Surprise Finding: Website Traffic Source Analysis at the Shopping Cart
Your website is a leaky bucket. The more traffic (water) you put in the top, the more it spills out. On average only 2-3% of website visitors convert, while only 3 out of 10 people that get as far as the shopping cart convert. That’s less of a bucket and more of a sieve.
So why are we as an industry so focused on website traffic and not on plugging the leaks? Most emarketers want traffic above all and are willing to invest big bucks to get it—through SEO, paid search, display advertising, referral, social media, and so on. These all drive traffic with varying degrees of precision.
However, if you analyze the traffic sources that drive conversions, as opposed to the overall volume of traffic, then a different story emerges. A member of the SeeWhy research team pointed out that email was often the largest source of traffic for ecommerce conversions. Email as a significant source of conversions, yes… but the largest? This warranted further analysis.
So we dug into the data. Using a sample of 60,000 completed ecommerce transactions across multiple sites from February 2011, we were able to look at where the traffic came from that generated those conversions.
Rather than look at site traffic sources as they arrive on the site, we decided to look at traffic that reaches the top of the shopping cart process. This is essentially when an item is added to the cart and can be considered a signal of intent. The traffic picture at the shopping cart looks very different from the sources at the homepage and other key landing pages.
At the shopping cart, email represents more than half of the traffic (57%), while direct traffic contributes 18% and display advertising is a meager 1.7%. (…)
In Part 1, we looked at how promotions can have a dramatic impact on the revenue that can be recovered with triggered shopping cart abandonment emails. In this blog, we look at the potential pitfalls of using promotions in shopping cart recovery campaigns and the different strategies you can use to avoid these issues.
We’ll take a look at the three broad types of problems you’ll encounter in turn:
Getting out of step with the customer
Viral spread of promo codes
Training the customer
1. Getting out of step with the customer
Problem: You set up a trigger-based campaign that sends follow-up emails to customers who abandon their shopping carts. You include a promotion to encourage customers to buy. Within a couple of days of going live, the head of the customer service team is in your office berating you about the angry customers they’ve had on the phone and the rebates given to customers. What went wrong?
Your campaign got out of step with what customers were doing. This has happened to some of the largest U.S. retailers and has caused some significant red faces.
The problem here is due to a delay between the customer abandoning their shopping cart and the email going out. A small percentage of customers will come back and buy before your remarketing email goes out; so when you make a promotional offer, you are bound to upset some customers.
Batch data transfer mechanisms, where data is sent say every 24 hours, all suffer to some degree from this problem. If you are stuck with one of these, then you are effectively restricted to only sending a single remarketing email without a promotional offer. These types of campaigns typically deliver less than a quarter of the return that you should get from a multi-stage campaign with all the bells and whistles. (…)
Analyzing abandoned shopping cart data from one of our customers last week made me sit up and take notice: Optimizations to their remarketing campaign and the introduction of a promotional discount for the first time caused their recovery rate to jump from 18% to 46%.
Wow. Clearly promotions can make a big impact.
There’s lots of academic research which shows that while promotions have limited affect on long-term sales, they enable marketers to grab market share by incentivizing consumers to stock up on their product at the expense of a competitor’s share.
But while analyzing this data, I wondered whether promotions work altogether differently in remarketing, compared with a promotion made at the point of sale.
A point-of-sale promotion encourages the customer to add a promoted item to their shopping cart and checkout. A remarketing promotion encourages a shopping cart abandoner to come back to the site and buy the items they abandoned. The difference here is, of course, that the abandoner has already considered—and rejected—your value proposition and decided not to buy.
A portion of your abandoners, of course, are still considering and have decided not to buy yet, but as we know from multiple studies, ecommerce leads go cold very fast. The latest study from MIT shows that 90% of ecommerce leads go cold in just one hour.
So perhaps promotions have a very different role to play when looking at shopping cart abandoners. Promotions are your chance to achieve two goals, when customers have already decided not to purchase from you:
1) Get the customer interested again
2) Recover the lost sale
Why Promotions Work So Well in Remarketing Emails
So what role exactly do promotions have as part of a remarketing campaign?
Uniquely, remarketing promotions give the ecommerce merchant the opportunity to change the value proposition after the initial one has already been rejected. (…)
In January, Booz & Company produced a report suggesting that stores on Facebook were set for explosive growth—a 56% CAGR over the next five years, which some commentators have hyped as 600% growth. In 2015, the report says, the market for products sold on social network sites will be worth $30bn, of which $14bn will be in the U.S.
According to some commentators, this is a ‘massive opportunity’ that ecommerce companies should jump on and get the ‘first mover advantage.’
Hmm. Many of the ecommerce heads who I’ve talked to that have Facebook stores are somewhat more cautious, referring to their stores as ‘an interesting experiment’ or ‘we’ve learned a lot and so has our Facebook shopping cart vendor.’ When asked about sales volumes, they typically report low single-digit sales.
All of this says to me that these are very early days, and while early adopters may want to plunge onto the bleeding edge, it’s frankly not right for most ecommerce companies. In fact, I’ll go further and suggest staying away for 2011, with a few exceptions.
A $14bn U.S. market may sound really big, but I wanted to see how big it is relative to forecast growth for traditional ecommerce.
By combining the data from Booz with the U.S. Commerce Dept. and mobile commerce forecasts from CODA, for the first time we can see where commerce on social networks sits relative to overall online sales.
As a percentage, in 2015, after that 600% growth, commerce on social networks will represent only 4% of all online commerce.
Mobile commerce is forecast to be three times bigger by 2015.
Here’s the data, published for the first time:
Of course, all forecasts are wrong: They’re either too high or too low. But for mainstream ecommerce it just doesn’t make a whole ton of sense right now to duplicate the ecommerce site, particularly when sales using the Facebook channel are miniscule. (…)
Last week at the Conversion Conference in San Francisco I had the pleasure of interviewing Tom Davenport. Tom, is a Babson College Distinguished Professor, and research/faculty leader and co-founder of International Institute for Analytics.
I blogged last week on his keynote to the conference. In the video interview below, I ask Tom specifically how we can apply his thinking to ecommerce and web analytics.
Some key takeouts from this interview:
If you’re an analyst, it may be easier to move to a different employer than struggle to make analytics relevant to your organization
Companies that compete on analytics often look to hire-in analytically oriented talent when building teams
More data in ecommerce means that there’s enormous scope to get competitive advantage from your data
Many ecommerce companies are still struggling to make web analytics actionable
The average number of web analysts per company is 0.25 full time equivalents