For the past 30 months, the ebusiness team at CD and DVD burning software-maker Roxio have been focusing on improving conversions. First they ran a battery of a/b tests, multivariate tests ... you name it. Sales surged. But Senior eBusiness Manager Michael Roehricht wasn't satisfied.
"We were focusing on optimizing the programs -- paid search, affiliates, banners, SEO. But then we realized the real challenge is how can we understand individual customers within these programs?
"We really have to look at the entire customer in a much more holistic, complete view. How can we build a one-to-one relationship with each customer and prospect?"
Here's how his team is evolving their programs toward one-to-one marketing on a reasonable budget:Step #1. Outsource tech (at last)
"One of the key messages is we realized we needed to outsource a lot of the technical components of our ebusiness strategy."
In the beginning most top ecommerce marketers built their own marketing and measurement programs rather than outsourcing. Third party software programs and ASPs simply weren't as evolved as the marketers needed them to be. Plus, the vendor-businesses were anything but stable.
Believing those days had passed, Roehricht frequently ran trials with potential vendors hoping for a winner. He's at last found stable, high-end vendors to power almost every part of Roxio's ebusiness technology, including:
- A/B, multivariate testing, and presentation scenarios - Ad serving - Email broadcasting - Affiliate management - Web analytics - Shopping cart
These days Web servers are about the only things Roxio handles in-house. "Now we can stand back and evaluate results and optimize for the bottom line."Step #2. Centralize all your customer and prospect data in one system
Roehricht was fed up with the all-too-common plight of multi-channel advertisers -- having no way to tie together stacks of reports from a wide variety of campaigns into one single view of the customer. So his team picked a single system to warehouse and report on data from all sources, including:
-> Traffic sources: Banner ads--Offline (print, TV, radio, in-store) advertising--Search optimization (SEO)--Paid search marketing (PPC)--Affiliate marketing--Email marketing to opt-ins--Referral link marketing (hotlinks on third-party sites)--Pre-installed OEM marketing--Refer-a-friend programs--Reactivation campaign (pops, email, etc.).
This data was tied into individual prospect's and customer's accounts when possible, so the team knew all the touches that a customer received and which he or she acted on.
-> Individual visitor's activities: Connection speed--Operating system--Click path--Types of offers responded to (discount, free shipping, none)--Page drop off--Cart activities--Purchases.
-> Demographic and "soft" data on groups and individuals: Country Zip code/postal code (when possible): Sociodemographic data overlays based on zip code.
-> General site improvement data--Usability labs--Focus groups--A/B tests--Multivariate tests.
Gathered in one place, this is a vast amount of data that could easily overwhelm a marketing team. To avoid this fate, the team focused on the few key tests and segments most relevant to improving conversions.
They changed the site's content management system so the site could dynamically serve up content, offers, hotlinks and pop-ups depending on what the database knew or assumed about each incoming visitor.
"Although one-to-one means looking at every user that hits a Web page, we actually apply very few high-level segment rules," notes Roehricht. These include:
- Country; Zip code to determine if offline ads might be in play; Past purchases and inferred loyalty; Traffic source.Upgrade the upgrades sales programs
For software marketers, the true profits start when you re-sell and up-sell customers to more and better packages. Roehricht's team learned three key lessons:
Lesson #1. Individualize the landing page based on key factors
-> Referral page
"Maybe you came from a support page on Microsoft because you're using a very old version of our software that is incompatible with Windows XP. That link may send them to our home page. We may not be able to control it. But we can leverage the fact that we know where they came from and dynamically build a message on that page saying 'We apologize you got an error. A new version that's compatible with your PC is now available."
-> Connection speed
If the consumer is using a very slow connection, then the landing page may be dynamically constructed to highlight the boxed software as opposed to the downloadable upgrade.
-> Search term used
When possible, the search term the user clicked through is displayed in the headline of the landing page as a hotlinked word they can click through for more info.
Users coming from outside the US are given their choice of language (with home-language assumed) and can order in a wide variety of currencies. Plus the team has built out a series of localized Roxio sites for countries with a critical mass of customers, such as the UK and Japan.
Although Roxio may broadcast the same email offer or newsletter to a large number of customers on the same day, the list is segmented by country. That way prospects in each country see the offer in their own language, with their own currency, at the best time for their time zone, and with appropriate links to country-specific landing pages.
Lesson #2. Don't assume email got through
If you're sending a multi-step email campaign -- such as four waves of an educational series to encourage upgrades -- you can't assume every email got through or was read.
So now Roxio's email team is using tools to dynamically change these emails based on which messages they knew an individual user has opened and/or clicked on. Plus, the team suppresses automated emails that were scheduled to be sent to users who've already converted. It's just wasteful and annoying to pitch users on products they've already bought.
Lesson #3. Customers' budgets and motivators may shift
Roehricht says his team has learned to never make fixed assumptions about individual customers. A customer whose past activities revealed strong price sensitivity may now be willing to spend more. A customer who once thrilled to free shipping may care more about features now. People change.
"You have to allow for that change dynamically and in real-time. Is there anything we can cautiously infer from optimizing little pieces here and there?"
As a result the team now allow the site to serve up dynamically the offer they suspect a particular prospect will find the most appealing, but they also hedge their bets by serving a supplemental offer or running an a/b test on a group of similar prospects.
The site, in effect, becomes a decision tree prospects work their way through--changing offers, content and emphasized paths not only based on that visitor's past activities but also on his or her current movements in real time.
Email driven from the site joins in this tree. A certain set of activities will cause you to get a certain email... or not. For example, if a prospect whose click path indicates price sensitivity abandons their shopping cart, Roxio may quickly email an automated, discounted offer for their cart's contents. International lessons learned
Roehricht emphasizes you can't assume tactics that work online in the US will also be best practices for the rest of world. Customer loyalty and lifetime value also varies greatly from country to country.
Example: "We've been surprised by how much higher email click and open rates are for Europeans, especially UK, for reactivation campaigns. We've focused the past six months on localized sites such as Roxio.co.uk and implemented all the tracking and reporting tools." End result: the increase in conversions is actually higher for Europeans currently than it is for US customers who may be more jaded by highly optimized sites.
Roehricht has also discovered consumers in Asia are more different from US consumers than anticipated. "They're much more into relationship building. We're educating users very slowly, giving them a sense we understand them." Brand building and awareness campaigns take longer to pay off, but once they work you may have a slightly greater barrier to competition. Useful links related to this article
Samples of Roxio's landing pages and emails http://www.marketingsherpa.com/sonics/study.html
Kefta - the conversion optimization tool Roxio's team uses to merge data and test new tactics across multiple media including email and Web: http://www.kefta.com
Omniture - Roxio's Web analytics software: http://www.omniture.com
Digital River - Power Roxio's shopping cart: http://www.digitalriver.com
Doubleclick - Send Roxio's basic email broadcasts (not dynamic or individualized content): http://www.doubleclick.com
Commission Junction - Power Roxio's affiliate program: http://www.cj.com