Author: Alun Rafique – CEO & Co-founder, Market Dojo | Read time: Part 1 – 3 minutes / Part 2 – 10 minutes
This article, written in two parts, looks at whether any sized tender can potentially benefit from a reverse auction.
Part 1 gives a little background to why we are asking this question and Part 2 goes into more depth, with original data, to truly understand if the savings from reverse auctions can apply equally to small as well as large tenders.
Market Dojo started out in 2010 with a view to making reverse auctions more accessible to a wider audience. Since then we have grown, and developed a wider toolset in sourcing and supplier engagement with the belief that there is a better way to manage data, mitigate risk and minimise costs.
However, the reverse auction is still key to our offering, available truly on-demand for as little as £500. This hasn’t changed from when we created Market Dojo.
Figure 1: Example of Market Dojo’s Dashboard
Our first client was a company called Hamworthy Combustion and this initial customer really tested our core principles. I remember the Ops director coming into our meeting asking the team to tell him “how they can save money, not how they can’t save money”.
It was a very positive, albeit direct, approach which resulted in a buyer, Emma, embarking on an adventure. Emma had never run an electronic tender.
The Ops director wanted an auction by the end of the following week which could be run on the day of the board meeting. At the same time, a factory consumables tender had just been instigated on only £20k of value.
In only one week, Emma was to create the auction, obtain supplier engagement and run the auction in front of the procurement team, stakeholders and the board.
Both myself and Nick Drewe (one of my other co-founders) were slightly apprehensive. During our consultancy days we would normally only run auctions on tenders >£1m, although consultancy fees for these types of tender were north of £30k. (Nowadays we support client auctions, only if needed, for a few £ ‘000’s).
Nevertheless, we needn’t have worried.
The auction ran flawlessly with the three local suppliers bidding furiously and Hamworthy saved over 25% on the tender which in turn paid for an annual licence of our software.
We were all thrilled, a win for ourselves, Hamworthy and the supplier. Many more small tenders were also successfully run over the following months.
Figure 2: Screenshot of an auction with two suppliers battling for first position
This really opened our eyes to the potential of what we had started.
Now after 12 years, we have had a UWE student, Ludovico Bettoli, run a project to understand if the savings from reverse auctions can apply equally to small as well as large tenders.
Following the statistical analysis of more than 50,000 products and services purchased via reverse auctions, conducted by procurement professionals from multiple industries in 21 different countries, he concluded that there is no correlation between the savings generated through reverse auctions and the auction value.
In other words, procurement professionals running low-value tenders, (<$10,000) can equally benefit from the savings generated by the implementation of reverse auctions. And at £500 to run an auction with Market Dojo, the ROI is also clearly there.
Also importantly, he confirmed factors such as bidding participants and the use of Ranked Auctions (rather than a simple RFx) are positively correlated to higher savings.
Learn more with our eAuctions Analysis (download PDF) or read on for Part 2 where we’ll delve into some of the analysis used to reach this conclusion.
In collaboration with Market Dojo, under-graduate student from UWE (University of the West of England), Ludovico Bettoli, confirmed through statistical analysis that the size (value) of a tender doesn’t affect the potential savings.
Ludovico analysed and correlated more than 50,000 products and services purchased with reverse auctions conducted by procurement professionals from multiple industries in 21 different countries.
A correlation is a statistical measure that tells us about the association between the two variables.
It describes how one variable behaves if there is some change in the other variable. For instance, if the variables are increasing or decreasing in parallel, then they have a positive correlation between them. If the change of one variable has no effect on another variable, then they have a zero correlation between them.
The presence of correlation between variables is evaluated through statistical tests; in this scenario considering the data analysed, Spearman’s Correlation test has been used to measure the strength and direction of association between the two variables (Saving % and Auction Value $).
Once the test is performed, it will result in a Spearman’s rank coefficient, also called “rs”, which ranges from -1 to 1.
Following this scale is possible to estimate the strength of the correlation:
0.0<=rs=>0.2 “very weak”
0.8<=rs=>1.0 “very strong”
The statistical test output will also provide a “p –value” which corresponds to the level of confidence; the smaller the p-value, the stronger the evidence of correlation.
A p-value of <0.05 corresponds to >95% confidence, and it is considered the maximum p-value to evaluate significant evidence.
The presence and strength of a possible correlation can also be visualised through a Scatter Graph.
Image Source: CQE Academy – The Scatter Plot and Linear Regression
Ludovic was given access to anonymised data from Market Dojo. This covered over 50,000 lots and savings were calculated from the current value and best bid.
This was the first time Market Dojo had looked at such a large data set using such a rigorous methodology.
After the analysis, Ludovico concluded the following findings.
“Spearman’s Correlation test between the reverse auctions current value in USD and the Percentage of Saving % resulted in an rs =.003 and p=.852, highlighting that there is no correlation between saving and event value.”
The reasoning for this is quite simple. As long as the suppliers participating in an event see the potential outcome as an opportunity, you can find that any tender will be equally enticing to a supplier, whether it is large or small in value.
Ludovico also looked at the relationship between the auction result and the number of participants bidding.
“Spearman’s Correlation test between bidding participants and Percentage of Saving resulted in an rs =.463 and p=<.001, there is a moderate positive monotonic correlation between the # of bidding participants and saving.”
Therefore, reverse auctions with an increased number of participants are likely to generate more savings.
This was also seen in some earlier work carried out by Market Dojo, which saw in ranked auctions a good increase in savings with up to 10 suppliers and then the law of diminishing returns kicks in. The findings are summarised in our eAuctions Analysis (download PDF)
Essentially an auction is more successful with increased interaction from the suppliers. Thus more suppliers will generally mean a higher number of interactions, hence a better outcome, and a host will be more likely to get the market price.
Next, Ludovico looked at the auction length and the savings.
“Spearman’s Correlation test between auction length and Percentage of Saving resulted in a rs =.136 and p=<.001; there is a very weak positive monotonic correlation between auction length and saving.”
This also supported what Market Dojo had seen previously. Generally most of the bidding is at the end of an auction so the length is not very relevant even with suppliers across multiple time zones.
In previous research by Market Dojo, we realised that as long as there is a dynamic close, which gives suppliers the ability to counteroffer, the final result is not impacted by the original set auction length.
Where there is no dynamic close, the auction length can have an effect, generally negatively, on the savings percentage, compared to when the dynamic close is switched on. Ludovico’s research looked at all events together, both with and without the dynamic close.
Finally, Ludovico analysed the savings generated from the different types of auction.
A one-way non-parametric ANOVA (Kruskal-Wallis test) test was performed to statistically investigate the distribution between Japanese, Ranked and Open auctions percentage of saving; resulting in a p value <.05 which indicates that distribution are not the same, hence Japanese auctions resulted in a lower distributed percentage of saving.
Meanwhile, statistically speaking, the difference between % of savings generated by open and ranked auctions is not significant. Nevertheless, ranked auctions are by far the most favoured type of auction used as they ensure confidentiality of supplier’s sensitive information, so suppliers are generally more willing to participate.
The conclusion that ranked auctions give the best savings result is born out by most of our clients opting to use a ranked auction.
The added ability for all suppliers to interact until the end of the auction is a big factor.
Whereas in an Open auction, only a bid that beats the lead bid can be placed, hence less interaction, and a Japanese auction has no supplier interaction, although is very suitable if you have low supplier liquidity or a monopoly.
It was wonderful to have many of our understandings so thoroughly validated. Thank you Ludovico!