Online retail is the wonder kid of retail – it is young, energetic, it is growing fast yet it is still in its infancy. Based on 2010 estimates online retail amounted for no more than 7% of total retail purchases, as seen below.
The figure may not be exact as it amounts for purchases that happen exclusively online. Users tend to mix retail channels in their quest for a better shopping experience. They might know the brick-and-mortar store brand and order online because it is more convenient. They might also discover the online store, find the product best suited and than “feel” it in the physical store.
Multichannel tracking has not changed that much since the days consumers would receive coupons in magazines and advertisers would track these coupons to get a better view on what’s efficient and what is not in their marketing efforts.
What is Multichannel Shopping?
First and foremost – what is Multichannel Shopping? As you probably have noticed or done so yourself, shoppers tend to use multiple ways of combining online and offline activities. Here are the most important:
Shopping across multiple channels (brick-and-mortar stores, online shops, mobile apps, phone order etc.). Consumers will try to use the best channel available at the time. Say you are a avid online shopper but this evening your brother celebrates his anniversary and you forgot to buy him a present. You will rush over to the closest store and buy something from there, after you have searched for that store and the gift online.
Using more channels to purchase goods from a single retailer. Users that are accustomed to a certain brand will try to buy as often as possible from that particular brand. They will mix offline and online purchases, depending on the specific occasion, while staying loyal that brand.
Using multiple channels to complete a purchase. Users will use multiple channels sometimes, to get the best deal / the easiest way to get the goods. They might browse the online store, order the product on the phone and purchase / pay for it in the brick and mortar store.
How can we track Multichannel Shoppers?
As retailers increasingly look for new ways of tracking consumers and increasing sales they use a combination of old(er) and new(er) strategies, such as:
Multichannel loyalty programs – this programs are usually extended CRM programs, using identifiers such as member cards, phone numbers, unique ID’s or others. Consumers are encouraged through loyalty points incentives to use their ID’s on the different channels
Multichannel consumer life cycle – tracking the consumer through different channels by combining online and offline purchase steps (Ex.:buy online, pay offline, support on the phone)
Track users through wi-fi and mobile use – a rather cutting edge yet extremely promising strategy of trading free on-location internet (everybody wants some), combining it with personal online data (such as Facebook user accounts) and seeking trends in collected data, in order to increase sales and understand the consumer life-cycle better.
In my last post I talked about the shift in consumer targeting that happened once the Internet went mainstream. Several highlights were the short history in consumer targeting, information regarding Amazon’s personalized recommendations and Apple’s usage of consumer data to increase music and app sales.
Now we’ll have a look at how two of the largest and fastest growing technology companies use consumer data and behavior to deliver ads. As Facebook and Google’s business model heavily relies on advertising they have to make sure ads are delivered efficiently to increase revenue.
However, trying to increase ads relevance and user experience can sometimes lead to unexpected (?) outcomes. Both companies had had their fair share of legal troubles regarding users privacy. For example last year Facebook user tracking practices lead to a request by US congressmen for the Federal Trade Commission to investigate the company. Apparently Facebook would track users web traffic even after they logged out. By linking browsing history, location and time of visit to account information (list of friends, preferences, browser) the company could potentially extend its user profiling to some very intimate data. Apparently the issue was corrected and now Facebook stopped linking browsing data to user profiles. Even so, the anonymized data can provide the company with some very good insights.
What are Google and Facebook’s revenues?
As stated above both companies rely heavily on advertising revenue. 96% of Google’s 2011 $37.9 billion revenue came from advertising. Industries that pumped most money in Google’s Adwords program were Finance and Insurance ($4 billion), Retail ($2.8 billion), Travel and Tourism ($2.4 billion) – source.
Meanwhile Facebook reported “only” $3.1 billion in advertising revenues last year. Even though the numbers are visibly lower than Google’s, Facebook advertising revenue increased 69% and topped Yahoo in 2011.
Having established that online targeting leads to generous revenues, let’s have a look at how Facebook and Google manage to efficiently target consumers using technology:
How does Facebook target users?
Facebook increase in popularity coined the term “social media”. This term describes web and mobile platforms where organizations or individuals communicate through different types of media (text, image, video etc.). As more and more users started using Facebook the available content increased, social links improved as users added more and more friends.
Facebook recognized the opportunity in consumer targeting using social preferences (Ex. “Your friend likes X Brand. You should too.”). Interestingly Facebook managed to give user profiles a real – life feeling by encouraging people to bring their friends along. Of course few people could recognize nicknames such as “MickeyMouse1982” so users started adding their real names, than their birthday, location etc.
Soon enough Facebook had a few hundred million demographic profiles at hand. These profiles were interconnected so influence groups could easily be determined. In a genius move Facebook introduced the “Like” button and later “Share”.
By using the “Like” button users would essentially hand over to Facebook their personal preferences.
As publishers saw that articles posted on Facebook were more likely to become viral and increase traffic they adopted the Like/Share widgets and later the Facebook Connect signup system. As these widgets could track user behavior by transferring traffic data back to Facebook the social network now knew what users were interested outside the platform.
Combining this data Facebook launched and improved in time their Facebook Ads platform. With more than 20% of all web traffic plus data on web traffic outside its social network, the company could potentially target ad delivery better than most other media companies. Let’s review what kind of data Facebook has at its disposal to target users:
consumer demographics: users enter their demographic information during signup or later as they use the social network
social networks: Facebook knows who is a friend of who, who is more likely to have his or her posts liked, shared or commented on. Basically it knows who is most likely to influence their peers actions with a granularity almost impossibly to achieve by others
consumer preferences: every time a user clicks a like or share button, comments, posts a status, photo or video it basically signals Facebook on some of his or her preferences regarding a wide array of things (music, products, news) that could later be used to show relevant ads.
web traffic: by tracking user behavior through like, share or social widgets Facebook registers data that even anonymized can show insights on a scale that no other company can
These are the most important factors in Facebook efficient ad targeting. Weather advertisers choose to use classic ads, sponsored stories or promote several posts the company takes into account this data to maximize exposure and engagement.
How do Google ads become “contextual”?
Probably the most disruptive technology company in the past two decades, Google relies on user data, behavior and semantics to deliver the contextually targeted ads.
To deliver ads, Google needs data. Where does it get it from?
Where does Google get data from?
indexed and ranked web pages: even though the number is not really known as Google is secretive about its data centers, it’s estimated that indexed data is stored in more than 30 data-centers. These data centers hold 35 to 50 billion pages at any given time. They are ranked according to an algorithm initially designed by Larry Page and Sergey Brin and improved in time.
web page analytics: Google Analytics is used by more than 10 million web sites. As Google hosts data regarding traffic and user behavior on these sites it can predict user behavior and ads most relevant to potential consumers.
email information: even though information is anonymized Google makes good use of mails hosted on it Gmail platform. With more than 350 million users in Jan 2012 the data flow through Google’s emailing platform is astonishing.
searches: Google responds to almost 3 billion searches every day. By analyzing searches and user paths Google can determine what are the most popular search results and how can this information be used to optimize ad targeting and delivery.
Google+ is the company’s response to Facebook’s rise in popularity. It already has more than 170 million registered users (mostly active). Having answered the need for information in social networking targeting Google further improved its advertising targeting capabilities.
Android is Google’s mobile operating system. Though buggy at start, Android is now on its way to world domination in terms of mobile operating system.
Basically Google knows a lot about a lot of potential consumers and uses these data to increase efficiency in ad targeting.
Having a look at how the likes of Amazon, Apple, Facebook and Google use research and targeting , we can surely say that conventional (old ?) knowledge on the matter is becoming increasingly obsolete. As technology replaces human input research and targeting becomes real-time.
Unfortunately some privacy issues arise when people become “users” or “consumers”. On this matter – soon.
Conventional (TV, print, radio) advertising often relies on research and targeting methods such as focus groups or demographic targeting to increase brand awareness and sales. These methods seem to be more and more outdated as targeting technology is already delivering better results.
A (very) short history of advertising research and targeting
In the past, as media was unidirectional (broadcaster to consumer), there were few ways retailers could efficiently target potential consumers. Advertisers would use consumer profiles and split purchasing options through demographic indicators (age group, location, education, sex etc.). By using statistic results they could outline marketing opportunities for certain demographic groups (Ex. “Women between 25 to 35 years, urban, having higher education are more likely to buy Product X”).
Having (theoretically) discovered a potential consumer profile they would then buy media in newspapers, radios or TV stations that would best appeal to that certain demographic group.
Of course this is just a skeletal description of the whole targeting process but it explains the process pretty well. Many companies have benefited greatly from this targeting and advertising system. Most of the brands we now know and buy were built this way. Even now, decades after the likes of David Ogilvy were setting up the rules on research-based advertising, the system is virtually unchanged.
“I notice increasing reluctance on the part of marketing executives to use judgment; they are coming to rely too much on research, and they use it as a drunkard uses a lamp post for support, rather than for illumination.” – David Ogilvy
How did the Internet change research and targeting?
Few could have predicted the impact Internet was to have on commerce and economy. Even less would have guessed how this initially “exotic” media would impact research and targeting.
20 years ago there was no marketing concept that could explain AdWords targeting and not be considered science-fiction.
Internet targeting and advertising renders most of conventional knowledge on research obsolete as technology has achieved what was once impossible. 30% of all human population is now in reach of all advertisers and they can now target more than just demographics.
Behavioral marketing is a concept that could not be possibly be achieved with conventional media. Using consumer behavior rather than demographics advertisers can target real time preferences and individuals rather than demographic groups. Say a user is known to have previously visited a car dealership website. He then browses websites in search of reviews on different car models. The car dealership could potentially target this exact user and serve him the most informative ads. Advertising ROI is sure to increase this way.
Some companies have become increasingly good at Internet research and targeting. One of them is now the most valuable company in the world in terms of market capitalization. Let’s have a look at how Apple, Amazon, Facebook and Google use large data to target and monetize consumer traffic.
How did Amazon, Apple, Facebook and Google changed consumer targeting ?
Amazon personalized recommendations
Amazon is well known for its personalized products recommendations. How can it do this? Short answer: large data on consumer purchases and mathematics. Longer answer: Amazon holds a patent on its product recommendations which you can have a look at here (issued in sept. 2006). Although rather technical it focuses on certain key elements:
user profiling: Amazon holds valuable data on user demographics and previous purchases. Using this data it can map users in specific consumer groups. User profiling combines shopping cart contents, item ratings and recent purchases as purchase intents seem to change in time.
similar products information: say you bought three SF books. Similar products would be other books in that category. Some of these books would be more popular in terms of item ratings, reviews, views and purchases.
item affinity is the probability of some products to be purchased together. Say you are buying a Kindle on Amazon. You are very likely to buy a cover or case to protect your device. That means these products have a high affinity index
driver items are those products that are most likely to drive traffic to store. Again – the Kindle, Amazon’s best seller is not only a driver item but also a platform that insures further product purchases.
user path: the consumer will follow a certain path until it ads a product to the shopping cart or confirms a purchase. These paths are very important as they can be used to “guide” consumers to products they are most likely to purchase.
Using these information (and probably more) Amazon can first map users in consumer groups (1), extract popular, affinity and driver products (2), compile most profitable user paths based on previous history and other users actions (3) and than recommend the items most likely to increase basket size.
Recently Amazon announced the launch of its Kindle Fire product. This product is built on a Android platform and uses a proprietary web browser called Silk. The browser optimizes web traffic by routing it through Amazon’s servers. As Amazon already holds information on user profiles (users will have to login to synchronize their book collection) and now data on web traffic it can further improve its recommendations.
Apple Genius recommendations
Although Apple does not explicitly state it monitors iOS user actions it doesn’t deny it either. If it does, however, it might access a huge pool on users data such as web traffic, mobile purchases, locations, call history, social networking information (through access to contacts information, call history, SMS and iMessage history etc.). Basically everything there is to know on its customers profile.
For now the most visible way Apple uses data to increase sales is iTunes Genius, the music and video recommendation system. iTunes Genius uses purchase history and iPod activity to recommend potentially interesting songs, albums or videos.
Although iTunes Genius probably uses a system similar to Amazon’s it is not yet known to be as accurate. The performance issues are probably connected to the number in customers and items on sale. Amazon has a wider products inventory and a larger pool of potential customers. This leads to a larger database and increased accuracy.
Technology based companies have changed the way we think of consumer targeting and advertising. Innovation lead to profits and behavioral targeting will probably develop in the future. Tomorrow we’ll have a look at how two of the largest advertising – revenue based companies, Facebook and Google, use large data to improve consumer targeting. Stay tuned.