The benefits of weighted auctions

Readers of this blog will have seen a few posts about the impact of new public procurement rules on the use of auctions.

Over the past year I have seen some pretty interesting results and I want to share a couple of illustrative examples. On the one hand these auctions can deliver increased savings (sounds counter-intuitive, I know, but it’s true). On the other hand these auctions can let higher quality suppliers win, even if they can’t compete purely on price.

Generalising, suppliers are either high quality/high price or low quality/low price. In your common or garden online reverse auction the high quality/high price supplier can’t compete on price with the low quality/low price supplier. This means that the low quality/low price supplier doesn’t have to work so hard to get into first place.

In a multi attribute or weighted auction, the quality is fed into the auction mechanism to produce a comparator score – this compares the total value of each offer. The winner of the auction is the one with the best comparator score, not necessarily the one with the lowest price.

The high quality/high price supplier ends up with good overall value for money, even if their price is relatively high. So a lower quality/lower price supplier has to bid far more aggressively on price than they would have done in a price only auction.

Two real-world anecdotes I’ve seen recently highlight what can happen when you start running these auctions:

1. An auction where the winner of the auction was the supplier with the highest price: Best value for money overall but highest price. The winner of the auction had a price about 10% above the lowest price in the auction, but because of their higher quality and better delivery they were able to win on a total value for money basis.

2. An auction where the winner of the auction was the supplier with the lowest price, but they had to compete very aggressively to match the overall value for money of the higher priced suppliers. The savings generated through the auction were 18% but if non-price factors had not been included into the auction then the saving would only have been about 12%.

Caveat: If this all sounds too good to be true, then that is because this kind of auction is hard work to set up. Additionally it is not appropriate in all situations. For example, if I want a 2 day lead time and I really don’t care if you offer me a 1 day lead time instead, or if I want trainers with 2 years experience at least and I really don’t care if you offer me trainers with 5 years experience then building in weightings for these attributes would be a waste of time.

The Trouble With Dutch Auctions

I am not a big fan of reverse Dutch auctions on the internet, from either the buyer’s or the seller’s point of view.

Dutch auctions were initially invented and adopted to speed up the sale of flowers in Amsterdam. Potential flower buyers would sit in a room which contained a clock. The clock would initially show a very high price – much higher than any flower buyer would be prepared to pay. The price shown on the clock would then tick downwards until it reached a price that one buyer was happy with. The buyer would press a button to “stop the clock” and could buy as many lots of flowers as they wanted at the price shown on the clock. If more flowers remained to be sold then the clock would continue downwards until all the flowers had been sold.

Software providers have adapted this kind of auction to internet procurement auctions, but the practical implementation leads a lot to be desired.

In a reverse dutch auction the price starts very low and gradually increases until a supplier “stops the clock” and offers to supply at the price shown on the clock. Apparently, the faster the clock, the more excitement and the higher prices would be (i.e. better for the auctioneer).

Of course, on the internet, you can’t legislate for the speed of bidders’ internet connections. So all implementations of online reverse Dutch auction that I have seen allow the bidder to place their best bid before the auction starts and then have the system work out when to stop the clock. The system gradually ticks upwards and eventually reveals the winning supplier’s bid to the buyer.

In other words, they aren’t auctions at all. There is no competition, no unleashing of those “animal spirits”, no nervous energy. In short, nothing of what makes auctions the best tool for allocating goods and services in a market.

As if that isn’t bad enough, the buyer only ever gets to see one supplier’s bid – so if there are any quality or other issues with that supplier the buyer is not able to switch to another supplier.

In fact, I am aware of one major retailer which has been updating its Dutch auction software to reveal all the suppliers’ bids. In other words they are turning Dutch auction software into Sealed Bid software.

The moral of the story – think carefully before running a Dutch auction on the internet. You’ll be running a second-class sealed bid. 9 times out of 10 the real solution would be to use a Japanese auction …. but that is a different story post.