Guest Post by Euan Wilcox (Regional Managing Partner – The Upper Storey )
There is a lot of talk, and honestly a lot of contradictory information, about Social Commerce. No wonder some brands might approach the whole opportunity with some trepidation and caution. Fortunately for me, my agency The Upper Storey, and Dell, we jumped headlong into Group Buying at the end of 2008 way before the over-hype made the picture significantly more complex.
Some people might question if group buying is a significant part of social commerce. I would agree, I think its true relevance to the potential field is probably minor. But it certainly fits under the broader definition of social commerce and is one of the most talked about today. For me there is no point arguing definitions and I find the Wikipedia definition more than satisfactory: Social commerce is a subset of electronic commerce that involves using social media, online media that supports social interaction and user contributions, to assist in the online buying and selling of products and services.
So by this definition, and all logic, social commerce has been around for quite some time. Commenting and User Generated Reviews are adopted everywhere in online commerce. What is different now is the “FB-phenomenon” where every marketer, business manager and CEO seems obsessed with Facebook as the quasi entry point into the power of social media. In fairness, social media is helping to fill a void in a marketing environment coming apart at the seams through declining effectiveness of traditional mass media. Social at least holds a great promise for the future and a marker of how things will change in the new digital environment.
My issue is not so much with the fact social media, and Facebook, have joined the social commerce party – they have every right to be there – but my issue is that people are not really questioning what is said to be a fundamental truth of social media thinking; “I am more likely to buy something because my friend likes it”. At the surface this is fine. Yet as a significant part of my media budget, my marketing time and effort – it does deserve more thinking.
My questioning and thought process of this fundamental truth is greatly influenced by my experience on the Dell social commerce program; Dell Swarm. For me it has put into perspective why social commerce can work and in what ways it is social. This gives me a lot of clarity on where to take this program and others like in the future.
Dell Swarm background
Born from a group brainstorming session with media agency OMD and Intel, our inspiration for the program was the China phenomenon of Tuángòu, which we wanted to turn into an online social media idea for Intel OEM partner; Dell. My trepidation at the time was not a confusing array of conflicting opinions, there was virtually no chatter at all then, but more the worry that we had ‘seen this all before’. People seem to forget (or just were not around to see) that group buying sites Mercarta, MobShop and Letsbuyit.com were big hits of the late 90’s, spending hand over fist on branding and awareness. So my trepidation was more about the fact that this was already a failed business model, and thus I did my best to research what I thought made them fail and why this time things might be different. In the end we were satisfied that these group buying sites of the past failed primarily because they tried to be all things to all people. They carried a wide range of products with no focus on a specific audience. So despite the hyped awareness, they had no critical mass to carry off a deal. Additionally the fact that ten years later there is greater sophistication in web technologies, including large scale social sites like Facebook, gave this idea a better fighting chance. So Dell Swarm was born, roughly the same time as Groupon, and two and a half years later we have our own data about what has worked and what has not.
Dell Swarm has run fully in three markets; Singapore, Canada and Australia. It has gone through at least 5 different stages of evolvement and development. The business model, product mix and discount model has evolved and developed as well. This makes it, in many ways, an excellent platform to draw learning’s from.
The most significant and obvious evolution was the change in product mix. Originally, in Singapore, the products were all laptops, many focused on the higher-end Intel based processors. In Australia however the weekly swarms mainly comprised of peripherals and accessories. Logically this was better from the audience POV, as there were more lower-priced, low involvement items to purchase. This model also allowed for steeper partner-funded discounts more in line with consumer expectations. Rightly or wrongly consumers expect large discounts on group buying. This may in part be their experience of services like Groupon where ‘up to 80%’ might be possible. But there is a very big difference between a Chicken Burger at Kenny Rogers or a massage at the local beauty salon to the consumer electronics sold on Dell Swarm. PC’s simply don’t have the margins people imagine them to have and so Dell cannot offer enormous discounts over the long term. Offering other products allowed for more flexibility in discounting and helped build a great critical mass of people generally interested in everything offered. This included software, printers, cameras, TV’s and of course some laptops.
The evolution also included the mechanics of how we discount products and the interface to participate in the offer. Like Amazon did for eCommerce in the late 90’s, Groupon has now done for group buying. There is a standard way to communicate that a deal is on, and by adopting that standard, we helped significantly improve the 1st time participation rate by new users. First time visitors grew to almost 60% of sales in Australia up from around 30% in Singapore, where the average user had to come back 2.1 times to eventually participate in the deal (the discount and product mix obviously made a significant difference too).
By all accounts the Australian Swarm was a business success at Dell. Selling out Printers in minutes, selling higher value items like TV’s in hours. We also gathered a lot of data about people’s future buying intentions and ‘wish list’. We offered some swarms to help satisfy their desire, and will be taking this part of the functionality even further in future iterations.
There is a lot more we can do – and will be doing as this gets more attention at Dell. Some improvements are really quite obvious as the service eventually scales up. They are all in development. But what surprised me originally was the fact that as Swarm became more popular from a sales point of view, with even more attractive offers, the less social traffic and less monitored social conversations were happening.
Declining traffic from social sources is not confined to Swarm. Data from Hitwise shows that upstream traffic from social sources to Groupon has declined from roughly 40%-50% in December ’09 to 8.31% in Dec 2010. Famously Groupon’s traffic and sales have exploded in that time period.
Dell Swarm behavior
My data from Dell Swarm troubled me even further. The statistics showed not only was social traffic on the decline, but the quality of social traffic was appalling. In the end there were two quite distinct groups: Buyers & Socialites. And by Socialites I mostly mean the catchall-simplified-pseudonym for social: Facebook/Twitter.
Buyers were easy to identify. They primarily came from sites focused on PC purchase and discussion. This include Gizmodo and Ozbargain. Sometimes it was activated with affiliate marketing but not always. These people were vivacious in both their buying habits and likelihood to return to the site. A person from Ozbargain was 104% more likely to buy and 41% more likely to return to the site than the average person.
The polarized opposites were the Socialites. Facebook (non-paid) traffic was 70.5% less likely to buy, and 72.3% less likely to visit again. Traffic from Twitter, which was only user generated sharing, had incredibly low loyalty. 96% of traffic leaving the site in less than 1 minute and only a minor percentage of return visitors. Facebook Ads did not fare much better. They had 158% higher bounce rate than average and spent on average 53% less time on the site.
Somewhere in the middle, across all sites, were the like-minded communities Dell has across the world particularly in Canada. The Dell Community for example. They spent roughly twice as long on the site, sales were negligible (it’s a low base), and repeat traffic was very low. They found it interesting, but they were not a great source of potential buyers.
In fairness to Facebook, there is some mixed data. In Canada, where they have 97,000 Dell fans, data was roughly the same as the Dell Community. They were sort of interested, but they don’t come back and their low numbers weren’t significant in total sales. Twitter is must be said can be used in other ways to help attract the right audience. This is planned for the future, but was not possible due to manpower restrictions. Twitter was actually a great traffic driver and large buzz initiator in Singapore. But these people were great talkers, mainly loving the idea, not buying the products.
Dell Swarm is social
Despite some gloom, for the social media evangelist within me there is some hope, a savior for social hype: Word of Mouth (WOM). It was no surprise to see ‘direct’ traffic grow in Dell Swarm Australia. What surprised me was the amount of that growth and the value of the traffic. It grew from low 20’s to high 50’s in percentage terms over a 3 month period on growing traffic volumes.
Direct traffic does not come from any referring domain or cannot resolve that referring domain. So in many ways it is a bit of a mystery. Of course people can book mark the domain, or type it in directly to a browser. New browsers auto-complete URL’s. There are many more reasons for which this article on Actionable Analytics has a good summary. Many of the common reasons I can discount through simple deduction. For example while the email database grew, we could see open rates and that could not account for the growth. Many of the IE browser issues account for some direct traffic, but that should be proportional, and looking across browsers, the data is the same. So this cannot account for the growth of direct traffic over time. We knew what advertising ran and it was comparatively very little. There are server side redirects, but just to move from Dellswarm.com.au to Dellswarm.com/au and we monitored social conversations where possible and saw social mentions actually dropped over time, not grew. Yet, overall the most interesting thing about this direct traffic was the fact that they were 5% less likely to be a repeat visitor (discounting eDM, book marking etc) and more astonishingly, 19.3% more likely to be a buyer.
Dell Swarm grew from a wider distribution of the “Dell Swarm” concept between different people. A person got to know that there are great bargains, but not by sharing individual offers through the online share functions but sharing the service name and what that group buying service offered as a whole. This was not all through online channels which mostly can be tracked, or through online advertising or key community channels which we could also track, but through a growing awareness of the term “Dell Swarm”. It was not via PR as there was virtually none in Australia by intention. So in general it got people talking, which by all accounts people spread amongst people driving them to come directly to the Dell Swarm domain, and be more likely to buy as a result.
Beyond the “Socialgraph”, the power of “Interestgraph”
Underlining the fundamental concept of most social shopping is the fact that I am more likely to like something because my friends like it. There is no doubt that this can be true, certainly for younger demographics (e.g teenagers), however I do believe it needs to be questioned.
If I look at my own socialgraph I think it provides a good illustration of some issues. I have about 350 friends (pretty typical for people I know, especially in this industry), and they are roughly broken down into 3 key buckets;
1. The Ancients; people I went to school or university with and my family and friends associated with my family. In the end, these are people I have known for well over 15 or 20 years.
2. The Colleagues; people I have worked with and met during work related activities. Having lived in many places and having worked in digital this is quite a large group.
3. The Club; being involved with a social sports club with much of its activity centered around Facebook, I have accumulated many friends associated with canoeing and dragon boating in the region.
So while this group of people might be a good predictor of how I might spend my social time, they are not a good or efficient representation of my likes and dislikes. Take for example Music; I can very safely say that my ‘friends’ (as defined by Facebook which is the default way people talk about social in this context), do not share my tastes at all. And I certainly don’t share theirs. This is true of almost anything I can think of. There are some exceptions which are mainly based around a shared interest. For example if you wanted to sell water proof watches with GPS and heart rate monitor, you would tap into over 100 very interested people. In fact we have had informal ‘group buying’ arranged through our Facebook group to buy products just like that.
There are other examples like travel or food, where what I see on my social does interest me. However overall I trust a persons recommendation because I feel they know something, or are similar to me – not because they are my ‘friend’.
Enter the power of the ‘interestgraph’. In many ways it overlaps a socialgraph but it is a far better predictor of what I would be interested in from a social commerce point of view.
An excellent description of the relevancy of Interestgraphs can be found on the Assetmap Blog. And much of this based on the description of Interestgraphs on TechCrunch where the valuation of Facebook over Twitter was put to question.
Of course ‘likes’ are a way of defining interests, however in Facebook you don’t follow someone just because of their interests or tastes. You ‘friend’ them based on personal contact or social interaction. Other services are geared towards an ‘interestgraph’ where you typically ‘follow’ people because of that interest. In Twitter for example the best use for me is following marketing professionals who expose me to new ideas and articles. Tumblr or StumbleUpon are other well-known examples. And there are whole host of new services which seek to collect and network people of similar interests and tastes. My favorite at the moment is Pintrest, which is a beautiful way to organize thoughts or interests and introduce you to people with similar likes.
From a commerce and shopping point of view this can manifest itself in many forms. From Nuji an online and offline shopping experience based on things you like, to Shopkick, a mobile location based service offering rewards for shopping ‘likes’.
Future of social shopping, and Dell Swarm
The social shopping scene is greatly hyped at the moment. I keep reading quotes like “It’s a matter of time—within the next five or so years—before more business will be done on Facebook than Amazon” (Sumeet Jain, Principal, CMEA Capital) and I am sold the big picture by social commerce infographics. But then again, when you start really delving into the topic, and reading the discussions and expert opinions, you get your feet back on the ground. The single best resource I can think of which covers this topic is Social Commerce Today (SCT) by Paul Marsden. Of course there are many articles painting a bright future, but look again, many are usually service providers looking to profit from this rising trend. Importantly there are many people engaged in the ‘industry’ who share their views and help keep perspective.
For me, I am exploring different avenues to help take the thinking behind Dell Swarm and the future of social commerce forward. This is pretty much in line with the thinking and trends which were identified by commentary from this years Crunchies awards and discussed on SCT.
- Not all ‘social’: many services use Facebook Connect as easy sign-in. But the networking and sociagraph defined from this is not greatly used. Understanding ‘likes’ of friends is a key data point, but the data is limited and inconsistent
- Deal based: in the end most services are based on Deals. It is the value of the deal which makes people talk in Dell Swarm. And it is no different in the vast majority of social commerce service.
- Vertical specialty: social shopping is getting more vertical. And I believe this is in reaction to tapping into a more defined and refined interestgraph. Once you share an area of interest it is easy to define who knows the most, or who shares similar thoughts and interests to yourself.
- Online events: social shopping is mostly about an event online. Be it a short time based deal, or exclusivity and scarcity.
- Clicks and mortar: extending real shopping into the digital world. Including real-world ‘likes’ which link back to online assets.
- Location based: social shopping and offers lend itself well to location
- Mobile: to make much of the above happens and to engage people wherever they are, especially in the process of establishing their interestgraph and preferences. And also helping to create action anytime
Above all, I am also keen to take further and better leverage so called ‘old school’ social commerce intelligence. User and customer reviews defined the social shopping category, and in the end these review make a big difference to future buyers interested in a particular product or service. And whether defined as ‘social’ or merely ‘behavioral’ I still feel the data gathered from ‘recommended’ lists is a better predictor of potential likes than my ‘friends’. Take Amazon for example; the socialconnector will tell you what products you might like based on what your friends like. This is based on pretty shoddy data (what they say they like) and off a pretty low base. It is also based on assumption that all friends are equal and all of us have a lot in common. Compare that to the Amazon recommendation engine; it uses a massive amount of data to suggest other products liked or bought with the product you are interested in purchasing. In a way it is based on an ‘interestgraph’ albeit anonymous data from people you don’t know. This behavioral information is a very good predictor of additional products I may want to buy.
What we do specifically with Dell Swarm will be based on time, budget, business and a whole host of other factors. Yet the game has just begun, and the scrabble to win at ‘social’ anything is just reaching fever pitch.