Price book cleanup before a migration: cutting 10,000 items down to what you actually sell

Most 10,000-item price books are about 1,000 items your techs sell and 9,000 items they scroll past. The 9,000 got there honestly: a supplier catalog loaded in one go, a consultant's template, three different people adding "capacitor" in three different ways. But when you switch software, every one of those items has to be mapped, checked and imported, and every one of them gets in a technician's way at the kitchen table on the new system too.

A price book cleanup before migration is the cheapest moment to fix that. This post gives you a method: rank every item by how often it actually appears on invoices, sort the book into four bins, standardize names, and import only what earns a place. The worked example uses a hypothetical 10,000-item book.

What the research says about cutting choices

There's a famous experiment behind the idea that fewer options sell more. In a 2000 study, shoppers who stopped at a tasting table with 6 jams were far more likely to buy than those who stopped at one with 24: 30% of the limited-display group bought a jar, against 3% of the extensive one.¹ It's a great story, and it's often oversold. A 2010 meta-analysis of 50 experiments found the average effect of "too much choice" was close to zero, with large swings between studies.²

The more useful evidence for a price book comes from retail assortment cuts. When an online grocer made dramatic cuts in mostly low-selling items, sales rose an average of 11% across 42 categories, and the researchers found shoppers "uniformly welcomed" the removal of redundant items, even when they reacted differently to losing sizes.³ A later study of a large assortment cut at an online grocer found overall store sales fell, with less-frequently purchased categories hit hardest.⁴

Read together, those studies say something practical: cut redundancy, keep real choice. Five near-identical capacitor entries are clutter. Good, better and best options on a system replacement are choices the customer wants. A price book cleanup should remove the first and protect the second.

Step 1: rank items by invoice lines, not by price

Pull every invoice line for the last 24 months: item code, item name, quantity, price, date. Count how many invoice lines each item appears on. Twenty-four months instead of twelve so seasonal and odd-year items get a fair chance.

Sort items from most lines to fewest and add a running total. In most bloated books you'll see a steep curve: a few hundred items carry most of the work, and a long tail has never been sold at all. Here's what that looks like in our hypothetical book.

Two cautions. First, if techs regularly sell from a catch-all item ("misc. repair", "custom labor"), those lines hide what was really sold; read the descriptions for the top catch-alls before you decide anything. Second, rank by lines, not revenue. A $12 part on 900 invoices matters more to a tech's day than a $9,000 system sold twice, even though the system matters more to your revenue. Step 3 protects the big-ticket items separately.

Step 2: sort every item into four bins

Add the 200 protected items from Step 3 and the active book is about 1,250 items, 12.5% of the original, covering roughly 97% of invoice lines plus the rare expensive jobs. Your numbers will differ. The shape usually doesn't.

How to spot duplicates

Duplicates rarely share a name. Look for items that match on two or more of these:

• the same price and the same category,
• the same linked materials,
• names that match once you strip punctuation, sizes written differently ("1/2 HP" versus "0.5hp") and words like "new" or "copy",
• one item that was used until a date and another that took over the day after.

When you merge, keep the item with the cleaner name and the most history, and write down the old code. You'll want it when someone searches an old invoice.

Step 3: protect the rare, expensive and required

Usage alone would cut things you must keep. Add a protect list before you finalize the bins:

• High-ticket, low-frequency work. Any item with a price above a threshold you choose (say $1,500) that sold even once in 24 months.
• Compliance and safety items. Permits, inspection fees, refrigerant recovery, disposal fees. Rarely the reason for a call, often required on it.
• Membership and warranty items. Member-only prices, warranty labor at no charge, inspection tasks that membership visits rely on.
• Option ladders. If a good, better, best set has one option that rarely sells, keep it: the cheap option often exists to make the middle one look reasonable. See good, better, best estimates.

Step 4: fix names and structure before you import

Renaming 1,250 items is a few days of work. Renaming them after the new system is live, while techs are learning it, is a few months of confusion. Pick a standard and apply it during the cleanup.

A good naming pattern is task, object, spec, condition: "Replace capacitor, dual run, up to 45/5 µF." It sorts sensibly, reads well on an invoice, and a tech can find it by typing the first word that comes to mind.

Structure matters as much as names. Flat-rate tasks are what customers buy; materials are what you stock and cost. Keep them separate and link the materials to the tasks, so a sold task pulls its parts for job costing and truck restocking. If you're deciding which jobs should be flat rate at all, see flat rate versus time and materials and the glossary entry on flat-rate pricing.

Step 5: re-price while you're in there

A cleanup is the one time you'll look at every surviving item. Check each kept task against your current labor rate and material markup, not the rate from the year the item was created. A price that was right at a $95 labor rate is wrong at $140.

Step 6: run a search test with real techs

The best check on a cleaned book isn't a spreadsheet. It's three technicians and a stopwatch. Before you import into the live system, load the cleaned copy into a test company and run this:

1. Pick the 20 tasks your shop sold most in the last year, in plain words ("swap a capacitor," "clear a drain line," "replace a water heater thermocouple").
2. Ask three techs, one senior, one mid-level and one newer, to find each item on a phone, the way they would in front of a customer.
3. Write down how long each search takes and whether they picked the right item.

You're looking for two things. Any task that takes more than a few seconds to find has a naming problem: the item exists but doesn't start with the word techs think of. Any task where techs pick different items has a duplicate you missed. Fix both, then run the test again with a different 20 tasks. Your price book is a search tool before it's a price list; test it like one.

Who should do the cleanup, and how long it takes

Don't hand this to whoever has a free afternoon. You need two people: someone who knows the field (a senior tech or service manager, who decides what a task is and whether two items are really the same) and someone who knows the spreadsheet (an office manager or bookkeeper, who runs the counts and keeps the files straight). Neither can do it alone.

As a rough example for a 10,000-item book: half a day to pull the exports and count lines, a day to sort the bins and build the protect list, two or three days to merge and rename the 1,250 survivors, a day to re-price, and half a day for the search test. Call it a week of part-time work over two or three weeks. Most of the time goes into names, which is why it's worth doing once, properly, before the import rather than after.

The vendor content trap

If your price book started from a vendor's content service, read your contract before you export. ServiceTitan's Terms of Use, for example, say that when a Pricebook Pro subscription ends you get 10 days to access your content to remove ServiceTitan-provided content, and only after ServiceTitan agrees that provided content (and content derived from it) is gone will it let you export your own.⁵ The same Terms bar you from transferring Pricebook Pro content to third parties.⁵

In practice that means two things. Know which of your items, descriptions and images you created and which came with the subscription, and plan to rebuild the provided ones. And do the cleanup while the subscription is still running, because ten days is not a lot of time.

Getting it out and back in

For ServiceTitan, the export is Pricebook, Import/Export, Export tab, export type "Pricebook (Settings, Materials and Part Link)", one business unit at a time, with "Include inactive" if you want the full history.⁶ The workbook has sheets for Categories, Services, Materials and Equipment plus three link sheets (ServiceMaterialLinks, ServiceEquipmentLinks, EquipmentMaterialLinks) and an Assets sheet. Images are stored as URL fields, and category images reference ServiceTitan's internal file path.⁷ Save the image files you want to keep; don't count on those links after your access ends. We list the other records worth taking in 14 things owners forget to export.

Do the bin sorting in a copy of the export, not in the live system. That way the original export stays your untouched archive, and the cleaned copy is what you import.

app.tradeswork.app · Price book with services, prices and margins

TradesWork imports the ServiceTitan pricebook workbook with all its sheets, keeps the service-to-material link sheets, and keeps each item's ServiceTitan code, so dropping a corrected export again updates items rather than duplicating them. Two honest limits: the ServiceTitan layouts were built from ServiceTitan's screens and help docs and are still being checked against real exports, and TradesWork doesn't pull live distributor pricing or build a price book for you. The cleanup is yours, which is the point of this post.