What can be learnt from Amazon’s website price glitch – everything a penny?

15th December 2014

Amazon sellers experience the downside of dynamic pricing

Having just come from a conference call on a project where User Journey monitoring has proved invaluable for protecting against pricing problems, I saw the news about Friday’s 1p price issues on Amazon’s website. Where a number of Amazon sellers have suffered from a different flavour of the same pricing problem. Dynamic pricing needs handling with care!

The dangers of dynamic pricing were clearly illustrated when small businesses lost tens of thousands of pounds over the weekend, after an error in RepricerExpress software caused their products to be sold for a penny on the Amazon website. Many keen shoppers bagged a bargain and tweeted the deals.

The problem was compounded because sellers were not in control of warehousing. Most retailers state that there is no binding contract until an item has been shipped. So once realising the error, sellers contacted Amazon to cancel the orders. However due to Amazon’s automatic distribution, many orders had already been sent out. And to top it all Amazon continued to charge sellers their usual fees.

Amazon did manage to cancel many of the orders but many sellers lost tens of  thousands of pounds and fear they are facing bankruptcy.  The Londonderry based software house Repricer Express sorry for the Amazon 1p glitch.

The RepricerExpress software is designed to keep businesses competitive by checking that products are the cheapest available.  Clearly, software that is trying to be the cheapest, is an algorithm that is fragile:  I don’t know the exact root cause but can imagine that two stores both trying to be cheapest, could potentially trigger a race down to 1p, without safeguards around the algorithm, which raises first lesson to be learnt:

This isn’t the first and it won’t be the last automated pricing disaster

Earlier in 2014 it was DIY retailer ScrewFix who suffered brand embarrassment when a price glitch on their site priced everything at £34.99. Many high value products were ordered including sit-on lawn-mowers at thousands of pounds.

At least in ScrewFix’s case they were in control of their own warehouse and could mostly prevent goods being shipped. They may have lost some money on sales that were ‘click and collect’, when the store opened at 8am next day. But they certainly suffered bad PR, lost credibility on social media platforms and an increase in time spent managing the fallout; order cancellations, refunds and high call volumes.

Lessons to be learnt from algorithmic Dynamic Pricing

1 – Ensure your site monitoring is providing a pricing safety-net

To mitigate pricing risks, a number of our clients incorporate a range of checks into their monitoring to catch pricing errors. Some choose to run random journeys to check if a large number of prices are all the same. Others choose to run journeys to identify if prices are found to be below a certain value, or if a price changes beyond a specified percentage.  Any such pricing irregularities trigger repeat tests and alerts.

It’s good practise to double check your own site’s active insurance against price failures.

2 – Be careful with roll-out of personalised pricing

Several vendors push the advantage of charging individual shoppers different prices, based on their past shopping and browsing history. This is all good when customers get rewarded for loyalty and repeat business. But when it is used to increase prices for some shoppers – for example higher pricing for iPad users over Android – consumers can quickly become angry if they feel they have been singled out to pay more.

There was a good article in eConsultancy about the dangers of personalised pricing just recently: Online price personalisation: should we be worried?

3 – Be aware how much you depend on your 3rd parties

It’s very common in the online retail sector now to depend on 5-10 3rd parties. Problems with these suppliers can prevent or slow down users shopping on the site.

And with the addition of dynamic pricing tools, like the RepricerExpress software that caused the problem on Amazon – we shouldn’t be too surprised to see more pricing problems into the future.

The Amazon 1p glitch may have impacted smaller retailers who live on top of the Amazon engine. But it won’t be long before the next ScrewFix comes along: a big name retailer with a price problem that again does brand damage, beyond the direct costs of the mistake.