Native vs TIFF production: which file types break when imaged

Native production wins wherever a file’s meaning depends on something that is not visible on the printed page, and loses wherever a stable, endorsable page is what the parties need. That is why almost no production is entirely one or the other. Of the fifteen file-type categories in our format risk planner, only two image cleanly; five lose content when imaged and eight need special handling or cannot be imaged at all.

When should documents be produced native rather than as TIFF?

When imaging would remove something a reviewer needs.

A TIFF production is a sequence of page images. Anything that is not rendered on the face of a page does not survive the conversion: a formula, a hidden row, a speaker note, an embedded object, a timestamp that only exists as a field. For a Word document or an email, nothing important lives off the page, so imaging costs nothing but conversion. For a spreadsheet, most of what makes it a spreadsheet is off the page.

Native production inverts the trade. The file arrives as it was created, complete, and opens in the application that made it. What it gives up is the stable page: there is no fixed pagination to cite, no surface to burn a redaction into, and no place to put a Bates number. Those are recovered with a slip sheet, a single generated page that holds the number and any legend while the native file travels alongside it.

So the question is never native or TIFF for a whole production. It is which file types go which way, and that is a question with a mostly settled answer.

Which file types break when they are imaged?

File-type categories by imaging risk, with the recommended form of production

Category	Imaging risk	Recommended form
Word documents and text PDFs	Images cleanly	Image with load file and extracted text
Email and attachments	Images cleanly	Image with load file, or near-native EML or MSG
Spreadsheets	Loses content	Native, with a Bates-numbered placeholder image
Presentations	Loses content	Native, with a Bates-numbered placeholder image
Files with macros or embedded objects	Loses content	Native, with a placeholder image
Scanned documents	Needs OCR	Image with OCR text
Photographs and screenshots	Needs OCR	Native or image, with OCR where text is present
Chat platforms	Special handling	Native or near-native, with unitization agreed in advance
Mobile messaging	Special handling	Native or near-native with sender, recipient and timestamp
CAD and engineering drawings	Special handling	Native, requiring specific discussion between the parties
Audio and video	Cannot be imaged	Native, with transcription if required
Databases and structured data	Cannot be imaged	An agreed report or export, not the underlying database
Password-protected or encrypted files	Exception-prone	Remediation before processing
Container files	Critical risk	Expanded, with exception reporting verified
Legacy or proprietary formats	Exception-prone	Native application or a conversion vendor

Categories and recommended forms are the dataset behind the format risk planner, which will tick through your own collection and total the three risk bands.

Two things are worth reading off that table. The first is how small the clean-imaging set is: two categories of fifteen. The second is that the failures are not all the same kind. A spreadsheet imaged loses content silently, and a reviewer cannot tell from the page that anything is missing. A database cannot be imaged at all, which at least announces itself.

What happens to a spreadsheet when it is imaged?

It stops being a spreadsheet and becomes a picture of one screen’s worth of it.

Formulas are the obvious loss. A cell reading 4,812,000 gives no indication of whether it was typed, summed from a column, or pulled from a sheet that was not produced. Hidden rows, hidden columns and hidden worksheets do not print, so they simply vanish, and nothing on the resulting page shows that they existed. Column widths decide pagination, so a wide sheet breaks across pages in whatever way the conversion tool chose, splitting a row from its headers.

This is the specific reason hybrid production exists, and the reason it has become the common default rather than an exception. It is also why the recommended form for a spreadsheet in the table above is native with a placeholder rather than native alone: the placeholder keeps the Bates sequence unbroken and gives a brief something to cite.

Presentations fail the same way with speaker notes, and files carrying macros or embedded objects fail it with the embedded object.

What must a hybrid production format settle?

A hybrid format is not a single decision. It is four, and a protocol that leaves any of them open tends to produce an argument later.

Which types go native. The list has to be explicit. “Spreadsheets and presentations” is a category; a protocol that says only “documents that do not image well” leaves the producing party to decide, which is the disagreement it was meant to prevent.

What accompanies a native file. A native alone has no number. The convention is a Bates-numbered slip sheet in the image stream and a link field in the load file pointing at the native, and both halves need stating.

How redactions are handled on the native types. This is where hybrid protocols most often fall apart, because a redacted spreadsheet cannot be produced natively without altering the file. The parties have to choose in advance: produce the redacted item as an image and accept the loss on that item, or agree a different treatment.

Where document boundaries fall for the types that have none. Chat and mobile messaging have no inherent document shape, so the unitization rule has to be agreed: a message, a conversation, a day, or a fixed window. The choice changes the document count by orders of magnitude from identical data.

This guide does not publish clause language for any of those, because clause language belongs in the protocol itself, drafted for the matter and checked against the court’s own requirements. A clause that looks liftable invites being pasted into a stipulation without review. What the four points above give you is the checklist to hold a draft against.

What does imaging add to an eDiscovery bill?

A separate line, charged on a separate basis from everything around it.

Format conversion is billed per production rather than per ingestion, so it lands each time a volume goes out, and it is frequently charged separately from both processing and export. The Fee Field Guide records wide variance in that line, and conversion is routinely billed at a multiple of processing.

The structural complaint is that conversion cost has no relationship to review effort. It is metered by bytes for work measured in documents, and those scale very differently: a gigabyte of scanned images is roughly 15,500 pages, while a gigabyte of plain text is closer to 678,000. Producing natively where the format allows it removes the line entirely for those file types, which is a cost argument that happens to point the same way as the evidentiary one.

How this works on a flat rate

Kavarn, the flat-rate eDiscovery service Konvergence Inc operates, includes productions and PDF conversion in the instance rate rather than metering them. Native and PDF production with metadata and load-file generation are part of the platform license, so the format decision is made on what the matter needs rather than on which option carries a per-gigabyte charge. TIFF is accepted on ingestion; TIFF output is not among the platform’s listed production formats, so raise it on a walkthrough if your protocol requires it.

That does not change any of the reasoning above. A spreadsheet still loses its formulas when imaged, and the protocol still has to settle the four points. What changes is that choosing the right form costs nothing extra, and re-running a corrected volume costs the time it takes. The pricing page publishes the rate. See what the platform includes, or book a walkthrough.

Sources

• File-type categories, risk bands and recommended forms: formats.json, the dataset behind the format risk planner.
• Format conversion billing: the Fee Field Guide dataset.
• Fed. R. Civ. P. 34(b)(2)(E) on the form of production, quoted in the ESI protocol guide.
• Pages per gigabyte by file type: volume.json, as used by the volume translator.