The Customer Experience’s Manager Biggest Dilemma: Web site or Web business? (The Reds vs the Blues)
The Reds believe the purpose of their site is to provide a great experience to their customers, not necessarily sell products. Multi-channel retailers with a small online presence (and little investment) use the Web as a branding tool. Such sites are generally pleasing to the eye, well designed, fun. The Blues believe it’s about results. The goal is primarly to sell online, not to complement the offline channel. Pure-play online businesses or retailers with a large online presence run their site like a business or a division with its P&L. Take a look at some Reds: Levi’s, Movado, Nike. Now some Blues: Tiger Direct, eTrade, eBay.
Why does this matter? It matters because many of the design and usability decisions are easy to make if you know you are Red or Blue. The problem is: many of us are undecided - we want to be business driven yet we like good looking sites. We’re torn when it comes to making simple usability decisions.
Two classic examples - font size, and Flash. Smaller fonts look better, yet numerous studies have small fonts dramatically deteriorate a site’s usability, and ultimately, results. Flash applications look great, but we all know that they can make pages a lot slower and almost never drive higher conversions. So why are we seeing so many Verdana-8pts commerce sites with a huge Flash animation on the home page? Because many retailers haven’t yet committed themselves to truly running their online division like a business.
What is Customer Experience? Hard to define in just a few words… I have met with enough Customer Experience Managers and Directors to know that it depends on your business, your size and the goals for the site. Generally Customer Experience involves Site Usability, Design, Performance. Customer Experience managers typically have a marketing or branding background - they are technical, and they extensively use Web analytics.
We need to think in terms of Customer Experience when we design a new site, a new home page or streamlined checkout sequence. Larger sites understand this well and often have an entire staff dedicated to this. Smaller sites however not always invest in Customer Experience - because the ROI is harder to prove than some shorter term projects.
Perhaps my biggest surprise in my X years of Web analytics is the fact that flashy and cool design seldom mean high conversion rates (remember boo.com). Customer Experience is not a design contest. In my opinion, Customer Experience is a conversion rate contest. As a warm up, take a look at this article on Site Design before we attack some of the bigger issues in the next post.
“Firefox Market Share Gains Continue” says this article on ClickZ. True but my numbers don’t say 10.28% like in the article - I am finding around 5% market share, 5.4% to be precise. I am computing these over more than 23 million sessions and very large Web properties. Not that it really matters - but it’s important to look at the actual numbers and not get carried away by the hype surrounding Firefox. Firefox is doing great but IE is not out as yet.
Firefox share according to my research: February ‘05 3.7%, March ‘05 5.1%, April ‘05 5.4%.
I urge anyone worried about the impact of users deleting their cookies to assess the impact of such practice on the business. In general, I support Jupiter’s findings (see this post on the matter) and find the debate around this issue healthy. However, no one (to the best of my limited knowledge) has done a good job of publicly explaining which reports are affected and by how much. When Research XYZ says 40% of users delete their cookies on a weekly basis, this absolutely does not mean your marketing campaign reports are 40% off.
To illustrate my point let’s look at the impact on the marketing campaign reports when you look at campaign results over 28 days (typical these days). I have run a very simple model where 25% of users delete cookies every week, 10% every 2 weeks, and 5% four weeks - and spread the deferred purchases over 28 days (1/n^2 distribution for deferred orders, 62% on day 1, 15% on day 2, 7% on day 3, etc… pretty realistic based on what I am seeing actually). I then looked at how many deferred orders would not be tracked because of the cookie deletion problem. It’s less than 10% for deferred orders. If you estimate that 65% of all orders for the campaign happen on the first visit, and 35% are deferred, the impact drops to 3.2%.
In summary, with my assumptions, the tracking would be off by around 3% over 28 days for marketing campaigns. Not a huge deal… Find the Excel Spreadsheet here and plug-in your own numbers/models. Enjoy!
Given the importance of marketing campaigns’ landing pages and the fact that 35 to 45% of campaign conversions are deferred and happen within 30 days of the initial site visit, analyzing points of re-entry after an initial campaign sounds… like a must.
Optimizing re-entry pages is a real challenge because marketers have no control over which pages are used to re-enter the site. The dream would be to force users to re-enter the site through a nice: “Still haven’t made up your mind on this product? Here’s a 5% coupon“: Simply unpractical - users can re-enter from anywhere.
To get my ideas clear I took a look at some numbers for a great mid-size retail site with a straightforward online marketing strategy: Search engine optimization, Paid Keywords and Email. Pretty clean, and perfect for what I needed.
I cannot disclose the details of what we found out, but at an aggregate level,
Initial Entry: Email
Re-Entry: 85.8% direct access, 12.6% email, 1.6% paid keyword
– Direct access vehicule: 11% organic search, 89% typed URL/bookmarked
Initial Entry: Paid keyword
Re-Entry: 66.5% direct access, 6.4% email, 27.1% paid keyword
– Direct access vehicule: 69% organic search, 31% typed URL/bookmarked
Finally, for both paid keyword and email users re-entering the site directly, about 75% of them land back on the homepage.
Use your favorite reporting tool to create a list to top re-entry pages for your marketing campaigns are treat them like landing pages. Most often, you efforts will focus on the home page and a handful of other pages which aren’t necessarly the popular initial entry points.
Jim Novo, a well know authority in loyalty marketing, made a few very interesting comments on the Frequent Visitor Program post which are well worth reflecting upon. I encourage reading the full post, here are a few key excerpts:
I designed and implemented what I believe was the first “points for page views” loyalty program on the web, for CBS / Sportsline (then SportsLine) back in 1997. […] The program was designed to differentiate SportsLine and generated not only tremendous increases in repeat traffic but also increased the conversion of free visitors to paid subscribers.
In many cases, it is not worth paying “50 points to know who you are” because you’re not worth it.
For example, let’s say I know you are a paying subscriber and have bought from the store. You also play Fantasy football every year. You are one of the most valuable customers I have. If I see your visit Latency (number of days between visits) starting to expand I have a problem; you are defecting. So when you hit a certain Latency, I automatically trigger an offer that says “you are special, we’re giving you double points for page views this month plus you can get a Joe Namath autographed football for only 2000 points”. We know the guy has 3000 points from the database, so the offer will burn points and get him back to the site.
Bottom line: You have to allocate points by current or potential value of the visitor, or you get killed like so many have with these programs.
I must admit I was a little surprised by this case study for a number of reasons. - First, I was initially thinking this type of program would apply to retail sites, not content. I was thinking airlines miles just like the ones I get when I purchase a DVD at Tower Records. - Second, I wasn’t thinking in terms of customer LTV but personalization, i.e. triggering targeted offers for a user who I know has accessed my site with a specific intent (The intent we infer by analyzing which products were looked at, added to cart, abandonned, etc). - Third, the real challenge for this type of program appears not to be technological, it resides in the actions you take to make these programs profitable, as Jim said “most of us aren’t worth the 50 points”.
OK - I admit I am a little colder on the idea of rewarding people with points or money or miles. Let’s go back to delivering some value to a user who accepts to identify herself upon accessing a site. Can we make a convincing case for allowing this person to be tracked?
Personally, I would be very willing to identify myself in every retail site I visit ONLY if that meant fewer and more targeted ads. But unfortunately, I get this feeling that it’s a lost cause - openly sharing my ID will get me even more phone calls at night during dinner.
I ran query on 5+ Million users. Here are the results for broadband penetration for the top metropolitan areas, data sample covers 4 full weeks.
(If you can’t see this image, click here).
While most studies estimate the broadband penetration to hover around 50% at home (see this interesting summary from WebSiteOptimization.com), the reality as seen from the eRetail side is dramatically different.
To clear my doubts, I ran a big query across half a dozen sites on distinct retail verticals over a full week, to elimate the week end effect. My sample is 3.6 Million users, and here’s what I found out: 71% broadband, 29% modem. Modem is defined as <56kbps (arguably an arbitrary definition).
I ran dozens of other queries. One of them surprised me a little bit, the Average Order Size for modem users is 9.5% lower than broadband users. This average is for the 3.6 Million sample, but when looking at each site individually I did see the same pattern: a significantly lower AOS for modem users.
Similar to earning miles when we make a purchase with our United Visa credit card, could we earn miles each time we accept to identify ourselves when entering a site? I’d be glad to Sign In for 50 miles - obviously I would have to trust that the site in question will not sell my email address to the world. As a nice side benefit, Signed In users can checkout in 1-click. And they get a personalized home page. How smart…
With leads to the following (more serious) business question: what is the value of knowing who exactly is visiting your site - and what he/she is looking at, adding to the cart, etc? Enormous I think. And if so, then ROI = Enormous / (Cost of a few times 50 miles) = still enormous.
Has this (or something similar) been tried? (Yes, this is a call for comments)
Google apparently is launching “Budget Optimizer”, see this page on the Google Help Center. Essentially, you tell Google how much you want to spend - and they maximize the number of clicks you get for that budget.
I don’t think this offering will compete well with modern bid automation tools in the retail space, because it is maximizing clicks, not ROI. Content sites and affiliates on the other hand will love this. It will be also interesting to see what Yahoo and MSN come up with.
In light of the Urchin acquisition a few day ago, it’s not too hard to predict Google will get into full blown keyword bid automation in the not-too-distant future. Customers will then have to decide if they trust Google enough to write them a blank check every month. I hope they don’t - we don’t need another Microsoft.
