Almost everybody who digitises a set of papers is quietly holding three beliefs, and all three are wrong in a way that costs something later. The first is that scanning is keeping. One copy, on one phone or one laptop, is not a backup — it is a second thing that can be lost, and devices are lost far more reliably than cupboards are. The second is that a scan replaces the paper. It does not: for most of the things people digitise documents in order to do, an office or a court wants the original or a copy certified from it, and a digitisation exercise is never a reason to destroy anything. The third is the quietest and the most expensive. OCR — the step that makes a scan searchable — is not accurate. It reads handwriting badly, reads faded carbon badly, reads text under a stamp badly, and it makes its mistakes without telling anybody. A file that is searchable but wrong is in one specific way worse than a plain picture of a page, because people believe what a search returns and stop looking. Set those three aside and the real value of this exercise becomes visible, and it is not the scanning at all. It is the index. Four hundred files carrying the camera’s own numbering are not an archive; they are the same cupboard, in a different cupboard. What turns a pile of paper into something usable is a consistent naming scheme, a list of what exists, and a structure that somebody who is not you could open and navigate — which is the test that matters, because the moment this is most needed is usually a moment when you are not the one looking. This page is about doing it once, properly: what to capture and what people miss, why colour and both sides are not fussiness, which material must never go through a feeder, what to ask before handing your papers to a stranger, and what to do with the originals afterwards, which is to keep them.
Almost every disappointing digitisation exercise we see rests on one of three beliefs, and each of them is reasonable enough that nobody questions it until the day it fails.
| The belief | What actually happens |
|---|---|
| “It is scanned, so it is safe” | One copy in one place is one accident away from gone. Scanning without a second location has not reduced the risk, it has moved it. |
| “It is scanned, so I do not need the paper” | The office you eventually deal with wants the original or a certified copy. The scan gets you to the counter; it does not get you past it. |
| “It is searchable, so it is readable” | Text attached by machine reading is approximately right and silently wrong in places. Searchable is not the same as accurate. |
Those three beliefs are the whole argument of this page, and the sections that follow are mostly an elaboration of them. If you read nothing else, read the three rows above and then the section on the index, because between them they decide whether the work you are about to do is worth anything in ten years.
A scan feels protective in a way that is out of all proportion to what it actually does. The moment the file exists, the anxiety about the paper reduces, and from then on nobody thinks about the arrangement again. That is the trap, because what has been created is not a second copy of the archive — it is one more object that can be lost, sitting alongside the first one.
Consider the realistic failure cases rather than dramatic ones. A phone is replaced and the gallery is restored imperfectly. A free account is created with an email address that later stops being used, and the recovery goes to that address. A laptop is handed down to a child and reformatted. A pen drive is put in the same cupboard as the papers, so one flood takes both. None of these is unusual, and all of them are quiet — you find out years later, when you go to look for something.
The working test is simple. Pick any single device or any single account you own and imagine it gone tomorrow, with no warning. If the complete set still exists and you can reach it, you have a backup. If the answer depends on that one thing, you do not.
Notice that this has nothing to do with scanning quality, OCR, naming or anything else technical. It is purely about where things sit, and it is the cheapest part of the whole exercise to get right. It is also the part most often left for later and then never done.
This is the belief that causes real damage, because people act on it in an irreversible way. The reasoning runs: the content is captured, the content is what matters, therefore the paper is now clutter. And so papers get thrown away, or stored carelessly, or left behind in a house being cleared.
What offices and courts generally work with is the original instrument, or a copy certified from it by somebody authorised to certify. A scan you made at home is neither. It is extremely useful — it tells you what you have, it lets you send something ahead, it lets somebody check whether a document is worth pursuing before anybody travels anywhere. But when the moment comes to file, register, claim or prove, the question is about the paper.
Our certified true copy guide goes into which kind of copy satisfies which kind of office, and that is the companion question to this one; this page does not repeat it. The point to carry away here is narrower and more practical: digitisation is never a reason to destroy anything. Not a stamped receipt, not an old mutation entry, not an envelope with a postal endorsement on it, not a page you cannot read and do not understand. Especially not that last one.
It is worth being concrete, because the term gets used as though it were a quality level rather than a step.
A scan is a picture. To a computer, a scanned page is an arrangement of coloured dots with no more meaning than a photograph of a wall. You can look at it and read it; the machine cannot. Optical character recognition is the process that examines those dots, decides which shapes are letters and numbers, and attaches the resulting text to the file — usually invisibly, behind the image, so the page still looks exactly like the original but can now be searched.
That one capability is what changes the character of an archive. Without it, a hundred and fifty files are a hundred and fifty files, and finding the one that mentions a particular plot number means opening them one after another. With it, you type the number and the archive answers. For a family's papers this is the difference between an archive and a heap; for a business with ten years of records it is the difference between answering a notice in an hour and answering it in a fortnight.
Used as intended, OCR earns its keep in one specific way: it tells you which document to open. That is a large benefit and it is worth paying for. It is also the entire extent of what it should be relied on for.
Place Order — Free · pay after work
Machine reading is good at clean modern print on white paper and progressively worse at everything else. What makes this a genuine hazard rather than a minor imperfection is that it never announces the difficulty. It does not leave a gap or flag a doubt; it produces a confident wrong answer that looks exactly like a right one.
Handwriting. Largely unreliable. A handwritten endorsement, a signature block, a clerk's note in a margin — assume these are not searchable at all, and index them by hand instead.
Faded carbon and old duplicates. The purple-grey second copy of a typed document is among the hardest material there is, and much of what families keep from the nineteen-seventies and eighties is exactly that.
Text under a stamp or seal. The impression sits on top of the letters. A person reads through it without effort; the machine reads the combination and produces nonsense in that one spot, which is frequently the spot with the date on it.
Old typewriting. Uneven inking, letters striking high or low, and an l doing duty as a 1 — all of it confuses the reading step.
Regional scripts and mixed-script pages. Results vary a great deal by script and by how clean the print is, and a page with two scripts on it is harder than either alone.
Tables and columns. The characters may be read correctly and the order be wrong, so the right figure ends up against the wrong row. This one is particularly unkind because every individual value is accurate.
None of this is an argument against doing OCR. It is an argument against trusting its output for anything that matters. The rule that follows from it is the next section, and it is short.
The discipline is one sentence: search the text, read the image.
Use the searchable layer to locate the page. Then open the page and read the thing itself with your eyes before you write a date on a form, quote a number in a letter, or tell anybody what a document says. Every error we have ever seen traced back to a digitised set came from somebody skipping the second half of that sentence — usually in a hurry, usually with a figure that looked plausible.
Be especially careful where a document has been copied from a searchable file into something else. A date typed from a search result into an application, a plot number pasted from extracted text into a letter, an amount carried from a scanned ledger into a reply — these are the points at which a silent machine-reading error becomes your statement.
This is also why we do not deliver an extracted-text file as the product. We deliver images with text attached for searching, and an index written by a person. The extracted text is scaffolding, not content.
If the three beliefs are the problem, this is the answer, and it is worth stating plainly because it is not what people expect to be buying. The valuable output of digitisation is not the folder of files. It is the list.
Four hundred images with names the scanner gave them are, functionally, no better than four hundred loose papers — arguably worse, because at least a pile of paper can be riffled through. What makes a digital archive usable is knowing what is in it without opening it: a line per document saying what it is, whose it is, what date it carries, what condition it is in, and where the original sits.
That list is also the only part of the whole exercise that does not degrade. Image formats change, software is discontinued, drives fail, services shut down. A plain list of what exists survives all of that, and lets anybody rebuild or verify the rest. We treat it as the primary product and the scans as the attachment, which is the reverse of how the work is usually sold.
The instinct is to start feeding paper into a scanner, because that is visibly progress. It is the wrong order, and it is expensive in a way that is not obvious until you are two-thirds through.
Scanning is labour. Sorting is thinking. If you scan first, you will spend the same labour on a six-page insurance circular from 2004 as on a registered sale deed, and you will do your thinking afterwards, on a screen, where the physical cues that help you — the quality of the paper, the staple, the envelope it arrived in, the fact that three sheets were folded together — have all been removed.
So the sequence is: empty the cupboard, make piles, then capture. The sorting usually takes an afternoon and it almost always produces a surprise, which is that the genuinely important material is far smaller than anyone expected. People estimate a thousand pages and find that eighty of them matter. That discovery is what makes the rest of the project affordable, because eighty pages can be done properly.
Place Order — Free · pay after work
Four piles is enough. More than four and the sorting itself becomes a project.
| Pile | What goes in it | How to treat it |
|---|---|---|
| Identity and personal | Certificates of birth, marriage, death; educational records; identity documents; name-change papers | Capture properly. Small pile, high value, frequently needed at short notice. |
| Property | Deeds, allotment and possession letters, mutation entries, tax receipts, society papers, plans, loan documents | Capture properly and in order. This is the pile where a missing page costs the most. |
| Money and compliance | Bank and investment papers, insurance policies, tax records, employment and pension papers, business filings | Capture properly for anything still live; capture quickly for closed and historical material. |
| Everything else | Circulars, statements long superseded, warranty cards, correspondence, duplicates of duplicates | Capture quickly or not at all. This pile is usually three-quarters of the volume. |
Two rules make the sorting go faster. First, when you cannot decide which pile something belongs in, it goes in the pile that gets the more careful treatment — indecision is itself a signal. Second, do not throw anything out of the fourth pile during the sort. Decide about discarding later, as a separate exercise, when you are not also trying to think about three other things. Our legal file management service does this sorting with you where the volume has become daunting.
Once the sorting is done, capture is mechanical, and good capture is mostly a matter of resisting four instincts that all point the wrong way: to save space, to save time, to tidy the edges, and to skip the blank side. The next four sections deal with those in turn.
Before them, the settings that matter, stated once. Scan at a resolution that holds fine detail rather than the lowest that looks acceptable on screen — faint endorsements and small print are what you are protecting. Keep the page flat and the lighting even. Capture the whole sheet including its margins. And do not let a scanning app's automatic improvements run on legal material: the sharpening, straightening, background-whitening and shadow-removal features are designed for whiteboards and receipts, and on a stamped document they can remove the very thing that makes it evidence.
If a scanning app offers a “document” mode that makes the page look crisper and whiter than the paper in your hand, that is the mode to avoid for anything legal. Use whatever setting gives you the page as it actually is, marks and discolouration included.
Black-and-white scanning survives from an era of expensive storage and it still appears as a default. For legal paper it is the single most damaging setting available, because of what it quietly discards.
The colour of a stamp. Blue, violet, black — sometimes the only thing distinguishing an office's impression from a reproduction of one.
Printed line versus ink entry. On a form, the difference between what was printed and what somebody wrote in is often carried by colour alone. In black and white a filled form can read as though it came printed that way.
Embossing and seals. A dry seal survives as a faint tonal shift. Convert to two tones and it vanishes completely.
Corrections and additions. A date altered in a different pen, a figure overwritten, an initial against a change — all of them become invisible or indistinguishable.
Stationery and watermarks. The tint of judicial or official paper, and anything held in the paper itself, goes with it.
Colour scanning takes the same time and produces larger files, and larger files are not a problem worth solving at the cost of evidence. Where a file has to be small for a particular purpose, make a small copy for that purpose. Never make the archive itself small.
This is the commonest hole in a digitised set, and it is entirely avoidable. Scan the reverse of everything, including the sheets you are confident are blank.
The reasons are specific rather than theoretical. Endorsements are written on the back. Receipts are stamped on the back. Continuation of a schedule frequently runs onto the reverse. Postal and courier markings, which establish when something was sent, are on the back of an envelope. Older official forms often carry conditions, instructions or a declaration on the reverse that the front refers to. And a page numbered “2 of 4” tells you something is missing only if you can see the numbering, which is sometimes only on one side.
The cost of scanning a blank back is a few seconds and a small file. The cost of not having it is discovering, two years later and possibly in front of somebody, that the thing being asked about was on a side nobody captured. For a two-sided set the duplex setting on a scanner does this automatically; for phone capture it is a habit you have to build deliberately.
Place Order — Free · pay after work
Automatic edge detection is a convenience feature, and on a legal document it is a liability. It is designed to find the paper and trim away the background, which is sensible for a printed page and unhelpful when the margins carry information.
What lives in a margin: a page number, a file or receipt number written in by hand, a clerk's initials against an entry, a pin or staple hole that shows a page was once attached to something, a torn edge that explains why a line is incomplete, a punch hole pattern that tells you which file the paper came out of. Trim the margin and the page looks neater and says less.
Capture the full sheet with a little of the surround visible, and leave it uncropped. The same applies to straightening: a page slightly askew is not a defect worth correcting, and automatic straightening on a page with a border or a stamp near the edge can shave off exactly the part you most wanted.
Compression is where a well-intentioned step destroys the thing being preserved, and it is worth understanding why, because the mechanism is not intuitive.
Image compression works by deciding which parts of a picture matter less and representing them more cheaply. The parts it judges to matter less are areas of low contrast and fine detail — which is a precise description of a faint carbon impression, a light seal, small-print conditions, a signature in thin ink and the edge of a stamp. Compress hard enough and those regions turn into smooth smudges. The file still opens. The page still looks like the page. The detail is simply no longer there, and no later step can bring it back.
Keep one full-quality archive set and never touch it again. Make every small, emailable, portal-sized version as a separate copy from the archive. The error that cannot be undone is compressing the only copy you have.
Portals do impose real limits and those have to be met; that is a conversion problem, not an archiving decision. Our PDF compression and resize service produces portal-sized files from a full-quality set, and our photo and signature resize service handles the dimension-and-kilobyte rules that application forms carry. Both work from the archive and leave it as it was.
A good phone photograph is a legitimate capture method and we are not going to pretend otherwise. For a great deal of ordinary paper it is indistinguishable in usefulness from a scan, and it has the enormous advantage of actually happening — a set photographed over three evenings exists, and a set waiting for a proper scanner does not.
Where it is adequate: typed or printed pages in reasonable condition, correspondence, statements, circulars, forms you filled in yourself, anything in the fourth pile, and a quick working copy of anything at all while the proper capture is still pending.
Where it is not: anything with a seal or a dry embossing, a stamp impression that may have to be examined, a faint or faded original, a signature whose appearance could matter, a page with very small print in the conditions, and bound material where the page curves away at the spine.
If you are photographing, four things make most of the difference. Put the page on a flat surface in daylight away from direct sun. Hold the phone square above the centre rather than at an angle. Keep your own shadow and the phone's shadow off the page. And turn off the app's automatic document-enhancement before you start, not after.
A practical middle path works well for most households: photograph everything quickly to establish what exists, then pick out the thirty or forty pages that deserve proper capture and have those done properly. Our scanning service handles that second set, and our document pickup and delivery service moves the paper if you would rather not carry it.
Fastened paper raises a question that has nothing to do with image quality: whether to take it apart.
The default should be no. A staple is information. It tells you that these four sheets were filed as one document, in that order, by somebody who treated them as belonging together. Separate them, scan them individually, and you have four pages whose relationship now exists only in whatever you remember about it. Where the fastening has to be removed for capture, note it in the index — that these pages were fastened together as received — and keep them together afterwards.
Practically: for a stapled set of a few pages, open the staple carefully if it is a simple one, capture, and re-fasten through the existing holes rather than making new ones. For anything stitched, ribboned, sealed or bound in the way registered instruments often are, do not take it apart at all. Capture it as it is, accepting that the inner margin may curve, and photograph the fastening itself so the record shows how the document was held together.
Place Order — Free · pay after work
Old paper behaves differently and deserves a different approach. Paper that has gone brittle cracks along fold lines rather than flattening, and the pressure of a sheet-feeding scanner can tear it outright.
So: nothing old goes through a feeder. Flatbed with the lid resting rather than pressed, or a photograph from above, and no attempt to force a fold open. Where a document has been folded in the same place for forty years, the fold is now part of its physical state, and flattening it is a conservation decision you probably do not want to make on a kitchen table. Capture it folded if that is how it is, and capture the folded edges separately if text runs across them.
Two specific hazards. Do not use adhesive tape to repair a tear before scanning — it discolours, it is difficult to reverse, and it can take the surface of the paper with it later. And do not attempt to clean a stained or foxed document; the stain is cosmetic and the attempt frequently is not.
Where paper is in genuinely poor condition, the honest sequence is to capture it now, at whatever quality is achievable without handling it more, and then treat the question of a replacement as separate. For many official records a fresh copy can be obtained from the issuing office, and our duplicate certificate service deals with that route. A crumbling original and a clean fresh copy are both worth having, for different reasons.
Site plans, sanctioned drawings, survey maps, large family trees and genealogy charts prepared for a claim: all of them are larger than the scanner and all of them are frequently the document somebody will most want to look at closely.
Three approaches, in descending order of preference. Have it captured whole on a large-format scanner, which gives one accurate sheet and is worth doing for a sanctioned plan. Photograph it whole from directly above at a sensible height, which is adequate for identification and orientation but loses the fine dimensions. Or capture it in overlapping sections, which preserves detail but produces a set of pieces whose relationship someone has to reconstruct.
If you use the third approach, do two things that cost almost nothing and save a great deal. Photograph the whole sheet first as an overview, so the pieces have something to be fitted against. And make sure each section overlaps its neighbours generously, including a recognisable feature in the overlap, rather than meeting edge to edge. Note in the index that the document is held in sections and how many there are, because a reader who finds piece four of seven and does not know there are seven will assume they have the document.
This is where an archive becomes usable or does not, and it needs about ten minutes of decision followed by complete consistency.
A scheme that works for nearly everyone: the date first, written year then month then day, followed by what the document is, followed by whose it is. So a birth certificate becomes something on the pattern of the year, the month, the day, then the words identifying the document, then the name. The virtue of putting the date first in that order is mechanical: any folder listed alphabetically comes out in date order automatically, in every piece of software, for ever, with no sorting feature required.
Use the date on the document, not the date you scanned it. If the document carries no date, use the year it clearly belongs to and say so in the index.
Keep names plain. Letters, numbers, hyphens. Avoid slashes, colons and quotation marks, which various systems refuse, and avoid very long names, which get truncated in ways that hide the useful part.
Spell out what the document is rather than abbreviating it. You will remember what your abbreviations meant; the person going through this after you will not.
Add a sequence for multi-page captures of one document, numbered with a leading zero so that page ten does not sort before page two.
Never renumber. Once a name is in the index, it stays. If something was named wrongly, note the correction in the index rather than rewriting history.
The scheme itself matters far less than using one. Any consistent arrangement beats a thoughtful arrangement applied to two-thirds of the set.
Shallow. The temptation is to build an elaborate tree, and the elaborate tree is where documents go to disappear.
Two levels is usually enough: a top level that mirrors the four piles, and inside each one a level by person, by property, or by year, whichever the material naturally wants. Deeper than that and you start spending your time deciding which of two equally reasonable folders something belongs in — and whichever you choose, the next person will look in the other one.
The deeper reason to keep it shallow is that the structure is not what you will use to find things. Searching is what you will use. Folders exist so that a human being opening this for the first time can see the shape of what is here; they are a map, not a filing system. A map with forty levels is not a better map.
One addition that earns its place: a single plainly named folder at the top containing the index and a short note about the whole set. It is the first thing anybody opens, and if it is the only thing they understand, they can still get to everything else.
Place Order — Free · pay after work
The index is a list with one line per document, and the test of it is not whether you can use it. It is whether somebody else could, because the occasions on which a family archive matters most are very often occasions when the person who built it is not available to explain it.
Six columns cover almost everything.
| Column | Why it is there |
|---|---|
| What the document is | In plain words, as a stranger would need it described |
| Whose it is | The person or property it concerns, named consistently throughout |
| Date on it | From the document itself; blank and noted if there is none |
| Where the original is | The actual physical location — which cupboard, which folder, which bank |
| File name | So the line and the image can be matched in either direction |
| Notes | Condition, pages, whether it is an original or a copy, what is unreadable, what is missing |
The notes column is the one that repays effort. “Second page torn, text at the bottom right not readable” is worth more than the scan of that page, because it tells a reader that nothing has been lost since — the damage was already there. “Photocopy only, original not found” is worth even more, because it stops somebody searching for an original that nobody has seen in twenty years. Record gaps as deliberately as you record contents.
An index that lives only in one place is the same failure as a scan that lives only in one place, with the additional problem that losing the index makes everything else much harder to use.
So: keep the index in an ordinary format that any computer can open and that does not need the software it was made in. Keep a copy of it wherever the files are kept, in each location, rather than in one central place. Print it — actually print it, on paper — and keep the printout with the original documents, because a printed list in the same cupboard as the papers is the version that will be found by whoever opens that cupboard.
The one-line test for the whole exercise: if you were unavailable and somebody who had never seen these papers had to find a particular document, is there a single sheet of paper in the house that would tell them where to start? If yes, the archive works. If no, it is a private system, and private systems end when the person does.
Where the papers are part of a family's planning rather than just its storage, this connects directly to the arrangements people make deliberately; our will drafting guide deals with that side, and an index is a practical companion to it rather than a substitute.
The threat to a digital archive is not usually a dramatic failure. It is twenty years passing. Applications are discontinued, companies close, a file type everybody used becomes a file type nothing opens without effort, and an account you created with an email address you no longer have is effectively gone.
The defence is boring formats. For scanned pages that means PDF, which opens everywhere and will keep opening because far too much of the world depends on it. For the index it means a plain spreadsheet or plain text rather than something tied to a particular service. For anything born digital, keep the original and also keep a PDF of it.
Avoid app-specific archives. A scanning app's own bundle format, a note-taking app's export, a cloud service's proprietary container — all of them read perfectly today and are a problem the day the app stops being maintained.
Avoid anything that needs an account to open. If the file requires signing in to a service to be read, the service is now part of your archive.
Avoid passwords on archive files unless you have a genuine reason, and if you use one, the index note has to say where the password is recorded. Encrypted files whose password died with their owner are a recurring and entirely preventable loss.
Keep it unexciting. There is no advantage in a clever format. There is a substantial advantage in a format that a stranger with an ordinary computer can open without asking anybody anything.
Where the output has to be presentable as well as durable — a bundle going to an office, a set being filed, pages that need numbering and a cover — our PDF legal formatting service handles that, and that is a separate job from archiving.
Three copies, in two different kinds of place, with at least one of them physically away from the others. That is the whole rule, and everything else is detail about how you implement it.
What “two different kinds” means in practice: not three cloud accounts, and not three drives in one drawer. One copy on a computer or drive you control, one copy in an online account, and one copy somewhere else entirely — a drive kept at a relative's house, a second drive in a locker — covers the realistic failures, which are a device dying, an account becoming inaccessible, and something happening to the building.
Whichever online account you use, the thing to check is not storage or features. It is recovery: whose email address is on it, whether that address is still used, whether the phone number on it is still yours, and whether anybody else in the family could get into it if they had to. An account nobody else can reach is a cupboard with one key.
And check the copies occasionally. Once a year, open three or four files at random from each location and confirm that they open. A drive that failed silently four years ago and a sync that stopped working after a phone change both look exactly like a working backup from the outside.
Place Order — Free · pay after work
This is a reasonable anxiety and it deserves a direct answer rather than reassurance. When you hand over a cupboard of papers for digitisation, you are handing over, in one bundle, the material somebody would need to impersonate you or to understand your affairs completely.
The commonest route — the photocopy shop on the corner — is also the least considered. Scans sit on a shared machine's disk until that machine is replaced, often with no deliberate deletion ever taking place, and the person operating it has no obligation to you beyond the job.
So whoever does this work for you, including us, ask four questions before the paper changes hands: where the files sit while the work is being done, who else can see them, what happens to the working copies afterwards, and whether you get an answer in writing. A provider who treats those as awkward questions has told you something useful.
Our own answer is in the next section, and our home visit documentation service exists partly for this reason: where a set is sensitive or simply too large and precious to move, the work can be done where the papers are, and the paper never leaves the house.
Stated plainly, because a vague answer here is worth nothing.
The paper comes back — all of it, in the order it arrived, with the fastenings as they were, against the same list we checked it in on.
You get the complete set on whatever you prefer to receive it on, together with the index, before anything on our side is removed.
Working copies are deleted once you have confirmed you have the set and can open it. Not before, because a delivery that fails after deletion is a disaster, and not indefinitely after, because keeping them is a risk we are holding on your behalf for no benefit to you.
We do not keep a reference copy of a client's documents as a matter of course, and we do not keep the index either. If you want us to hold the set so that a future piece of work can refer to it, that is a decision you make explicitly and we confirm in writing.
Nothing is shared with anybody, for any purpose, including anything described as verification, improvement or convenience.
The one thing we will keep is a record that the job happened — what was received, how many pages, what was returned and when — because that record protects both sides if anybody later asks what was handed over. It contains no document contents.
Once a set is captured, the paper needs a decision, and the decision is almost never “discard”. What it should be is “sorted by how often it is needed and how hard it would be to replace”.
| Kind of paper | Where it belongs |
|---|---|
| Irreplaceable and rarely needed — registered instruments, old deeds, original certificates of a person no longer living | The most secure place you have, touched as seldom as possible. Record the location in the index. |
| Replaceable but needed often — current identity and address documents | Accessible. These are the ones to carry a certified copy of rather than the original. |
| Replaceable and rarely needed — old statements, closed policies, superseded records | Boxed and labelled with what the box contains and which years. Scanned is enough for working purposes. |
| Genuinely redundant — duplicates of duplicates, circulars, expired warranty papers | The only pile worth discarding, and only after the set is captured and verified. |
One habit worth adopting before any of this: stop carrying originals. The reason a surprising number of irreplaceable documents are lost is that somebody took the original to an office because that is what was asked for, and then the original travelled home in a bag. A certified copy satisfies most requests, and where it does not, the original goes and comes back the same day with somebody who knows what they are carrying.
A digitised set produces something that looks official and is not, and the distinction is worth holding firmly, because it is the point at which a well-organised archive can mislead its owner.
What you now hold is a photographic record of your own paper, made by you or on your instructions. Nobody has attested that it corresponds to anything. It is excellent for knowing, sending, checking and deciding. It is not an attested copy, it is not a certified copy, and printing it does not make it either.
So: a scan is not a substitute for a certified copy, a notarised copy, an attested copy, a true copy issued by a record office, or an original. When an office or a court asks for one of those, the archive tells you what exists and where it is. Getting the thing the office wants is a separate step.
Our certified true copy guide sets out which kind of copy satisfies which kind of requirement and who can issue it, and our certified true copy service obtains them. Where what you actually need is a fresh issue from the office that made the record in the first place, our duplicate certificate service goes that route instead, which is often faster and always cleaner than certifying an old photocopy.
Place Order — Free · pay after work
The commonest real version of this job is not an orderly archiving project. It is a cupboard, a trunk or a steel almirah belonging to somebody who has died, being opened by people who do not know what is in it and are not sure what matters.
The approach that works is deliberately unhurried. Do not sort by importance first, because you cannot yet tell what is important. Sort by subject: everything about one property together, everything about one bank together, everything about one person together. Subject sorting can be done without understanding the documents, and the understanding tends to arrive as the groups form.
Then capture everything in the first three categories, including every page you cannot read and every paper whose purpose is a mystery. The unreadable half-page with an office stamp on it is precisely the item that turns out to matter, and it is also precisely the item somebody throws away during a clear-out because it looks like nothing.
Keep a short note, as you go, of anything that looks like it should exist and does not — a deed referred to in a tax receipt but not present, a policy mentioned in a letter with no policy document, a bank passbook for an account nobody has mentioned. That list of absences is frequently more useful than the set itself.
Where a succession, a transfer or a claim is already under way, digitisation stops being an archiving exercise and becomes a working tool, and the priority order changes completely.
What offices, banks and registries typically want to see established is a chain: who held the thing, what document says so, who the people entitled now are, and what connects them. So capture along that chain first, in order, rather than capturing the easy material first. Everything proving the holding, then everything proving the relationship, then everything proving the absence of competing claims, then the rest.
The second change is that the index becomes the live document of the process. Add two columns to it: which office or institution has asked for each item, and what was sent to them and when. A claim involving four institutions generates four slightly different demands, and the thing that goes wrong is not the documents — it is losing track of which version of which set went where. Our legal heir certificate guide covers what that chain usually has to establish, and our legal file management service keeps the running record while the process is open.
For a business the question is not sentimental and it has a sharper edge: when a notice arrives, it asks about a period, and whoever answers it has to produce that period's records quickly enough to matter.
Which means the organising principle for business papers is different from a family's. It is by year and then by kind, not by subject, because demands arrive framed as years. Within a year, keep the categories that a demand will use: statutory filings, tax, employment and payroll, contracts and their amendments, correspondence with an authority, and licences with their renewals.
Capture the filing acknowledgements, not just the filings. The question is usually whether something was filed and when, and the acknowledgement is what answers it.
Capture every amendment with its parent. A contract whose two addenda are in a different folder is, in practice, a contract nobody can state the terms of.
Capture licences together with every renewal. A current licence alone does not show continuity, and continuity is frequently the thing in question.
Keep a one-page list of what exists for each year, including years where the answer is “almost nothing”. Knowing a year is thin is better than discovering it under a deadline.
Our annual filing guide deals with the compliance calendar that generates much of this paper, and the two exercises fit together: a company that knows what it filed, and can find it, answers in a morning what otherwise takes a fortnight.
Often the paper has already been scanned — by a relative, a previous owner, a departed office manager, a shop — and what you have inherited is a drive or a folder. That is a mixed inheritance and it deserves scepticism rather than relief.
The failures are predictable. Files named by the scanner, in sequence, with no indication of contents. Only the front of each page. Black-and-white throughout. Everything compressed to make it fit on something. One folder containing four unrelated matters. Pages of one document split across two folders because of when they happened to be scanned. And no list, so there is no way to know what is missing.
None of that makes the set worthless. It makes it unverified, which is a different thing. The next section is how to find out which you have.
Place Order — Free · pay after work
Six checks, in this order, and they can be done in an evening for most sets.
| Check | What you are looking for |
|---|---|
| Does everything open | Pick twenty files at random across folders. A set with a corrupt region usually has it in one place. |
| Front and back | Count pages against the paper for one multi-page document. If only fronts were captured, it will show immediately. |
| Colour or not | Open anything with a stamp. If the stamp is black, the set was captured in black and white and the seals are gone. |
| Detail at full zoom | Zoom into the smallest print on a page. If it smears rather than sharpening, the set was compressed. |
| Completeness | Take three documents you know exist in the paper and try to find them in the set. Three misses out of three means the set is partial. |
| Is anything searchable | Search for a word you know is on a page you have open. If nothing is found, no OCR was done — which is fixable without rescanning. |
The useful outcome of these checks is rarely “start again”. Usually it is a short list: these forty pages need recapturing because only the front was done, these ten because the stamp is unreadable, the whole set needs OCR and an index. That is a much smaller job than the original one, and knowing which forty pages is the entire value of the checking.
It is worth being clear about what the government's document locker is and is not, because it sits next to this subject and gets confused with it.
Its value is that documents issued into it by an authority arrive as that authority's own electronic record, and an office that accepts them accepts them on that footing. That is a different and stronger thing than a scan you made, and where a document is available that way it is usually the better route.
What it does not do is turn your cupboard into an archive. Papers you upload yourself are your own uploads, with whatever standing your own copy has and no more. The range of documents issued into it is growing but is nowhere near a lifetime of family and business paper, and acceptance in practice still varies between offices. Our DigiLocker assistance service sets it up and pulls in what is available, and our digital document verification service checks whether an electronic document somebody has given you is what it claims to be. Neither replaces the exercise on this page; they sit alongside it.
An increasing share of what matters was never a sheet of paper: an electronically signed agreement, a tax acknowledgement downloaded as a file, an electronic challan, a registry extract, a bank statement that only ever existed as a download, an email that is itself the record of a decision.
These need a different kind of care, and the mistake is to treat them more casually because they look tidy. Three rules.
Keep the original file as it was received, untouched. Printing it and scanning the print destroys exactly what gives it standing — the embedded signature or the electronic record behind it. A printed copy is a convenience copy.
Keep the covering message with it. The email or portal receipt that delivered the file establishes when and from whom it came, and it is routinely discarded.
Put them in the same archive. An archive split between “the scanned papers” and “the downloads folder” is two archives, and the index exists precisely so that there is one.
Our electronic signature guide covers what gives an electronically signed document its standing, and that is the question to settle before you decide how casually to treat the file.
Almost every application portal imposes limits, and they are frequently unreasonable: a few hundred kilobytes for a multi-page document, fixed pixel dimensions for a photograph, a signature within a narrow band of sizes, and a total upload cap that was set years ago.
This is where people damage their archives, because the quickest way to satisfy the limit is to compress the file you have. Do not. The limit belongs to one submission on one day; the archive is supposed to last decades.
The discipline: archive at full quality, derive at whatever the portal wants, and never write the derived version back over the original. Name the small copies in a way that says what they are for, and treat them as disposable, because they are.
Our PDF compression and resize service makes portal copies to a stated limit, and our photo and signature resize service handles the dimension and file-size rules that forms carry. Both take the archive as input and leave it exactly as it was.
Place Order — Free · pay after work
Old family and property papers are frequently not in the language the current process runs in, and the digitised set makes this visible for the first time rather than solving it.
Two separate problems. The first is reading: machine recognition of regional scripts, and especially of older printing and handwriting in them, is weaker than for clean modern English print, so expect such pages to be effectively unsearchable and index them by hand with a plain description of what each one is. The second is standing: where a process needs the document in another language, what it needs is a translation that the office accepts, not a translation for your own understanding.
Keep the two apart in the archive. The original stays the original, in its own language, uncropped and uncorrected. A translation sits beside it as a separate item with its own line in the index, noting who made it and whether it was certified. Our certified translation service produces the kind that offices accept, and the index should say clearly which of your items are translations, because a translation mistaken for an original is a problem that surfaces at the worst moment.
Worth setting out, because all four are things people quietly hope it will do.
| It cannot | Why |
|---|---|
| Make a weak document strong | A scan records what the paper says. If the paper is incomplete, unstamped or unregistered, a perfect image of it is a perfect image of that. |
| Replace an original or a certified copy | What offices work with is the instrument or a copy attested by somebody authorised. Your own capture is neither. |
| Recover what is no longer on the paper | Faded, torn away, obscured by a stain — if the eye cannot find it on the sheet, no capture or processing will produce it. |
| Tell you what you are missing | Only reading the set can do that. The index does it if somebody writes it carefully; the scanner never does. |
And a fifth, which belongs here even though it is about people rather than paper: it cannot make somebody else understand your affairs. That is the index's job, and the index is written by a person who decided to be clear.
A fair description of the last few years, so you can place yourself.
A cupboard and no idea. Papers accumulated over decades, nobody certain what is in there, usually prompted by a death, a sale or a move.
Four hundred photographs on a phone. Captured in a hurry during something urgent, never named, never sorted, and now unusable for the very purpose they were taken for.
A drive from somebody else. Scanned by a relative or a previous manager, unverified, and now the only version anybody can find.
A live process and scattered papers. A claim or a transfer already running, with each institution asking for a slightly different set.
Ten years of business files and a notice. A period named in a demand, and records that exist but cannot be produced in time.
A careful person doing it properly. The least common and the easiest — usually somebody who has just watched a family go through the first case on this list.
Before any paper moves, the quickest useful thing is a rough picture of the volume and the condition. That is a ten-minute job.
A photograph of the whole lot as it sits — the open cupboard, the trunk, the stack of files. It tells us the volume better than a number would.
Your rough count of how many pages, to the nearest hundred. Nobody expects accuracy.
Photographs of three or four typical pages, including the worst-condition one you can find. The worst page sets the method for the set.
What any of this is for, if there is a purpose — a claim, a sale, a notice, a process with a date on it. Purpose changes the order of work.
Anything oversized or bound, mentioned specifically, because those need arranging rather than scheduling.
Whether the paper can travel, or whether the work has to happen where it sits.
Nothing has to be sorted first and nothing has to be explained. Sorting is part of the work, and a set arriving in no order at all is the normal case rather than an awkward one.
Place Order — Free · pay after work
In order, so you know what is happening to your papers at each stage.
| Stage | What happens |
|---|---|
| Check in | The paper is counted and listed as received, before anything else, and you get that list. It is what the return is checked against. |
| Sort | Into the four groups, with you if you want to be there. Anything doubtful goes into the more careful group. |
| Capture | Colour, both sides, uncropped, full quality, nothing enhanced. Fragile and oversized material handled separately. |
| OCR | Applied across the set so it is searchable, with no claim made about its accuracy. |
| Name and structure | One scheme across everything, two folder levels, nothing renamed afterwards. |
| Index | Written by hand, one line per document, with the notes and the gaps. This is the part that takes the time. |
| Return and verify | Paper back in the order it came, against the check-in list; set delivered; you confirm it opens before our working copies go. |
Two things we will tell you during the work rather than at the end: if a document is in worse condition than it looked and should be replaced from the issuing office, and if something referred to in the papers is plainly absent. Both are more useful early.
A good deal of what surfaces during a digitisation job is not about scanning at all. Somewhere in the fourth hour, somebody holds up a page and asks whether a property was ever transferred, whether a document still has any effect, whether an old policy can be claimed on, or what to do about a name that is spelt three different ways across four certificates.
Those are not questions about image quality and we do not answer them as though they were. What we can do is make them answerable: put the relevant papers in order, show what exists and what does not, and hand over a set somebody competent can read in an hour instead of a week. That is genuinely most of the difficulty in such questions.
Where the answer needs somebody entitled to give it, we point you to the right kind of professional instead of guessing. Our advocate directory lists verified advocates by city and by the kind of work they do, so the choice stays yours and the conversation starts with somebody who has already handled the sort of matter you are describing.
Court work is for your advocate, whose fee is engaged and paid by you directly; we do not quote, collect or share it.
Certify anything. We are not an authority and cannot attest that a copy corresponds to an original. Where you need that, we obtain it from somebody who can.
Tell you to destroy paper. No set of ours comes with that advice, and if somebody doing this work for you offers it, that is a reason to be careful.
Correct a document. Not a spelling, not a date, not a figure, not a mark we think is an error. The archive records the paper as it is; anything that needs correcting is corrected by the office that issued it.
Improve an image into something it is not. No whitening, no reconstructing a faded line, no removing a stain. Enhancement on legal material is alteration with a friendlier name.
Claim OCR is accurate. We deliver it as a search tool and say so in writing, because a set sold as reliable text is a set somebody will quote from.
Hold your set as a matter of course. Working copies go once you have confirmed delivery, unless you ask us in writing to keep them.
Backdate anything. No file, no index entry, no note. The dates are the dates on the paper and the dates the work was done.
This is priced by the set rather than by the document, which makes it the one page on our site where we genuinely cannot name your figure in advance. It opens at ₹999, a straightforward set runs to 1 – 3 days, and the complete figure reaches you after we have seen the volume and the condition — not before, because a quote given over the phone for a cupboard nobody has opened is a guess with a number attached. Nothing is collected until you have accepted that figure.
What moves the figure is page count first and condition second. Four hundred clean typed pages are a smaller job than eighty pages of brittle, folded, stamped and handwritten material, and the second of those is the one families usually have. Oversized sheets, bound volumes and anything that has to be captured where it sits are arranged separately and told to you separately.
Other people's charges stay other people's. Whatever a record office charges for a fresh copy, whatever a notary or a certifying authority charges, and whatever a translator charges reach you as their own figures — we do not mark them up and we do not fold them into ours.
And the part that costs nothing, which is where most of this page's value sits: sort before you scan rather than after, scan in colour and scan the back of every page even when it looks blank, leave the archive uncompressed and make small copies separately when a portal demands them, pick one naming scheme in ten minutes and then never deviate from it, write the index for a stranger rather than for yourself, print that index and keep the printout in the cupboard with the papers, put the set in three places with one of them away from your house, and keep every sheet of paper you just scanned.
One subject that revisits old years relentlessly is tax, where a notice about a year five back needs the return, the statements and the proofs from that year. Our income tax notice guide sets out what to keep per year and why a reply that has to be reconstructed from memory is the weak kind.
The hardest case for producing an old invoice is a business that has already closed, because assessment for the registered periods stays possible long afterwards. Why those records have to outlive the business is explained in GST cancellation — it is not an exit, it is a reckoning.
A food safety inspection is mostly a request for records — the water report, the medical certificates, the pest control visits, the purchase traceability. What has to be producible is listed in FSSAI registration — you do not choose the tier.
The reason to make the set is not tidiness — it is that once a copy exists, no handover is a disaster. Where documents actually get lost, and the one line that makes a loss recoverable, is set out in document pickup and delivery — the handover is the risk.
Place Order — Free · pay after work
Scanning and indexing settle how a set is stored and found. They do not settle whether what is in it is still true: a perfectly digitised folder of superseded documents is a perfectly digitised folder of superseded documents. Our guide to keeping a file current takes it from there — sorting every item by what day would make it false, and separating history without destroying any of it.
A scan is not a backup and a searchable file is not the document — and almost every disappointing digitised set we are handed went wrong at one of those two points. We sort a cupboard into what deserves care and what does not, capture in colour and on both sides and uncropped at full quality, apply recognition so the set is searchable while telling you plainly where machine reading cannot be trusted, name everything under one scheme that sorts itself by date, and write the index by hand — what exists, where each original physically sits, what is damaged and what is missing — because the index is what somebody who has never seen these papers will actually use. The paper comes back in the order it arrived, checked against the list we made when it came in, and our working copies are deleted once you have confirmed the set opens. Send a photograph of the cupboard as it stands and your rough page count; looking at it and telling you what the job really is costs nothing, and no money is taken until you have the full figure in front of you.
Two doors, both free. Clients search a factual directory of enrolled advocates. Advocates apply to be listed on it — no fee, no commission, nothing paid in either direction.
Search Bar Council enrolled advocates by what your matter is about, by court, or by city. Searching and sending a request are both free.
Enrolled advocates anywhere in India can apply to be listed. Your entry is published only after we verify your enrolment number with your State Bar Council.
Would rather not pick anybody? Tell us the matter instead — one form goes to every advocate at once, and the first to take it up calls you.
This directory carries no ratings, no reviews, no rankings and no fees — only the factual particulars the Bar Council of India permits, published at each advocate's own request. Browse the network · Terms for Advocates