Matching customer enquiries to your product catalogue

The problem

Distributors and traders with a large catalogue receive customer lists in every shape. One customer sends a clean spreadsheet. Another sends a scanned PDF. A third pastes a list into an email using their own shorthand, such as "LED batten 20W", "M8 bolts zinc", "carton 5-ply", "brake pad set" or "toner cartridge".

An experienced salesperson knows what these mean and which catalogue item fits. But a list can run to hundreds of lines, and the work of reading, searching and pricing takes hours. The people who do it well are also the people you want talking to customers. Our team has built systems for this kind of work, and the central lesson is that matching quality, not reading the file, is the hard part.

How it works

1. Ingest the enquiry. The system reads PDF, spreadsheet and email text, including scanned tables, and pulls out each line with its quantity and unit.
2. Normalise the description. Each line is broken into attributes such as type, size, material, brand and standard, and unit differences are resolved.
3. Match to the catalogue. The system searches your catalogue using exact attributes and semantic search, then ranks candidate items with a confidence score.
4. Apply pricing rules. Prices come from your price lists and rules. Where an item is unavailable or a close alternative exists, the alternative is shown beside the main match.
5. Prepare the draft. A priced draft is generated, with matched lines accepted and uncertain lines flagged.
6. Review. A salesperson confirms the flagged lines. Their choices are stored and improve later matches.

Catalogue and price-list ingestion

Matching is only as good as the catalogue behind it. We build an ingestion step that loads your product master and supplier price lists, cleans inconsistent names, extracts attributes and keeps a history of price changes. Your price lists remain the source of pricing truth. The model never sets a price.

Inputs, AI action and output

Input	What the AI does	Output
Enquiry as PDF, spreadsheet or email	Extracts lines, quantities and units	Structured line list
Free-text description	Normalises attributes	Attribute set per line
Catalogue and price lists	Matches items and scores confidence	Ranked candidates and alternatives
Pricing rules	Applies price and discounts	Priced draft
Salesperson corrections	Learns preferred matches	Better suggestions next time

Integrations

The service runs on Google Cloud and connects to your ERP or accounting system for the product master, stock and customers, and to email for incoming enquiries. The review screen is a simple web app. It builds on our document AI and generative AI work.

Guardrails and human review

• Every match shows a confidence score and a short reason.
• Lines below your threshold are never auto-accepted. They go to the reviewer.
• Attributes you mark as critical can always require confirmation when they are missing.
• The system keeps an audit trail of each match and who approved it.

Data you need to start

• Your product master or catalogue as a spreadsheet, with codes, descriptions and units.
• Current price lists and discount rules.
• Thirty or more past customer lists, with the quote your team sent for each.
• A list of the abbreviations and synonyms your trade uses.

A hypothetical walkthrough

This is a generic illustration. A spreadsheet has a line reading "carton 5-ply 12x10x8, 500". The system reads the type, the layer count, the size and the quantity. Two catalogue items fit, and one is the customer's usual choice. It scores high and is accepted. A line reading just "toner black" has several possible matches, so it goes to the reviewer with the candidates listed and the missing detail named.

What changes for your team day to day

Salespeople spend less time searching the catalogue and more time on negotiation and follow-up. New joiners become useful sooner because the system shows candidates and reasons. The catalogue owner gains a new job: fixing the messy names the matcher keeps tripping over, which helps every other process.

Rollout plan

We begin with your catalogue and price lists and a batch of past customer lists with the quotes your team sent. We prototype matching and measure how often the top suggestion agrees with your salespeople. Then we add pricing rules and the review screen, run it alongside your team on live enquiries and tune thresholds. Only after that do we widen the set of customers and formats. After go-live, we maintain it under an AMC.

Who it is for

Distributors and traders with a large catalogue, many similar items and frequent multi-line enquiries, across electrical, hardware, packaging, spare parts and similar categories. If your customers always use your item codes, a simpler tool is enough. See the AI quotation automation guide or contact us with a sample enquiry.