbers and personal details of people who are not in the case.

The last intake step is a list of what the set contains. The post on how to index discovery documents shows how to code each document's date, author and type so the team can find a statement or a letter without paging through the box.