Should I organize by set or by value?

Both, at different levels. Split by value first: high-end goes in the case, everything else goes in boxes. Then organize within each of those by set, then by card number. Value decides which container a card lives in; set decides where it sits inside that container.

How detailed should a location be?

Detailed enough to find the card in one trip, and no more. "B4" is usually enough if B4 is sorted by set and number. Recording a row and slot position is precision you will not maintain, and a location that is wrong is worse than one that is vague.

Do I need to fix my whole backlog before this pays off?

No. Fix the cards that carry your inventory value and the containers you open every show, then clean up the rest as cards pass through your hands. A system where the expensive half is correct is far more useful than a full cleanup you never finish.

What about multiple printings of the same card in one set?

Treat each printing as its own entry, identified by the printed number and the variant written your standard way. Holo, reverse holo and non-holo copies can differ in value by a wide margin, so collapsing them into one row makes your inventory total guesswork.

Naming and Organizing Stock So You Can Find It

A search that returns forty results is the same as no search at all. The fix is not a better app, it is four fields filled the same way every time.

Here is the moment that decides whether your inventory system is real. A buyer puts a card on the table and asks what you want for it. You type a name into your phone. If one row comes back, you quote a number and the sale moves. If forty rows come back, you are reading through printings and conditions while the person in front of you decides to just check TCGplayer instead.

Most dealers blame the tool. It is almost never the tool. It is that the same card got entered three different ways over six months, and none of those entries say which box it is sitting in.

The four fields that do all the work

You can track a dozen things about a card. Four of them decide whether you can find it: set, number, condition, and location. Everything else is nice to have.

Set separates the thirty Charizards you might own into the one the buyer is holding. Use the printed set name, not a nickname you invented.
Number is the tiebreaker that never lies. Names get reused across two decades of releases; the number plus the set is close to unique.
Condition is the difference between a card worth eighty dollars and the same card worth thirty. If it lives in your head instead of the record, your inventory value is fiction.
Location is the one people skip and the one that costs the most. Knowing you own the card is worthless at 11am on a Saturday if you cannot put your hand on it in twenty seconds.

If a card in your system is missing any of those four, treat the entry as unfinished, not as done.

Pick one way to write a name and never deviate

Inconsistent naming is the biggest reason searches fail. A card entered as "Umbreon VMAX alt" in March and "Umbreon VMAX Alternate Art" in July is two different things to any search box, and you will find one of them and sell the other twice.

Write the name as it appears on the card, then put the qualifier in a fixed position with fixed spelling:

Same words for variants every time: Holo, Reverse Holo, Alt Art, Full Art, Secret. Pick one spelling and write it down. The choice matters less than the consistency.
No personal shorthand. "Moonbreon" is what a buyer will say out loud, but if half your entries use the nickname and half use the printed name, your search is split in two.
Skip punctuation you will not retype at the table. Apostrophes and colons are landmines when you are typing one-handed.

If you are starting from a spreadsheet, do this cleanup before you import. Fixing four hundred rows in a spreadsheet is an afternoon. Fixing them one at a time on a phone is a month of small annoyances, and moving inventory off spreadsheets covers that pass in more detail.

Consistency beats correctness. A slightly odd convention applied every time serves you better than the perfect convention applied half the time.

Let the catalog fill the fields you keep getting wrong

Set names and card numbers are exactly the kind of data humans enter badly. They are long, they repeat, and they are boring to type when there is a line at the table. Every typo becomes a card you cannot find later.

This is the part worth handing to software. CardOps has a scanner: point the camera at a card and on-device recognition matches it against a catalog of more than 20,000 Pokemon cards, then attaches a live market value. Set, number and printing come from the catalog rather than from your thumbs, so the same card scanned in March and in July lands under the same name both times.

That consistency carries the rest of the system. Market values are tracked by set and condition, so an entry with the right set attached is an entry with the right value attached. An entry with the set typed wrong is a card you own, priced as a card you do not. Type as few identifying fields as you can get away with.

Condition is a field, not a memory

Condition needs to be recorded when the card enters your stock, while you are holding it under decent light and not in a hurry. It never gets more accurate later.

Grade	What it means on your table
NM	Pack fresh or close. No visible wear at arm's length or under a light.
LP	Minor edge wear or light surface marks. Sleeve-ready, small discount.
MP	Obvious wear, scratches, or whitening. Priced accordingly and described up front.
HP / DMG	Creases, water damage, heavy whitening. Bulk or filler pricing.

Two rules keep this honest. Grade against the card, not against what you paid for it; the urge to call an LP card NM is strongest right after you overpaid, which is when it does the most damage. And when you are torn between two grades, take the lower one. A buyer who finds the card better than described comes back. One who finds it worse does not, and tells people.

Keep graded copies as separate entries from raw copies of the same card rather than as a note on one row. They are different products with different prices and different buyers. More on that split in graded versus raw, and on wear itself in how condition affects value.

Label the boxes so the record can point at them

Your record is only as useful as your storage is findable. If every box is labeled "Pokemon", the record cannot tell you anything actionable about where a card lives.

Give every container a short, permanent name and write it on the container itself. Not a description, a name.

Numbered boxes. B1 through B12, written large on two sides so it reads from the floor and from the table. It never has to change when the contents do.
Purpose names for what you touch every show. CASE for the glass, HITS for the high end, BULK for the commons.
Labeled dividers inside a box. A box holding 800 cards needs internal structure. Divide by set and keep the dividers in a fixed order.

Sort inside each container the same direction every time: by set, then by number. Alphabetical by card name feels natural and falls apart the moment you own four Charizards from four sets.

The high-value rule is separate. Anything above your "I would be sick if this walked" threshold lives in the case, in a fixed spot, and gets counted at the end of every show. The show day setup checklist covers how that maps onto the table.

Keep the record and the boxes from drifting apart

Every system is accurate on the day you build it. Drift starts the first time a card moves without the record moving with it. A card sold from the case and not marked sold. A card pulled for a trade and put back in the wrong box. A buy that went into the bag and never got entered.

Enter at intake, not later. A card that is not in the record is not in your inventory. The end-of-show pile of unentered buys is where organizing systems go to die.
Refile into the container the record names. "I will put it back properly later" is how a card becomes lost while still being owned.
Audit one box between shows. A rolling count is something you will actually do. A full inventory is something you will schedule twice and cancel twice.

It helps that the record is on the phone in your hand rather than a laptop in the car. CardOps runs search, sales, trades and buys on the device, with no account to sign into and nothing waiting on convention-center wifi, so entering a card at intake costs seconds instead of a trip to the back. Data lives on your device and in your own private iCloud, and Pro adds CSV import and export in both directions.

An hour of setup before your next show

You do not need to reorganize everything to get most of the benefit.

Write the convention down. One note, five lines: how you write set names, variants, and condition.
Label every container. Marker on tape, two sides, short names.
Fix your top fifty cards. The cards carrying your inventory value get correct set, number, condition and location first.
Sort the case in the order you will reach for it. It pays for itself in the first hour of the show.

Do not clean the long tail in one sitting. Bulk gets fixed when it gets touched, and most of it will get touched at the next show. If you are still moving from binders and a spreadsheet, migrating from binders to an app walks through the order to do it in.