Skip to content

VendorIQ

AI assistant

How can we help?

See how VendorIQ fits your procurement workflow.

Procurement research

More RFQs, lower prices? What research says about procurement savings

Research on bid counts and prices: evidence from Denmark, Finland and B2B purchasing, study limitations, and a practical way to measure savings.

One more comparable quote can change the outcome of a purchase. How much it helps depends on the market, specification, delivery terms and baseline. Research gives buyers good reasons to develop competition among suppliers. It does not justify promising that every additional RFQ email will reduce the price by a fixed percentage.

This article brings together empirical research and auction theory, then applies their findings to working with bills of materials, email and quotes for industrial components. We distinguish the researchers’ findings, our operational suggestions and illustrative calculations. All purchasing examples, businesses and amounts in this article are illustrative; they do not represent VendorIQ customer results.

First, agree on what you are counting

“We sent ten RFQs” can describe very different situations. One company sent the same specification to ten relevant distributors. Another found ten email addresses, three of which belong to the same organisation, four concern different product categories, and the remaining addresses do not accept purchasing enquiries. The message count is identical, but the prospects of obtaining a useful quote differ.

In everyday purchasing, distinguish between:

  1. Invited suppliers — separate businesses approached with the RFQ.
  2. Replies — including declines, specification questions and acknowledgements.
  3. Quotes — responses containing prices and fulfilment terms.
  4. Comparable quotes — proposals covering the required items, quantities and acceptable terms.
  5. Available purchasing decisions — options the buyer can actually accept within the required timeframe.

This distinction has support in the literature. Ido Millet, Diane Parente, John Fizel and Ray Venkataraman analysed thousands of procurement auctions conducted by a large multinational company. They described supplier participation as a multistage process: invitation, acceptance, logging in and active bidding. They highlighted participants who appeared in the system but never submitted a bid. A supplier’s presence therefore does not necessarily amount to active competition. Millet et al., “Metrics for Managing Online Procurement Auctions”, 2004.

We suggest applying similar measurement logic to email RFQs. A “we are checking” response helps track progress. A priced quote for a compliant item expands the set of prices that can support a decision.

What do 2,223 Danish tenders tell us?

Kasper Bek Aagaard and Jesper Gregers Linaa analysed 2,223 tenders from 2015–2022. They compared the contract price with the estimated value before bidding. In the model allowing separate results for different bid counts, two bids were associated with a price 4.6% lower, four with a price 9.7% lower, and six or more with a price 12.8% lower, relative to tenders receiving one bid. Aagaard and Linaa, “The Impact of Competition for Public Contracts on Public Finances”, 2024, figure 5.

Every percentage uses the same baseline: one bid. Do not add these values together or assign them successively to the second, fourth and sixth RFQ. They are estimates for groups of tenders covering goods, services and works.

The authors discuss the challenge of identifying causality: a contract’s type and attractiveness can affect both participation and price. Their analysis accounts for several contract characteristics, but it is not a randomised experiment. It also does not measure the full administrative cost of the process or the quality actually delivered. Methods and limitations, sections 3–5.

Denmark: bid count and contract price

Estimated price reductions relative to one bid: 2: 4.6%; 3: 6.7%; 4: 9.7%; 5: 11.7%; 6+: 12.8%. Whiskers: approximate 95% confidence intervals.
Price reductions relative to one bid, as reported in figure 5, model 2. Goods, services and works; 2,223 tenders. The whiskers are approximate 95% confidence intervals reconstructed from Appendix 2. Source: KFST (2024), figure 5 and Appendix 2.
Data and calculation method

Intervals use coefficient ± 1.96 × robust standard error, on the percentage-approximation scale used in the report. They are not an exact redrawing of the original confidence intervals. Every group is compared with one bid; the values are not cumulative savings.

All differences are relative to one submitted bid.
BidsLower priceApprox. 95% CI
24.6%1.1–8.1%
36.7%3.2–10.2%
49.7%5.8–13.6%
511.7%7.8–15.6%
6+12.8%8.3–17.3%

Finland: why the second bid deserves attention

In a 2026 KKV working paper, Oskari Ijäs examines Finnish procurement from 2017–2025. The main sample contains 18,251 tenders. The author matches similar contracts to reduce differences between the groups being compared. In the main analysis, moving from one to two bidders is associated with a winning bid approximately 9% lower. Ijäs, “Are We Paying Too Much? The Cost of Low Competition in Public Procurement”, KKV, Working Papers 1/2026.

The separate analysis of products is particularly relevant to component buyers. In table A.4, the matching estimate uses 5,344 observations. The coefficients for two and three bidders are −0.082 and −0.132 respectively. Converting from logarithms gives approximately 7.9% and 12.4%, relative to one bid. These are product-sector results, which differ from the estimate for the full market. Table A.4, “Products”, 2 NNM column.

This is a competition authority working paper. Matching uses observed characteristics and does not automatically eliminate every difference between contracts. The results describe Finnish public procurement; applying them to purchases at a particular plant requires a separate check.

Finland: prices in product procurement

Estimated price reductions relative to one bid: 2: 7.9%; 3: 12.4%; 4: 10.2%; 5: 17.8%; 6: 13.1%; 7+: 12.2%. Whiskers: approximate 95% confidence intervals.
Product procurement only; matching estimate (NNM), 5,344 observations. Baseline: one bid. Whiskers show approximate 95% confidence intervals. The groups contain different contracts: the fifth and seventh bids do not establish an optimal bid count. Source: Ijäs (2026), KKV, table A.4, Products / NNM.
Data and calculation method

Reduction = 100 × (1 − exp(β)). Interval endpoints transform β ± 1.96 × SE using the same formula. The intervals use the rounded coefficients and standard errors published in table A.4. They describe uncertainty relative to one bid, not significance of differences between adjacent groups.

All differences are relative to one submitted bid.
BidsLower priceApprox. 95% CI
27.9%5.3–10.4%
312.4%9.4–15.2%
410.2%6.1–14.2%
517.8%11.8–23.4%
613.1%1.8–23.0%
7+12.2%3.9–19.8%

Does the evidence extend to private B2B purchasing?

Public procurement studies are useful because they make it possible to observe many tenders and their prices. Industrial buyers also work with individual discounts, framework agreements, stock availability and continuing supplier relationships. Research on private companies therefore provides another useful perspective.

Oktay Karabağ and Barış Tan studied data from an organisation running electronic auctions for corporate customers. The number of participating suppliers, auction format and product category were significantly associated with the price reduction achieved. The authors measured that reduction against the lowest initial bid, making the baseline explicit. Karabağ and Tan, “An Empirical Analysis of the Main Drivers Affecting the Buyer Surplus in E-Auctions”, published online in 2018.

This finding supports investigating competition in B2B purchasing as well. It does not establish a universal discount for every additional email recipient. An auction with successive pricing rounds and a conventional email RFQ follow different processes, so their outcomes should not be treated as interchangeable.

Earlier research by Andrés Gómez-Lobo and Stefan Szymanski found a similar direction of association for British refuse collection contracts: more bids were associated with a lower service cost. This provides evidence from another market, whose characteristics differ from industrial automation supplies. “A Law of Large Numbers: Bidding and Compulsory Competitive Tendering for Refuse Collection Contracts”, 2001.

Why might another quote help?

It is useful to translate the research into mechanisms a buyer can examine in a specific purchasing decision. Another supplier may have a lower acquisition cost, a more favourable stock position, a shorter delivery route or a different margin policy. Reaching that supplier creates an opportunity to discover a price the existing contact list did not reveal.

Competition provides another mechanism: a supplier expecting a credible comparison has a reason to submit a proposal capable of winning the order. This requires a real possibility of purchasing from another business. An RFQ sent only to collect paperwork may leave both the decision and commercial terms unchanged.

A well-known theoretical result by Jeremy Bulow and Paul Klemperer illustrates the value of another participant: under specified assumptions, a selling auction with one additional bidder can generate more revenue than an optimal negotiation with a smaller group. This result concerns a selling model and its assumptions. It is not a direct estimate of a purchasing department’s savings, nor does it demonstrate that procurement negotiations are unnecessary. Bulow and Klemperer, “Auctions Versus Negotiations”, “American Economic Review”, 1996.

The operational implication is more modest: before another negotiation round, check whether the current quotes represent the available market. Finding a relevant second source may sometimes be more valuable than spending a long time negotiating a single price.

Why might a larger recipient list be insufficient?

Dakshina De Silva, Thomas Jeitschko and Georgia Kosmopoulou studied supplier entry when firms do not know how many rivals they face. Their model shows that a larger number of potential participants can, under certain conditions, even lead to higher procurement costs. The cost of preparing a bid and the decision whether to participate matter. The authors relate their analysis to construction contracts. “Entry and Bidding in Common and Private Value Auctions with an Unknown Number of Rivals”, 2009.

This does not support a simple recommendation to invite fewer companies. It shows why the potential supplier pool and actual participation should be measured separately. In technical purchasing, we suggest starting with the questions a supplier needs answered before preparing a quote:

If a controller’s part number is incomplete, ten more emails may produce ten questions about the same missing parameter. Completing the information may then be more useful than widening distribution.

Compare quotes for the same purchasing requirement

Counting attachments received for a bill of materials is not precise enough. Five suppliers may return five spreadsheets while the most expensive line still has only one quote. We therefore suggest measuring competition at the level of the items and quantities that can actually be ordered.

Consider an illustrative list with 20 lines. Eight suppliers receive the RFQ, five reply, and three provide prices. Two comparable quotes are obtained for six lines. Competitive quote coverage is therefore 6 / 20 = 30% of lines. Saying “we have three quotes” conceals that gap.

It is also useful to measure coverage by value, using a previously agreed reference price. The situation differs substantially depending on whether the six competitively quoted lines account for 80% or 5% of the basket’s value. Describe the reference price and apply it consistently, including to lines that receive no new quote.

Comparability requires checking the part number, manufacturer, variant, unit, quantity, currency, freight and delivery date. A proposed substitute needs a separate technical decision. A saving calculated using a product that cannot be used in the project is not an available saving on that purchase.

How do you calculate the actual price difference?

Consider an illustrative order for ten identical sensors that meet the same specification and required delivery date. All amounts below exclude tax; we omit taxes and costs that are identical across this example.

Unit price and the cost of the whole order

Ten identical sensors: A, PLN 2,100; B, PLN 1,960 plus PLN 60 delivery, total PLN 2,020; C, PLN 1,990.
Illustrative calculation, excluding tax; not research data or a customer result. B has the lowest unit price (PLN 196), while C has the lowest order total (PLN 1,990). The horizontal axis starts at zero; order totals include delivery.
Illustrative calculation, not study data or a customer result. All amounts are in PLN.
SupplierUnit priceDelivery per orderTotal cost
A21002,100
B196602,020
C19901,990

If the baseline is supplier A’s current quote, selecting C offers a potential difference of PLN 110, or approximately 5.24%. Supplier B has the lowest unit price, but C is cheaper once delivery is included. This is a straightforward reason to compare the cost of the entire required order.

If the company already had B’s quote before extending the RFQ, however, reaching C produces a benefit of PLN 30, or approximately 1.49% of PLN 2,020. The same final purchase can therefore have different correctly calculated savings, depending on the question being asked. Record the baseline before reporting the result.

Where payment terms, customs duties, commissioning costs or deliveries from several locations differ, extend the comparison to include material cost components. Start with the elements that actually affect the decision. For each, identify the document or assumption needed to reproduce the calculation.

Quote differences, realised savings and recovered time

A procurement report should distinguish three outcomes:

Supplier C’s illustrative PLN 110 advantage remains a quote difference until the order is fulfilled on those terms. Similarly, saving 30 minutes of price transcription creates capacity to do other work. It becomes a reduction in cash expenditure only when it actually lowers costs, such as paid overtime.

When comparing the economic value of a process, a valuation of working time can be included. State the rate used and avoid double counting the same benefit. An hour freed up to find a better quote and a lower purchase price are two observed outcomes of the process; they do not automatically represent two independent cash inflows.

How many quotes should a buyer collect?

The studies discussed here do not establish one correct number for every purchase. We suggest linking effort to value, risk, urgency and the availability of alternatives. These are operational rules to test, rather than thresholds derived from the publications.

For recurring purchases of standard components, check whether the company routinely relies on one source despite an available market. For low-value purchases, the work of handling another quote may consume its advantage. Where a part is blocking machine commissioning, securing a confirmed delivery may matter more than reducing the price further.

A single source may be justified by technical compatibility, warranty requirements or exclusivity. Record that situation. Judging the purchasing department against an arbitrary quote-count target without considering these constraints encourages collecting documents that do not improve the decision.

Obtaining the next useful quote takes preparation, contact, clarification and review. Weigh its potential benefit against the risk of delay as well. The assessment can be simple: does another attempt have a realistic chance of producing an acceptable alternative worth the time spent?

How can you test this in your own purchasing?

Rather than applying a published percentage to an annual budget, we suggest a short pilot with a scope agreed in advance. It can cover one category of recurring purchases where alternative suppliers can be found without changing the required parts.

  1. Define the scope. Record the category, period, eligibility criteria and exceptions, such as breakdowns or manufacturer exclusivity.
  2. Fix the baseline. Select a current quote or another justified reference price, including quantity and terms.
  3. Keep the specification consistent. Give suppliers the same information and record subsequent scope changes.
  4. Measure the stages. Count invited businesses, replies and comparable quotes separately for each BOM line.
  5. Measure the work. Record time spent preparing RFQs, clarifying requirements and comparing responses. Elapsed time between an email and a reply is not the buyer’s working time.
  6. Check fulfilment. Reconcile the quote with the order, invoice, freight and actual delivery date.
  7. Include the whole scope. Retain purchases without a price improvement and those where no additional quote was obtained.

After the pilot, assess whether market coverage improved, what price difference was achieved and how much work it required. A simple before-and-after comparison still does not isolate exchange rates, seasonality, order quantities or manufacturer price changes. If the organisation needs a stronger causal conclusion, plan a comparable group of purchases and seek advice on the evaluation method.

What role can VendorIQ play?

The practical obstacle may be the effort of handling additional quotes: matching responses to line items, reconciling different spreadsheets, finding missing prices and returning to the inbox. VendorIQ supports work from the bill of materials through RFQs and quote comparison to a recommendation presented to the buyer. Read more about supplier quote comparison and RFQ automation.

Test the tool’s value in your own process: is another compliant quote easier to obtain and assess, does comparison take less time, and can the basis for the decision be reconstructed? AI should not independently change BOM parts or accept a proposed substitute. The purchasing decision remains with the person responsible.

Research provides a reason to review available alternatives systematically. The result of a particular purchase should be established from its quotes and fulfilment. The VendorIQ calculator lets you examine a scenario; compare its estimate with the documented outcome of your pilot afterwards.

Sources and how they were used

These publications were checked on 6 November 2026. We use the full KFST and KKV reports and publication descriptions supplied by publishers and academic institutions. We identify the different types of evidence in the text; this is not a systematic review of all literature or a meta-analysis.

Explore your workflow

Let’s talk about your RFQs.

About 30 minutes: your current process, a VendorIQ example and the scope of a possible pilot.

Request a presentation