Gloria is made up, but most office managers will know her. She runs the back office at a 40-person property management firm in San Diego, and a client asked her for a lease addendum from 2014. The original was in a box in a storage unit. The emails about it were in the inbox of someone who left the company in 2019. She found it eventually, a day and a half later, with dust on her sleeves.
That kind of lost time adds up across a whole company. When the American Productivity & Quality Center surveyed 982 full-time knowledge workers in 2021, it found they spent about 2.8 hours a week looking for or requesting information. That’s roughly a working day every three weeks, gone to searching.
Gloria’s firm wasn’t short of information. It had too much of it, spread across boxes, spreadsheets and old software no one could log into anymore. Turning that into something people can actually use takes more work than buying a scanner, and most of it happens before anyone presses “scan.”
How the mess builds up
No one decides to make their records hard to find. It happens slowly. Accounts keeps paper invoices because they always have. Leasing uses spreadsheets, and someone in maintenance saves everything to a shared drive with folder names like “NEW stuff (2).”
Then people change jobs and systems get replaced, and the few people who knew where everything lived are gone. Gloria’s addendum is a perfect example. The paper copy and the email trail both existed, in two places no one would think to check together.
Owning documents is very different from being able to use them. A company can store thousands of neatly boxed files and still struggle if nobody can find the one they need on a Tuesday afternoon.
Scanning is the easy bit
Gloria’s first plan was to buy a fast scanner and have an intern work through the storage unit over the summer. A friend who’d tried that warned her off it.
Scanning a page gives you a picture of the page. Saved as “Scan_0047.pdf” in a folder called “Archive,” it’s barely easier to find than the paper original. The firm would have swapped dusty boxes for digital ones.
For firms with large archives, outside document digitization services can convert paper into searchable files and index them by date, document type, client or custom fields. Some, like Corodata, also scan individual records from storage on request, which helps when the archive is too big to convert all at once. Gloria still needed to work out what her firm actually wanted before she called anyone.
Deciding what to convert first
Her most useful afternoon was spent with a whiteboard. She listed which records people asked for every week, which ones the firm had to keep by law and which ones nobody had opened in a decade.
The answer was obvious once she’d written it down. Leases, tenant files and the last seven years of invoices needed scanning first. Board minutes from the 1990s could stay in their boxes, safely stored, until someone actually needed them.
Starting small saved money and gave staff something useful within weeks rather than months. It also showed her which categories of paperwork caused the most trouble.
The prep work nobody sees
Before a single page goes through a scanner, someone has to get the paper ready. Staples come out, torn pages get taped, and folders full of mixed documents have to be sorted. In Gloria’s archive, half the problem was coffee rings and faded thermal-paper receipts. One whole box had been water-damaged during a leak in 2016.
Older records tend to have their quirks. Handwritten notes in the margins, carbon copies and oddly sized forms all slow things down. It’s tedious work, but skipping it means poor scans and missing pages later, which costs more to fix.
Making the text searchable
After scanning, optical character recognition (OCR) turns the printed words in each image into text a computer can search. That lets someone search for a tenant’s name or “addendum” inside the documents themselves instead of guessing at file names.
OCR still makes mistakes. A faded “8” can come out as a “3,” and a smudged signature line can come out as nonsense. Handwriting and complicated layouts cause the most errors. For anything financial, legal or medical, someone should check the extracted text against the original before it’s relied on.
The same caution applies to newer AI extraction tools, which can pull totals or dates from invoices in seconds and sometimes do it confidently wrong. A few basic habits for checking AI-generated content go a long way here.
Labels that match how people work
Indexing is where the archive starts to feel useful. Each document gets tags such as document type, date, property address or tenant number, so people can filter for exactly what they need.
The trick is tagging things the way each team actually searches. Gloria’s accounts team wanted invoices sorted by vendor and billing month. Leasing wanted everything grouped by property and unit. HR wanted staff files filed by employee number and nothing else. Agreeing on those labels before scanning saved a painful re-sort later.
Who gets to see what
Paper has one accidental security feature: it’s a hassle to copy. Once a file is digital, it can be forwarded to the wrong person in two clicks. Gloria’s tenant files had bank details and ID copies in them, so access had to be tightly controlled from day one.
She set up permissions so each team saw only what it needed. Everything sensitive was encrypted, and the system logged who opened which file. Retention rules made sure old records were deleted on schedule rather than piling up. A proper identity governance checklist helps if you’re working out who should have access to what as the archive grows.
Where AI comes in
Most of this groundwork matters even more once a company starts using AI. Tools that sort documents, pull out key figures or answer staff questions in plain English all depend on the quality of what they’re fed. Give them blurry scans with no labels and they’ll make the mess worse, only faster.
Gloria’s firm is now testing an assistant that answers questions like “when does the lease at 14 Harbor View end?” It works well on the newer, well-indexed files. On the older ones, it sometimes gets things wrong.
Questions people usually ask
Q. Is scanning the same as digitizing?
Not quite. Scanning makes an image. Full digitization adds text recognition, indexing, quality checks and a place in a document management system.
Q. Do we have to convert everything at once?
Usually not. Start with records people use often or have to keep by law, and store the rest safely until they’re needed.
Q. Can AI sort our scanned files automatically?
It can help a lot, especially with documents that follow a consistent format. Odd or sensitive documents still need a person to check them.
Q. What keeps a digital archive useful years later?
Clear naming and good labels matter, along with sensible permissions. Someone also has to own the archive and keep it tidy, backed up and in line with retention rules.
As for the 2014 addendum, it’s now tagged by property, tenant and year, and it takes about four seconds to find. Gloria has started on the boxes from 2008.
Related: What AI Websites Know About You Before You Type a Prompt
