Amazon made to pay the price

To its cost, Amazon discovered that it is easy to get your fingers burned when evidence of price differentials are freely...

To its cost, Amazon discovered that it is easy to get your fingers burned when evidence of price differentials are freely available on your own Web site

In a remarkable but little noted exercise last month, Amazon was obliged to give rebates to several thousand disgruntled customers in the US.

Remarkably, these customers had received exactly what they'd ordered, on time, and at exactly the price they negotiated. So why weren't they happy e-business punters?

Well, as it turns out, Amazon had been caught abusing one of the great gifts that electronic businesses find at their disposal.

Earlier in the summer Amazon decided to try an experiment: it would price its discounts according to what crude customer information it had available. So according to how seasoned its customers were, where they lived, and what equipment they used, the company decided to charge different prices. It would throw its pricing model to the wall, and see what stuck.

Now market testing is an honourable tactic in bricks and mortar businesses, and not without legitimate precedence. The salesman always wants to know more about the mystique of the transaction. Who are you? Why are you buying this now and is your decision prompted by poor weather in Ohio or by a propensity of housewives to buy commodity X on a Wednesday afternoon?

The result was that Amazon punters could log in at random, and receive different pricing for the same DVD disc. What galled many long-standing customers, was that they could log in as a new user and receive substantial discounts compared to when using their true identities.

The savings weren't explained as discounts or offers - with the usual terms of qualification or expiry that apply to coupons - but simply, and baldly, presented as the going rate.

You can imagine what happened next. Faced with not one, but a range of "true" market prices, the punters demanded to know why they weren't getting the lowest. Which, when you think about it, is entirely reasonable. For in effect what Amazon was doing was the e-commerce equivalent of employing a store assistant to follow you around the shop, and marking the price up accordingly. It was caught playing God.

In traditional pricing trials, customer ignorance ensured that the buyers were ignorant of such manipulations. Mail order catalogue sellers could print different editions, with different pricing schemes for each targeted post code, and measure the response.

And as Hal Varian, head economist at the University of California Berkeley, and co-author the seminal e-commerce cookbook Information Rules told me: "I wouldn't call on my neighbour just to check their mail-order catalogue prices".

However, just as the Internet removes temporal and geographical difference among sellers, it can remove such price ignorance among sellers. In this case, the word rapidly spread around DVD buyer forums on the Internet.

That might not be the case with your customers. But given the arsenal of customer data mining tools being proferred by vendors, it's easy to forget that trust, not technology, is the basis of enduring relationships.

And if you're B2B or B2C, the fact this poster-child of electronic commerce was humbled into making such a concession should be a warning that some alluring temptations of e-commerce ought to be resisted.

No matter what your marketing whizz-kids tell you. Don't scratch that itch.

Andrew Orlowski is a San Francisco-based editor at The Register, an online IT newspaper

Read more on Business applications