AI image

Supplier Intake Became a Layer. In the PIM RFP It Is Still One Line

Supplier data intake is now a separate layer, yet most PIM RFPs give it one line. Five requirements to add before you compare vendors.

On 7 October 2026, PIMvendors, a Dutch firm that describes itself as an independent PIM consultancy, published the first part of a series called Trends in PIM. The author, Stephan Spijkers, puts the thesis in one sentence: “Getting data into a shape the PIM can trust is becoming someone else’s product.” (PIMvendors, Trends in PIM, Part 1, updated 18 October 2026.)

His evidence comes from three directions. Specialist tools now sell supplier intake on its own, with separate pricing and their own APIs. Retail and syndication platforms put a staging step in front of the system of record. And PIM vendors have started to answer with intake APIs of their own; he points to inriver’s Spring 2026 release. The PIM stays the system of record. The work of getting supplier data to its door is moving out.

A disclosure before going further: openProd is one of the three specialist tools the article names. We did not commission the article, pay for the mention or know about it in advance. Read what follows with that in mind.

He is right about the shift, and his advice to buyers is sound. He gives them three things to check with a vendor: where bad data gets rejected, who owns the channel that faces the supplier, and what the PIM expects at its entry point.

The checks come after the document that decides the answer

I think the article is too polite about timing. Those three checks belong to the vendor conversation. By the time that conversation happens, the RFP has been written and the budget approved, and in that document supplier intake is usually one line.

A vendor answers the question that was asked. A line that reads “import of product data from Excel and CSV” gets a truthful “yes, supported” from every vendor on the shortlist. It separates nobody, and it funds nothing. The checks from the article then have no place to land. No requirement exists to score them against, and no budget line would pay for a different answer.

So the shift Spijkers describes is visible in vendor price lists and release notes. It is not yet visible in the document buyers use to choose.

What the briefs we receive ask for

LemonMind, the agency that built openProd, has delivered more than 70 PIM implementations over 15 years, and each of them started with a brief or an RFP. We have yet to receive one that asks for the thing the article is about: take the supplier’s file as it arrives, turn it into records in our model, and show us what you were unsure of before anything is written.

What briefs ask for instead are the mechanisms their authors have already seen.

What the brief saysWhat it specifiesWhat it leaves without an owner
”Import of product data from Excel and CSV”The file format the importer acceptsWho reshapes each supplier’s file into that format
”Repeatable, scheduled import per supplier”One mapping per supplier that someone maintainsThe first file from a new supplier, and the file whose layout changed
”Supplier portal for entering and uploading data”A channel and a loginWhat happens when the supplier sends a PDF catalogue anyway
”Integration with ERP”Master data: item number, price, stockTechnical attributes that exist only in supplier documents

The wording in the table is paraphrased from patterns we see repeatedly. It is not a count, and no line is quoted from a single customer.

The reason is not carelessness. Nobody writes a requirement for something they have never seen work. Until recently the honest answer to “can the system read a 300-page supplier catalogue and map it to our classification” was no. So the work went to people: product managers, junior analysts, an agency’s hours. Work that is assigned to people does not appear in a software requirement, and what is not in the requirements is not in the comparison.

What the missing line costs

It helps to see what the work looks like once it is measured. In a pilot this autumn an industrial distributor gave us 12 files from six suppliers: catalogues, brochures, data sheets and one multi-sheet spreadsheet in seven languages. The run produced more than 9,000 attribute values in about 67 minutes of machine time, followed by up to two hours of human review and roughly 550 decisions, most of them taken in bulk. The full numbers are in the write-up of that pilot.

Those figures describe intake when it is a system’s job. It has a start, an end and a log. When the RFP does not name it, the same work still gets done. It turns up in one of two places.

Where the work landsHow it is paid forWhen finance sees it
Inside the implementation, as a custom importer per supplierChange requests against the project budgetAfter vendor selection, as a variance
Inside the buyer’s team, as manual preparation of filesSalaries, permanentlyNever as a line, only as headcount

Neither is wrong in itself. The problem for whoever signs the business case is that neither was priced during selection, so the shortlisted vendors were never compared on it.

The same pilot showed a second gap. The customer’s existing model, close to four thousand attributes, matched 37% of what the supplier files carried. A requirement that stops at “import into our model” says nothing about the rest.

A section you can paste into the RFP

Before the requirements, make one decision that Spijkers also asks for: name who owns the supplier-facing channel. A retailer-style portal, a specialist layer and the PIM’s own intake path are all defensible. Running several in parallel without deciding is how the spreadsheet comes back.

Then ask for outcomes, and ask each vendor to show them on your files.

  1. Input as it arrives. The solution accepts supplier product data in the formats suppliers send today, including PDF catalogues and data sheets, multi-sheet spreadsheets and XML, without the supplier or our team reformatting the file first.
  2. Our model is the target and the only definition. Extracted values are mapped to our existing product model, read from the PIM. The intake step does not keep attribute definitions of its own. This is the direct answer to the risk the article names for a specialist layer: “another place where attribute definitions can drift from the PIM model”.
  3. What does not fit is kept and reported. Values with no matching attribute are retained and listed for a modelling decision. They are not dropped at import.
  4. Review before write. Nothing reaches the PIM without a preview. Uncertain values are flagged, decisions can be taken in bulk, and the log shows who approved what.
  5. Writes go through the PIM’s own entry point. Data enters through the PIM’s API or import interface, so the PIM’s validation rules apply. No direct database writes and no private scripts.

Add one commercial line: intake is priced separately, with the setup cost, the cost per supplier or per file, and the cost of a repeat run from a supplier already mapped.

These requirements are not written around one product. A PIM’s native intake API can meet some of them, a mapping tool others, a specialist layer others again, and vendors will be stronger on different lines. That is the point of asking. If you want a method for checking the answers, we described a trial you can run on your own files with any tool, and a comparison of the tools in this layer, ours included.

What changes when the line exists

With the section in place, vendors stop giving the same answer. Some will say the capability is native, some will name a partner, and at least one will say it is out of scope. Each of those is a useful answer, because it can be compared and priced before the contract and not discovered after it.

Spijkers is clear that neither route is free. A specialist layer means another vendor and another contract. A PIM-native path keeps ownership in one place but may lag on difficult PDFs and many suppliers at once. The RFP is where a buyer can see the price of each option before the choice is made.

His closing line deserves to be quoted in full: “The expensive choice is pretending the peel is not happening while your team still remaps the same supplier workbook every Monday.”

Which line of your RFP pays for that Monday?

If you are writing a PIM RFP now and want to see what requirement 1 looks like on your own data, send us three supplier files as they arrived and book a session. You will get them back as records mapped to your model, with the uncertain values marked.

Source: Stephan Spijkers, “Trends in PIM, Part 1: The onboarding layer is peeling off from the PIM”, PIMvendors, published 7 October 2026, updated 18 October 2026, pimvendors.com. Quotations are from the version read on 13 November 2026.

See it working on your own files.

Book a short demo and we will run a real supplier file through openProd with you.