Supersession: The Most Under-Modeled Relationship in Industrial Catalogs

Every industrial catalog has supersession data. Almost none model it as a first-class relationship with direction, date, quantity ratio and source. This article explains why, and what it costs.

Supersession — the manufacturer’s replacement of one part number with another — exists in every industrial catalog. It is also the relationship most likely to be stored as a free-text note, a flat two-column spreadsheet, or a product specialist’s memory. The under-modeling costs sales, creates returns, and wastes the most expensive resource on the team: the people who know which parts replaced which.

How supersession enters the catalog

Supersession data rarely arrives through a dedicated channel. It shows up as a side effect of other processes:

Price-list updates. A part number disappears from the new price list. A new number appears in the same product group. Nobody labels the change as a supersession — it is just a delta between two files.

ERP status changes. A product specialist marks a part as “discontinued” in the item master and types the successor’s number into a notes field. The notes field is free text. The successor number may or may not be correct, current, or formatted consistently.

Phone calls and emails. A manufacturer’s rep mentions on a call that part A has been replaced by part B. The sales rep writes it on a sticky note, or in a Slack message, or nowhere. The information exists for the duration of the sales rep’s memory.

Supplier portals. The manufacturer’s website shows “replaced by [new number]” on the old part’s page. The catalog team may or may not check the portal. The portal may or may not be up to date.

None of these channels produce a structured record with direction, effective date, quantity ratio and source. They produce signals that a structured record could be built from — if anyone is building.

What under-modeled means

Under-modeled supersession is not the absence of data. It is data without structure:

What the catalog has	What it needs
A two-column spreadsheet: old part, new part	Direction (old → new, one-way), effective date, quantity ratio, source, confidence, status
A free-text note: 'replaced by AB-2050'	A typed REPLACED_BY edge linking two product identities
A product specialist who knows the history	A data model that captures what the specialist knows, with evidence, so it survives their departure
A single-hop replacement	A chain that follows through intermediate replacements to the current active part
No record at all — the part just shows 'out of stock'	At minimum, a discontinued status with a flag indicating whether a successor exists

What it costs

Lost sales

A customer searches for the old part number. The catalog returns “no results” or “out of stock.” The customer goes to a competitor whose catalog resolves the old number to the current one. The sale is lost not because the product is unavailable, but because the catalog cannot connect the old number to the new one.

Wrong-part orders

A customer orders the old part number because the catalog still shows it as orderable. The order fails at the warehouse (the old part has no stock and cannot be replenished), or worse, the warehouse ships remaining old stock that is a previous revision. The return costs more than the sale.

Manual resolution time

Every phone call or chat where a sales rep has to look up “what replaced this part?” is time spent on resolution that the data model should have handled. For a distributor with thousands of superseded parts and dozens of sales reps, this is a measurable labor cost — minutes per call, multiplied by hundreds of calls per week.

Broken AI and search

AI-powered search, recommendation engines and chatbots rely on the catalog’s data model. If supersession is a free-text note that the model cannot parse, the AI cannot resolve a search for an old part number to the current one. The AI fails for the exact same reason the customer does: the relationship is not modeled.

Why it stays under-modeled

If the cost is clear and the data is available (in price lists, ERP notes, product specialists’ heads), why does supersession remain under-modeled?

No system owns it. ERP owns the item master. PIM owns enrichment. The catalog tool owns presentation. Supersession is a relationship between products, not an attribute of one product, so it falls between systems.

No process ingests it. Price-list processing extracts prices and availability. It does not extract supersession signals. Supplier onboarding creates new records. It does not link them to the old ones they replace.

The ROI is invisible. The cost of a missing supersession link is a search that returns nothing. Nobody measures “searches that should have returned a result but did not” because the absence of a result is silent. The ROI of modeling supersession is the revenue from those silent non-conversions — real, but hard to quantify without instrumentation.

The scope feels infinite. A distributor with 500,000 SKUs and 20 years of history may have tens of thousands of superseded parts. Building the full supersession table from scratch feels like a project with no end. The practical answer is to start with the parts that customers actually search for, and extend outward from there.

Where Claro fits

Claro makes supersession a first-class, sourced relationship. When a price list changes, Claro detects the candidates. When a product specialist confirms a replacement, Claro records it with the source, date and quantity ratio. When a customer searches for an old part, Claro resolves the chain to the current active part and shows the evidence behind each step. The relationship that was in a product specialist’s head is now in the data — and it survives their departure.

Glossary

What is supersession?

The definition: shapes, direction and what a supersession record needs.

Guide

Model supersession chains

Data model, chain resolution, splits, merges.

Playbook

Build a supersession table from price-list changes

Turn price-list diffs into candidate pairs.

Comparison

Supersession vs interchangeability vs equivalence

Who decides, direction, evidence and risk.

FAQ

Why is supersession under-modeled in industrial catalogs?

Because it arrives as a side effect of other data — a part number disappearing from a price list, a note in an ERP status field, a product specialist mentioning it on a phone call — rather than as a structured feed. Most systems store it as a flat old/new pair without direction, date, quantity ratio or source, if they store it at all.

What does under-modeled supersession cost?

Three things: lost sales (a search for the old part returns “not found” instead of the successor), wrong-part orders (a customer orders the old part because the catalog still shows it as current), and manual resolution time (a sales rep or product specialist spends minutes per call tracing part-number changes that a data model should have resolved automatically).