From Excel to PIM: a 5-step migration guide for B2B SMEs

Based on our internal estimates at PixeePIM, around 73% of B2B SMEs still distribute their product catalog from an Excel file or Google Sheets. It works — up to a point. Then it breaks. This article walks through the warning signs and outlines a concrete 5-step migration to a real PIM, written for an e-commerce manager or SME owner — not for a CTO.

Excel is not the problem — it's the tool's natural limit

First, let's be clear: Excel is a fantastic tool. It's great for prototyping, modeling a catalog as it starts to grow, running ad-hoc calculations, preparing a one-off export. Most product catalogs we see at PixeePIM started in Excel — and that was the right call at the time.

So the problem isn't cultural ("Excel is old school"). It's structural. Excel is a spreadsheet, designed for one person to manipulate data at a given moment. It's not a system for shared, versioned, multi-channel data. Once your catalog reaches several thousand SKUs, multiple sales channels, and several people editing at the same time, the tool has left its home turf.

The right reflex isn't to blame Excel. It's to recognize that you've hit its ceiling, and to move to a tool whose native job is product data: a PIM (Product Information Management).

The 7 signals that say "Excel is no longer the tool"

If you recognize three or four of these situations, you've outgrown Excel's comfort zone:

• The file is over 50 MB — it takes 30 seconds to open, freezes on sorts, and breaks past 50,000 rows. Beyond 100,000, Excel often refuses to load at all.
• Multiple people edit at the same time — result: version conflicts, files named catalog_v2_final_OK_juliesEdit.xlsx, and nobody knows which copy is the source of truth.
• Variants explode as duplicated rows — a t-shirt in 4 sizes × 5 colors = 20 rows, each with the same attributes (material, brand, supplier) copy-pasted by hand. Update the brand: 20 cells to fix.
• Images and tech sheets live in a separate Drive — Google Drive, Dropbox, a NAS. The link between the Excel row and the image only exists in the head of the person who uploaded them. The day they go on holiday, nobody knows.
• Export to Shopify, PrestaShop or marketplaces is manual — copy-paste, manual column adjustments, image reprocessing, and start over for the next export.
• Nobody knows which attribute is "official" — you have a column "description", a "description_2", a "description_v2", a "description_marketplace". None of them is the source of truth.
• Regulatory compliance becomes untraceable — REACH, AGEC, DPP, EPREL. When a customer asks for a product safety sheet, you have to dig through a tab, then a Drive, then a salesperson's email inbox.

Before migrating, measure what you actually have

Before shopping for a tool, you need numbers. Without these, it's impossible to pick the right PIM plan or anticipate the migration effort. Ask yourself five questions:

• How many active SKUs? Not the number of rows in Excel — the number of products actually sold in the last 12 months. Often only 30 to 50% of the total.
• How many distinct attributes? Count the used columns in your Excel file. If you have 80, you'll have 80 attributes to model. If half are empty 95% of the time, you can drop a lot.
• How many suppliers or data sources? 1 importer? 5 brands you distribute? 30 suppliers each with their own format? This decides whether you need automated imports or whether manual import is enough.
• How many sales channels? E-commerce site, Amazon, eBay, B2B portal, retail EDI. The more channels, the more you'll need to manage different content per channel (long Amazon description vs short site description).
• How many read/write users? 2? 10? 30? This drives plan selection and the training effort.

A typical B2B distributor SME with 8,000 active SKUs, 40 attributes, 5 suppliers, 3 channels and 4 users fits comfortably in a PixeePIM Free plan for evaluation, then in a Starter plan for production.

Step 1 — audit and clean the current Excel file

The worst idea would be importing 50,000 dirty rows into a brand new tool. A PIM started on dirty data = a PIM broken within a week. Before any import, do a cleanup pass:

• Duplicate EANs — use a COUNTIF formula to spot barcodes appearing twice. Decide: duplicates to merge, or genuinely distinct products miscoded.
• Empty or incomplete rows — a product without a title, EAN, or supplier is a ghost. Either complete it or archive it.
• "Someday to delete" columns — you know, the ones nobody has filled in two years. Decide: keep and fill, archive separately, or delete.
• Broken formulas — the #REF!, #N/A, formulas pointing to a deleted cell. Convert to values before export.
• EAN formatting — Excel loves converting an EAN-13 into scientific notation 4.2E+12. Force text format, or use a validation tool. See our EAN Manager article to automate this validation.

Plan for 2 to 5 days of cleanup for a 10,000-SKU catalog. It's unglamorous but unavoidable. You'll do it now, or you'll do it later — painfully — inside the PIM.

Step 2 — model the target catalog

This step happens on paper or in a diagram tool (Miro, Whimsical, even a notebook). You don't touch the PIM yet. You sketch the target structure.

Three key distinctions to nail:

• Products vs variants — a "red t-shirt size M" is a variant of the product "t-shirt". The product carries the brand, description, generic images. The variant carries the size, color, EAN, stock, price.
• Attribute families — all clothing shares the same 15 attributes (material, cut, care, etc.). All power tools share another set (wattage, voltage, runtime). Defining these families avoids ending up with 200 attributes where 90% only apply to one product type.
• Categories and hierarchies — your e-commerce site tree, your Amazon categories, your B2B aisles. A product can sit in several categories, in several trees (one for the site, one for marketplaces).

Invest 1 to 2 days on this modeling. It's the highest-return cost of the entire migration. Bad initial modeling = weeks of rework six months down the road.

Step 3 — pick the tool and create a test environment

Criteria to weigh for a B2B SME:

• SaaS vs self-hosted — for 95% of SMEs, SaaS is the right call. No server, no IT team, no upgrades to manage. Self-hosted (Akeneo Community, Pimcore) only makes sense if you have a motivated in-house IT team.
• Free plan to test — demand one. Testing a PIM for 2 weeks on a sample of your real catalog is the only way to validate the choice.
• Native English (or your language) support — documentation, onboarding, email support in your team's working language is a significant comfort.
• Regulatory compliance — GDPR of course, plus the industry modules if they apply to you: AGEC, DPP (kicking in from 2026 for several sectors), REACH/RoHS for electronics.
• Native connectors — check that your main channel (Shopify, PrestaShop, Amazon, Mirakl, etc.) is natively supported, not via custom development.

PixeePIM offers a Free plan sized for this testing phase: 2,000 products, 1 supplier, all modules including BYOK AI, no credit card required. Enough to migrate a representative sample (an entire category, for example) and validate the tool fits before committing budget.

Once the test is conclusive, the Starter plan at €490/month opens production: 50,000 products, 1 channel connector, automated FTP/EDI imports, native regulatory compliance.

Step 4 — import in staging mode

You've cleaned your Excel (step 1), modeled your target structure (step 2), picked your tool (step 3). Time to import — in staging mode, never directly in production.

Concretely:

• CSV export from Excel — a UTF-8 file, comma or semicolon separator, consistent encoding. Verify accented characters survive.
• Column mapping — the Excel column supplier_ref maps to the PIM attribute supplier_ref. The column desc maps to description_long. Document this mapping in a file — you'll come back to it.
• EAN validation — a serious PIM detects invalid EANs (bad checksum), duplicates, malformed strings. See our article on EAN validation.
• First completeness audit — how many products have a full description? A weight? An image? The completeness score gives you a snapshot in seconds.
• Image upload — the often-forgotten part. If your images are on Google Drive named IMG_0023.jpg, now is the time to rename them with the EAN or internal reference so they can auto-link.

This phase usually takes 1 to 2 weeks for an average catalog. Not because of technical complexity, but because you discover data issues you hadn't anticipated and need to fix them on the fly.

Step 5 — switch to production, wave by wave

A successful migration is never a "big bang". Break the production switch into successive waves to limit risk and let the team ramp up gradually.

• Wave 1 — one product category, one sales channel. For example: all "hand tools" products published only on your e-commerce site. 200 to 500 SKUs. Observe for 2 weeks.
• Wave 2 — add a second category and a second channel. Keep observing.
• Wave 3+ — expand category by category until you cover the full catalog and all channels.

Throughout this, keep Excel running in parallel for 2 to 4 weeks as a safety net. The team won't feel fully comfortable until they've gone through several real cycles. Once confidence is built (and the PIM is identified as the single source of truth by everyone), archive the Excel and communicate clearly: "from this date, the PIM is the only official source".

Pitfalls to avoid

• Wanting to migrate everything at once — the temptation is strong, especially when the current Excel is painful. Resist. A wave-based migration drastically cuts risk.
• Underestimating team training — even a well-designed PIM requires 2 to 4 hours of onboarding per user. Block the time. Don't drop a login and a link on your team and walk away.
• Not updating the image Drive in parallel — if the PIM becomes the source of truth for product sheets but images keep living in Google Drive with no sync, you're rebuilding the original problem. The PIM's DAM (Digital Asset Management) must become the single source of media.
• Forgetting marketplaces — Amazon, eBay, Mirakl operators each have their own attribute format. Anticipate the mapping at tool selection time.
• Picking a PIM too powerful for actual needs — Akeneo Enterprise for 5,000 SKUs is using a semi-truck to deliver pizzas. Pick a right-sized tool, and move upmarket in 18 months if growth justifies it.

FAQ

How long does an Excel-to-PIM migration take?

For an SME with 5,000 to 15,000 SKUs and a relatively clean Excel file: 4 to 8 weeks total, including 1 week of scoping, 1 to 2 weeks of cleanup, 1 to 2 weeks of staging import, and 2 to 4 weeks of wave-based production switch. For a very dirty or much larger catalog, plan for 3 to 4 months.

Do I need a developer to migrate?

No, for a modern SaaS PIM like PixeePIM. CSV imports, column mapping, attribute configuration are all point-and-click. A developer becomes useful if you need automated imports from supplier APIs or an ERP, but not for the initial migration.

What happens to my linked files (PDFs, images, tech sheets)?

A modern PIM includes a DAM (Digital Asset Management) that stores and links these files to products. During migration, you upload your media and link them via the EAN or an internal reference. Over time, the PIM becomes the single source for product sheets, images and technical documents.

What if I change my mind partway through?

That's the main reason you keep Excel in parallel for 2 to 4 weeks, and migrate in waves. If the tool doesn't fit, you've only lost scoping time and the first wave — not your entire catalog. Demand a full CSV export from your PIM at any time: it's your contractual safety net.

How much does a SaaS PIM cost for an SME?

A SaaS PIM right-sized for a B2B SME runs between €0 (Free plans, enough for evaluation or very small catalogs) and €500 to €1,500/month for serious production with automated imports and multiple channels. Compare that to the hidden cost of an unmanageable Excel: time lost in double entry, marketplace errors, customer returns, regulatory non-compliance. See the PixeePIM pricing page for the full plan details.

Can I migrate from Google Sheets or Numbers, not just Excel?

Yes — the process is identical. All these tools export to CSV, the standard input format of a PIM. The pitfalls (UTF-8 encoding, EAN formatting, column separator) are the same.