Catalogue vs Free-Text Requisitions: Killing Maverick Spend

Maverick spend is the money that leaves your organisation outside the deals you negotiated: the requester who types "assorted fittings" into a free-text line and orders from whoever answers the phone. Catalogues are the usual cure, and they work, but I have watched too many sites try to catalogue everything, stall for a year, and revert to free-text in frustration. The honest answer is a hybrid. This guide shows how I split a spend base into catalogue, contracted price list, and disciplined free-text, then move a site there in ninety days.

The real trade-off

There are two competing goods here and you cannot maximise both. A catalogue, whether a punch-out session into the supplier's own storefront or a hosted price file loaded into your ERP, gives you structured line items: a fixed description, a part number, a unit of measure, a contracted price, and a category code that never has to be guessed. That structure is what makes spend analysis, price compliance, and three-way matching possible. The cost is effort. Someone has to build the catalogue, agree the pricing, map the units, and keep all of it current as suppliers change packs and prices.

A free-text requisition line is the opposite bargain. The requester types what they want, picks a supplier, and moves on. It is fast, it handles anything, and it needs no setup. What it produces is unusable spend data: "pump seal", "seal for pump", and "mechanical seal 40mm" are three strings describing one item, bought at three prices, that no report will ever group. You gained speed and lost the ability to see what you actually buy.

A decision table by category

The mechanism should follow the shape of the spend, not a blanket policy. High-frequency, standardised, catalogue-friendly items reward a punch-out or hosted catalogue. Commodity items bought against a firm contract only need a price list. Genuinely unpredictable spend should stay free-text with a mandatory category code so it is still visible. Here is how I map common categories.

The pattern is not arbitrary. Catalogues earn their keep where the same items recur at volume. Price lists suffice where the range is narrow and the contract is firm. Free-text is the honest answer where no two orders look alike, provided you never let a line through without a category code.

A worked spend-analysis example

Here is a pattern I see again and again when I profile a site running mostly on free-text. Take a year of roughly 4,000 free-text requisition lines. Run them through even a rough cleanse, normalising supplier names and grouping obvious duplicates, and the picture that tends to emerge looks like this. Treat every number below as an indicative field pattern, not a benchmarked statistic; your real base will differ, but the shape rarely does.

Read what that is telling you. The chaos of 4,000 free-text lines and thousands of unique strings collapses to about fifteen suppliers and a few hundred genuinely repeatable items. If you catalogue only the top 20 percent of item lines, roughly 160 items, you capture on the order of 70 percent of transaction volume. That is the whole argument for a hybrid in one sentence: a small, targeted catalogue against a handful of suppliers does most of the work, and you never have to touch the long tail. Again, these are indicative proportions from field experience, not audited figures, but they are why "catalogue the vital few, price-list the commodities, free-text the rest" is a sound default.

This exercise is also the natural next step after you have a clean requisition-to-order flow in place. If your PR to PO to GRN workflow already captures structured lines, the spend analysis above is a report you can run rather than a data-cleaning project you have to fund.

The maintenance burden nobody prices

The reason catalogue programmes stall is almost never the initial build. It is the maintenance, which is real, ongoing, and routinely left out of the business case. Before you commit a category to a catalogue, price these four things honestly.

Refresh cadence.

A hosted catalogue is a snapshot. Suppliers add and retire SKUs constantly, so someone has to agree how often the file is reloaded, monthly or quarterly, and who chases the supplier for the updated file. A punch-out shifts that burden to the supplier, which is exactly why punch-out wins for large, fast-moving ranges like IT and stationery.

Price validity dates.

Every catalogue price should carry a valid-from and valid-to date. Without them, a price agreed last year keeps flowing into new orders, and your "compliant" catalogue quietly becomes a source of overcharging. Someone owns the calendar of expiries and renegotiations.

UOM conflicts.

This one bites hardest. The supplier sells a box of 100; your storeroom issues by the single unit. If the catalogue loads the supplier pack size but your item master expects the issue unit, every receipt and every stock figure is wrong by a factor of a hundred. Reconciling the supplier pack UOM against the storeroom issue UOM is unglamorous, error-prone work, and it has to happen for every catalogued line.

Who owns the item master.

A catalogue is only as good as the item master behind it. If no single role owns adding items, retiring duplicates, and enforcing the naming standard, the master rots and the duplicates you cleansed out grow straight back. Name that owner before you start, not after.

A 90-day sequence to a workable hybrid

If a site is 95 percent free-text today, do not announce a catalogue mandate and freeze ordering. Operations will route around you within a week. Move in three phases that never block a live requisition.

Days 1 to 30: See the spend and stop the bleeding.

• Pull the last twelve months of free-text lines and run the cleanse from the worked example above: normalise suppliers, group duplicate item strings, rank by volume.
• Make a category code mandatory on every free-text line from day one. This alone rescues your analysis and costs nothing operationally.
• Identify the vital few: the top 15 or so suppliers and the top 20 percent of item lines. That is your catalogue target list.

Days 31 to 60: Load the easy wins.

• Stand up punch-out for one or two high-volume, long-tail categories where the supplier already runs a storefront, typically IT consumables and stationery.
• Load a hosted catalogue for one standardised category you control tightly, usually PPE or standard MRO, and resolve the UOM mapping properly before go-live.
• Put contracted price lists behind the narrow commodity categories: chemicals, filters, lubricants. No full catalogue, just the rate.

Days 61 to 90: Govern and hold the line.

• Assign the item-master owner and the price-validity calendar as named responsibilities, not shared ones.
• Report catalogue coverage weekly: share of lines and share of value flowing through catalogue or price list versus raw free-text.
• Route free-text lines above a threshold value in a catalogued category to a buyer for a quick check, so maverick spend gets caught without stalling the genuine one-offs.

At the end of ninety days you will not be at 100 percent catalogue coverage, and you should not want to be. You will have the vital few under catalogue or price list, the long tail visible through mandatory category codes, and a governance rhythm that keeps it from decaying. That is a workable hybrid, and it is where the maverick spend actually dies. The approval discipline that sits alongside it is worth building in parallel; I cover that in purchase requisition and approval automation.

The bottom line

Catalogue where the same items recur at volume. Price-list where the range is narrow and the contract is firm. Free-text where no two orders look alike, but never without a category code. Chase coverage of the vital few, not coverage of everything, and price the maintenance burden before you commit a category, because that upkeep, not the build, is what sinks catalogue programmes. Do that and your spend data becomes something you can actually negotiate from. If you want the full cost picture behind these platforms and their upkeep, I break it down in what procurement software really costs.

A note on independence and references. I am not affiliated with any procurement platform vendor and take no commissions. Product mentions are illustrative, not endorsements. Punch-out interoperability is commonly built on the cXML standard cxml.org and the older OCI approach; confirm which your ERP and suppliers support before you design a catalogue programme. Suites such as Coupa package catalogue, punch-out, and free-text controls together, but the trade-offs in this guide apply whatever tool you run. All percentages here are indicative field patterns, not benchmarked figures.