How to Automate Supplier Quote Collection and Comparison

Why supplier quotes take so long to collect

The delay rarely sits with the supplier. It sits in the gap between sending the request and holding a set of answers you can compare. A buyer writes an email, attaches a spreadsheet, and waits. One supplier replies with a PDF, another replies with a phone call, and a third answers a slightly different question from the one you asked. Someone then copies every number into a comparison sheet, chases the missing lines, and asks again for a revised price that should have been in the first reply.

The cost of that work is easier to estimate than most owners expect. Hobson and Company's ROI research, as reported by TradeCentric, describes the end-to-end manual purchase order process as taking roughly 8 to 12 minutes per order. That figure covers purchase orders rather than quote requests, so treat it as a lower bound for the re-keying alone. Quote collection adds the chasing, the reminders and the reconciliation on top.

Published cost-per-order benchmarks disagree with each other. APQC procurement benchmarking has been quoted at $35.88 to $506.52 per purchase order in an older 2006 report, which a certification consultancy repeats in its own write-up. A more recent APQC range, as TradeCentric reports it, runs from about $14 to more than $54 per order, though that source does not state a year. The wide spread matters more than any single number. TradeCentric notes that APQC attributes most of the gap to how the work is structured, not to the industry. Use these figures as context, not as a budget line.

A better number is the one from your own business. Take three inputs: how many quote requests you send in a month, how many minutes each one takes from first email to a comparable row in your sheet (including chasing), and the fully loaded hourly cost of the person doing it. Thirty requests a month at 20 minutes each is 10 hours of work. Multiply that by your real cost and you have a figure you can defend in a budget meeting.

Slow collection also has a price that does not show up in the timesheet. A quote that arrives after the decision is made is wasted. A supplier who was never asked about delivery terms can make a cheap quote look better than it is. Buyers who wait for everyone to reply often accept the first complete quote instead of the best one.

Manual and automated quote collection compared

How to automate it, step by step

You do not need a full procurement suite to begin. Many teams run this process on a form, a sheet and an automation tool. The order below is the one we would follow, and each step can be built and tested before the next one starts.

1. Write one standard request. Turn the specification into fixed fields: item description, quantity, unit, required delivery date, delivery location, quote validity period, currency, delivery terms, payment terms, and any certificates the item must carry. Keep a free-text notes box at the end. The fixed fields are what make later comparison possible.
2. Choose the channel for each supplier. Email is the default and works for almost everyone. For suppliers who mainly work in chat, WhatsApp Business Platform can carry the same request through an approved provider. Billing has changed: from 1 October 2026, Meta charges for service replies beyond 1,000 free messages per business phone number each month. Utility template messages are charged per delivered message at rates that depend on the recipient's country, so check the rate for each country you contact.
3. Collect replies in a fixed format. Use a form in Microsoft Forms, Google Forms or a supplier portal inside an RFQ tool. Make the price, currency, validity and delivery fields mandatory. For suppliers who insist on sending their own PDF, attach a template they can fill in and ask them to use it next time.
4. Route every reply into one place. An automation tool such as Microsoft Power Automate, Zapier or Make can watch a shared intake mailbox or a form. It saves each attachment to a folder named by request number and creates one row per supplier per line item. Pick the tool your team already uses, because switching later costs more than the first setup.
5. Read the PDFs and images. For documents that look like invoices, Microsoft's Azure Document Intelligence prebuilt invoice model extracts fields such as vendor name, totals, tax and line items. It is generally available as version 4.0 and supports invoices in 27 languages. Microsoft's documentation lists invoices, utility bills, sales orders and purchase orders as its supported types, so a supplier quote with an unusual layout may need a custom model. A language model can help turn a free-text reply into fields, but every extracted value should be checked against the source document.
6. Build the comparison sheet. Use one row per supplier per line item. Columns should include unit price, currency, tax treatment, delivery days, validity date and a completeness flag. Add a formula for landed cost so that a cheap unit price with high delivery charges does not win by default. If your business already runs an ERP such as Odoo, ERPNext, Microsoft Dynamics 365 or SAP, the comparison can live there. Do not add a second system only for this step.
7. Set the rules for flags and reminders. A quote that misses a line, expires before the decision date, falls outside a price band you set, or gets no reply after a fixed number of hours should trigger a reminder or a message to the buyer. Let the tool send the reminders on a schedule. Cap the number of reminders so that suppliers are not pushed into silence.
8. Keep the award with a person. The sheet ranks the offers, and the buyer decides. Record the reason for the award in the same row so that the next buyer can see why a supplier was chosen. Review the flagged queue once a week for the first month.

What breaks, and how to prevent it

The most common failure is a supplier who replies in a shape the system did not expect. A quote with prices per pallet when the form asks for prices per unit will look valid and will be wrong. Prevent this by stating the unit in the request, making the field mandatory and flagging any reply that does not match the template, rather than letting the tool guess.

The most expensive failure is a misread number. A decimal point in the wrong place, or a currency symbol read as a different currency, can change a ranking. Prevent it with arithmetic checks. Line quantity multiplied by unit price should equal the line total, and the line totals should add up to the quote total. Any mismatch goes to a person. Set a value threshold above which every quote is checked by hand before it enters the comparison.

Quotes also go missing. They land in a spam folder, arrive on a former employee's address, or come through a chat channel where the message template was never approved. Use one intake address or form for every request, and check the queue of unmatched items every week. A missing quote that nobody noticed is worse than a late one that someone chased.

Comparisons can mislead when the quotes are not like for like. One supplier may exclude delivery, another may quote a price valid for seven days, and a third may charge tax differently. Fix delivery terms and validity in the specification, then compare on landed cost per row. If the terms still differ, the sheet should say so rather than ranking the quotes as if they matched.

Supplier price lists are commercially sensitive. Limit who can open the comparison sheet, and make sure the tools you use have data terms that suit your country and your supplier contracts. Ask your tool provider where the data is stored and who can access it.

Finally, automation cannot do everything. It cannot tell you whether a supplier will deliver on time, notice that a low price leaves out installation or commissioning, negotiate a better offer, or handle a dispute with a supplier you have worked with for years. It moves and checks the data you have defined. The judgement stays with the buyer.

When not to automate

Some situations do not justify the setup. If you ask for quotes only a few times a year, a standard request template and a calendar reminder will do the job. If every job is custom engineering with a new specification each time, a form cannot capture the work, and the saving will be small. If the award depends mainly on relationships and negotiation rather than on price and terms, automation will speed up the paperwork but not the decision.

Another poor fit is a supplier base that answers only by phone and will not put numbers in writing. The first job there is to get suppliers to use one format, which is a commercial conversation rather than a software project. Finally, if your items have no agreed codes or units, clean that up first. Automation copies whatever mess is in the data, only faster.

What it typically takes

The cost of this work is mostly time from the person who knows the buying process. Pricing for the tools varies, and the time is the more reliable thing to plan for. We did not find a published benchmark for setup hours, so the following is our working estimate rather than a sourced figure. A first version with one request template, one intake form, one comparison sheet and a set of reminder rules usually takes a few weeks of part-time work to build. Testing with two or three real requests takes a few sessions. Expect to spend about a month watching the flagged queue and tuning the rules before you trust the output without a second look.

A sensible starting point is to take ten quote requests from the last quarter and record, for each one, the hours from first email to a finished comparison. That gives you a baseline. After the first month of use, measure the same thing again and compare. If the numbers do not move, the rules are wrong or the process is too irregular for this approach.