{{ searchResult.published_at | date:'d MMMM yyyy' }}

Loading ...
Loading ...

Enter a search term such as “mobile analytics” or browse our content using the filters above.


That’s not only a poor Scrabble score but we also couldn’t find any results matching “”.
Check your spelling or try broadening your search.


Sorry about this, there is a problem with our search at the moment.
Please try again later.

John Lewis says the new personalised recommendation tool on its website was a key factor in driving a 27.9% increase in sales over Christmas.

The tool, created by RichRelevance, provides customers with recommendations on fashion items by analysing shopping behaviour alongside the relationships between products and product categories.

John Lewis head of online delivery and customer experience Sean O’Connor said the tool helped increase sales in the five weeks to December 31 2011 beyond the usual spike expected during the Christmas period, and in comparison to the previous year.

When any shopper comes to our website, we want to provide them with the same personalised customer service we would if they visited us in one of our shops.”

He said that the recommendation and email personalisation platform delivered tangible results by offering customers relevant products.

Product recommendation works particularly well in the fashion category as it recognises shopper behaviour, patterns and recommends items of interest not only by product type, but by brand as well.

The tool is not integrated into social media so recommendations do not take into account what the customer's friends have bought or viewed – something John Lewis should possibly consider enabling as it has almost 317,000 Facebook fans.

However O’Connor said the tool and its recommendations are tuned in to "crowd shopping". 

This takes into account not only what the individual customer is doing on the site at that moment in time, but what other shoppers who are similar in product views have done before. 

We feel this is the very essence of social shopping: taking into account not just the individual customer's experience but those of the wisdom of the crowds."

At the moment the tool is only accessible via web, but O’Connor said the retailer was looking at introducing it to mobile.

We want our customers to have the same level of service and to provide a personalised shopping experience across all shopping channels.”

John Lewis has been quick to adopt mobile technology, launching a mobile optimised site in 2010 and trialling a virtual QR code store in Brighton in December.

It also introduced free Wi-Fi into its stores, a move which O’Connor says has been a big step forward in helping customers make an informed choice.

They can quickly and easily access our mobile optimised website, or use our iPhone app. Customers are free to access the whole of the web, including competitor sites to test our price commitment, however it primarily enables us to extend our John Lewis online content and services into our physical shops in a way that is convenient for them.”

David Moth

Published 3 February, 2012 by David Moth @ Econsultancy

David Moth is Editor and Head of Social at Econsultancy. You can follow him on Twitter or connect via Google+ and LinkedIn

1704 more posts from this author

Comments (13)

Save or Cancel

Rob B

I don't think that 27.9% of growth would be directly attributable to RichRelevance.

Firstly, ask how it's measured. Do RichRelevance say "we suggested this product, they clicked on it and then two weeks later they went and bought it"? Because that's not directly attributable, by any means. I'd hope they were attributing sales that were clicked on in the same session.

Secondly, sales growth would have happened if John Lewis has just populated these blocks with "best sellers" or similar.

Apart from that, there seem to be issue with the suggestions - for example suggesting the same product in a "customers who viewed this went on to buy" block is a waste of space - e.g. http://www.johnlewis.com/67423/Product.aspx

Suggesting womens slippers and coats on a men's product page is a waste of space too - e.g. http://www.johnlewis.com/23235/Product.aspx - if you click to view more suggestions, you see a toilet roll holder and a Cath Kidson purse - not relevant at all.

And also, other research published on Econsultancy has proved that cross-selling on product pages isn't that effective - I'd say that is backed up by the real examples above.

over 5 years ago

Matthew Curry

Matthew Curry, Head of Ecommerce at Lovehoney

I'm going to question this. Firstly, what was this increase factored against? Previous period, previous year, what? We all had a bumper fun time this December due to the lack of snow not screwing up customer confidence in receiving orders.

Secondly, of the 27.9% growth (which is terms of the growth some folks had year-on-year over December, is actually tracking behind), how much was from increases in visitors, conversion rate, and basket size & AOV? Recommendations can only affect the later 2, and indirectly the second by exposing substitute products with a higher propensity.

I wouldn't mind seeing a slightly more in-depth analysis of this, rather than attributing it all in a kinda sensationalist way.

over 5 years ago


Bhavin Shah, Web Analytics and Optimistation Manager at John Lewis Partnership

Hi Mattehew

Just in case you want to see what the orginal press release.



over 5 years ago

Vikki Chowney

Vikki Chowney, Head of Social at TMW

@Matt Indeed. Bumper fun time. 27% increase was in the five weeks to Dec 31st in comparison to the same period last year.

JL wouldn't be any more specific about the data, but was quite clear about the direct link between this tool and the results. We have to take their interpretation on board to a point.

@Rob You make a valid point about the attribution of this, and I think about that up-sell/cross-sell research almost daily. But, with different demographics come different behaviours. No study is ever de facto.

over 5 years ago

David Moth

David Moth, Editor & Head of Social at EconsultancyStaff

Hi both, the stat is a year-on-year increase in the five weeks up to December 31. I should have clarified this in the original article, so apologies for the confusion.

@ Rob, if you read the article it does say that the 27.9% increase wasn't entirely attributable to RichRelevance, but that it was a key factor. I asked John Lewis how much of this was actually just down to the usual boost in shopping around Christmas, and they said the recommendation tool was also a factor in the increase above the normal festive rush for gifts.

over 5 years ago


Nick Tsinonis

It's about time people took recommendations very seriously.

I've been looking at John Lewis and their site had ZERO personalisation last year, so actually utterly believable.

Amazon claim that 35% of sales come from recommendations.

We're also in the business of recommendations and personalisation and provide services for for smaller corporate customers but we are also seeing similar performance figures.

Personalisation, Recommendations, and Discovery create more activity, higher conversions and more loyalty for customers. Customers love personal assistance and discovery, so why not help them?

over 5 years ago

Matthew Curry

Matthew Curry, Head of Ecommerce at Lovehoney

Thanks for clarifying folks. I think the issue is that last December was so different to the previous years, any kind of direct comparison is going to be a bit screwy.

We run a permanent 10% control group on our recommendation tech, so only 90% of visitors get recommendations, and I can see exactly how much of any growth is attributable to it. Since these things are generally outsourced services, with monthly costs, there were be a demising value of returns vs a solution with a one-off development cost, as well variable effectiveness vs monthly shopping trends.

You have to be able to see, very specifically, how much value is being derived rather than implying causation.

over 5 years ago

Nick Tsinonis

Nick Tsinonis, CEO at RecSys Ltd

This is VERY believable seeing they did not have ANY personalisation on their site last year.

Amazon.com reckon they do about 35% of their sales through personalised recommendations onsite and on emails, so I can see how this could be the case on John Lewis if they are intelligently targeting users.

We see similar results with our implementations. Th results can vary depending on how well the client implements on different areas of their site as well as personalised emails.

Go see Amazon for how many ways they use Recommendation technology. They are masters at it.

over 5 years ago


James D

I wouldn't say John Lewis's recommendations are that personalised, or even that intelligent. Seems to just be powered by crowd-wisdom. Not sure how that falls in the 'personalisation' camp.

over 5 years ago


Ben H, Personal at Interflora

I wonder if John Lewis could possibly confirm how much of the 28% yr on yr growth was down to the huge delivery issues and impact on online sales in December 2010?
Releasing stats as part of an agreed press release is fairly common, but come context would aid the credibility of such stats.

over 5 years ago


Depesh Mandalia, Head of Digital Marketing at toucanBox

As Matt states its difficult without a fair 'base' from which to compare any uplift in a similar way to MVT. Even with a 10:90 split (personalised VS non-personalised) served in real-time, this isn't always pure, but much better than comparing YOY with and without personalisation. I'm sure JL will get to this level of detail if they haven't already

Press releases are notorious in leading the reader a particular way but I'm inclined to believe there was a % uplift as it states, perhaps somewhere between 0 and 27.9%? :-)

@Rob B there's always an element of testing and optimising to get the recommenders honed which is probably why JL are displaying women's shoes on the men's product page. There's the balance between the wisdom of the crowd (ie. women may be purchasing for themselves and their partners and vice versa which can get picked up as a trend) and common sense (ie show men's products in the men's section) which both the recommender system and the operator need to get right. This doesn't come out of the box.

over 5 years ago


* *, Product Strategy Director at Peerius Ltd.

@Depesh You are spot on in asserting that there needs to be a balance between the outputs of a recommender system and the specific application or strategy of an organisation where it is implemented. As a provider of a personalisation service [So slightly biased in my response :)] I know how important it is to have dedicated account managers and merchandising specialists to tailor each implementation on an ongoing basis with a retailer. Strategies, consumer behaviour and market conditions change all the time and while we can automate a lot of things to take these factors in account there needs to be some intervention to make sure that it is always optimised.

@James, I suggest you read "The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations" before making assertions that crowd based inputs have no place in recommendation or personalisation technologies. What is the value in personalising an experience solely on the past behaviour of one individual (Which is of course an important input) rather than combining it with cluster trends to enable personalisation which influences future behaviour?

I say well done to John Lewis for recognising the value of recommendations and personalisation, now let’s get some more detail as Matt suggested!

over 5 years ago


Tom Sheepshanks, -

@Daniel Being able to understand past behaviour so that you can infer insight and then predict future behaviour is invaluable. The problem comes form not having a holistic view of the user. The data sets being used (i.e. transaction history, on-site behaviour etc) are not enough to begin to serve relevant recommendations.

This is why I implore anyone reading this to understand the value in social data, which is only starting to be used for commercial gains. Imagine being able to contextualise every action a person has ever undertaken on a social network so that you understand not just WHAT someone may want to buy something but WHY they would like to.

Being able to merge a users social data with JL's transaction history means that you would know that they have recently started talking/liking/engaging with topics about fitness/health/weight. We would also know that they bought a pair of running shoes last week, but we can now INFER that they may well like other running equipment (leggings/headphones/hat/gloves etc) and so serve them a RELEVANT recommendation.

Or you could flip the whole process and create simple, yet sophisticated, advocacy programmes by incentivising users to recommend products to their friends online. The clever part being we already know which friends will find the product relevant.

Powerful stuff indeed.

over 4 years ago

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Daily Pulse newsletter. Each weekday, you ll receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.