Why new isn’t always better: Learning from Marks & Spencer’s online sales crash

If you knew in advance that your new web site would reduce sales when you turned it on, would you still insist it went live?

UK retailer Marks & Spencer’s (M&S) online sales dropped by 8% after they switched over to their new web design.  The M&S share price fell – Shareholders were furious.

(Marks & Spencer Shareholders Attack Online Sales Fall)

Oops… They should’ve tested first.

But Marks & Sparks didn’t think they needed to test.  Like most companies building a new website, they assumed that new would be better.  In fact, M&S still don’t see how the sales crash could have been avoided.

Here’s what the M&S finance director said:

“We already flagged this earlier in the year that the settling-in period for the new site will impact sales…It is a bit like going to the supermarket for milk, they’ve moved it and you can’t find it immediately.”

I don’t buy this.  And neither should you!  When a supermarket changes its merchandise layout, they always test proposed changes in a pilot store first.  If M&S had a decent testing plan in place, they would have tested the new site against the old one – and caught this sales nightmare before it hit the bottom line.

M&S pic

The new design (they call it “magazine format”) looks fantastic.  But it’s sales that matter.  Apart from the change towards a harder-to-navigate layout, there are many changes that may have contributed to this customer abandonment.  And all could have been tested individually if anyone had cared.  Some changes that should have been tested are:

– Scrolling panes (which usually fail in true A/B tests, although designers love them)
– Reduced navigation options
– Fewer calls to action
– Reduced range of products and less visible content above the fold

And so on…

And by testing I don’t mean some quick usability study that gives you a “feel” for how people use your site.  I mean true A/B testing; using live sales data and statistically significant results that will indisputably tell you which option will result in higher sales.  The most important thing to remember about site testing is that old and new versions should be tested simultaneously to make sure the changes actually do improve sales.

Was the M&S sales disaster due to the new web design or perhaps some other market condition? Unless the old and new sites were run concurrently, through the same market conditions, they’ll never know.  Say you sell air-conditioners on-line.  The day after you launch your new-look site, there’s a month-long heat wave.  Will you give your new web design the credit for the increase in sales?

There are many tools you can use to test sites in-house, and they’re already available in most content management systems.  Last week I spoke with 7 firms planning new websites.  None had any pre-testing planned.  And three told me they would start testing sometime but sequentially, after the new site goes live.  A recipe for disaster!

Please don’t change over to your new site without TESTING it against your current site first. 

You may have paid a designer a packet to come up with the sexy new look, but that doesn’t mean it will sell more for you.  Proper A/B testing is an easy way to avoid the pain Marks and Sparks are feeling right now.







Leave a Reply

Your email address will not be published. Required fields are marked *