March 16, 2005
In the past four years VistaPrint has served more than four million customers online... and counting. The 40-person marketing team has a budget of $20 million per year, and they test everything possible with it.
In our exclusive Case Study, you'll hear how 10 of VistaPrint's most fascinating tests turned out, including:
-> Offers in emailed order confirmations
-> Site exit pops (multiple tests)
-> Cross-selling during the cart checkout process.
On the face of it, VistaPrint doesn't sell anything that hundreds or even thousands of other printers can't sell you as well. Heck, anyone's neighborhood print shop can run off a few business cards.
If your product is a fairly easily-found commodity, you need superlative marketing to win and keep millions of customers.
CMO Janet Holian based her initial campaigns on free sampling. Anyone visiting VistaPrint's site could order a packet of personalized business cards at no cost beyond an obviously token shipping and handling fee.
Holian crossed her fingers and hoped product quality combined with ongoing email offers would convert samplers to buyers.
It was a fairly soft offer as sales tactics go. Although it was working well (and it continues to be the site's control offer to this day), she needed to get quicker payback from new accounts.
One day early on Holian wondered, "What if we test adding a harder offer cross-sale in the sample checkout process?"
The Web design team popped in a paid-only offer for a matching set of return address labels to go with sample business cards. "All you had to do was say, 'Yes I want this discount.'"
The day the test launched, Holian posted a betting sign on the office wall so everyone could predict results.
"28% of checkout traffic took the upsell. It's a number higher than what anybody in the entire company guessed. That one offer got us to the point of being cash flow positive."
It also had a profound impact on VistaPrint's corporate culture.
"Testing is at the heart and soul of VistaPrint," notes Holian. Every single employee considers it part of their job to help out eagerly -- not just marketing but the engineering department, service, fulfillment, and management.
For example, the Web development team has a rolling three-week testing schedule. In week one, tests are developed for a variety of parts of the site. In week two, tests launch and collect results. In week three, the team analyzes results, implements winning controls, and plans the next batch of tests.
Plus, no test result is ever final.
Trynka Shineman, Director of Retention, manages a rolling quarterly test schedule featuring a specific testing focus for each month. This helps her make sure no significant campaign tactic is overlooked in the course of the year including tests for email, site design, pops, and ads.
High-impact tactics are systematically scheduled to be re-tested every year or sooner. "We rotate testing on creative, tweaking offers, subject lines/headlines, new variations. We are always surprised by results. It's so fun to say, 'What do you think is going to happen here?'" says Holian.
VistaPrint now boasts more than four million customers, an annual marketing budget of $20 million, and 15 different national ecommerce sites including Canada, Sweden, Italy, and Australia.
Not bad for a printing business that opened its doors in 2001.
Here are 10 of the tests you might find the most significant.
Test #1. Offers in emailed order confirmations and receipts
MarketingSherpa data shows your customer admin emails may have a higher open rate than any promotional email you send. So why not add offers to these emails?
VistaPrint tests have been overwhelmingly successful, so offers are now a regular part of the standard creative. Plus, partners such as eBay Europe now add VistaPrint offers to their emailed confirmations and it works "extremely well." (Something to test with your super affiliates.)
Test #2. Emails to abandoned carts
VistaPrint's discovered it works best to email personalized reminders (including individual cart details) to shoppers three days and then again 30 days after a cart is abandoned. The premise for the second email is the fact that the cart is about to be wiped from the site's memory to save on server space.
Test #3. Text versus HTML email
Yes, that old chestnut is still worth testing. (In fact, now more than ever with filters stopping HTML sometimes.) Shineman says, "We've found it's better to mix it up a little bit. Sometimes one will outperform the others. I'll send maybe 15%-20% text-only and the rest HTML."
(Note: This means sending occasional text-only messages to customers who have signed up for and can view HTML.)
Test #4. Interactivity in landing pages
Should you send new prospects to a page that educates them about your offer, products, and brand or should you drop them off at the start of an interactive tool so they can start shopping right away?
Increasingly, VistaPrint is finding it's better to get prospects actively involved designing their cards and to describe the particular offer on the back-end. (This is a trend we've noticed with plenty of other sites, from Classmates to Frederick's of Hollywood.)
Test #5. Landing page navigation
Should you include all your standard site navigation on landing pages? VistaPrint tests reveal the answer may be no.
The team currently only includes navigational items that are critical to the conversion process the prospect is directly involved in at the moment. Most landing pages do not include the rest of site navigation -- only a stripped-down version if any at all.
Test #6. How many cross-sell offers can you add to the cart?
If one cross-sell offer works well in the cart checkout process, such as matching fridge magnets to go with your business cards, how about adding a choice of two offers?
Failure. Two offers was too complicated. Shoppers were willing to click to add a single no-brainer upsell to their cart, but a second offer meant they slowed to consider... and that was deadly for conversions.
Test #7. Free shipping offers
Although free shipping's proven a huge winner for many eretailers, "anytime we test a discount against free shipping, the discount always wins," says Holian.
Why? "Our customers are looking for something specific. They'll react to an offer for that rather than a generic free shipping offer."
Test #8. Pop-Unders
"Pop-ups and -unders used for acquisition are working less for us," says Holian. However, "pop-unders on our own site maximize results incredibly."
The team started testing site pop-unders in November 2003, just about the time when other sites such as AOL and iVillage were making public announcements about how they were reducing pops to curb user dissatisfaction.
Why take the risk? "Because we test everything," Holian says firmly. "It's never too scary. If a test outperforms a control, it's worth it."
VistaPrint found pop-unders for education (hey, we have a new product) worked fairly well, but pop-unders featuring specific product offers worked even better. "Our pop revenues increased 200% in the last eight months since targeting offers."
Through methodical tests, the team discovered the best offer depended on where a visitor was when he or she left the site. A newbie who bailed after only seeing the home page would react best to the traditional free business card offer. Someone who left midway through the checkout process reacted better to a discount, etc.
Test #9. Are two pops better than one?
"We tested double pops, with one showing a broad offer and the other with a single product offer," explains Shineman. "We found success depended on the strength of the offers. If they are equally strong [ie., convert equally well] then two pull better than one."
"However, if one is much weaker, then the weak offer cannibalizes the strong one."
Test #10. Generic versus cultural Web design
Although VistaPrint hires a native of each country to be in charge of marketing for his or her own country's site, the sites are constructed using generic templates. Language and prices are translated of course, but not overall style.
Holian figured it might not be optimum, but this system saved a great deal of wear and tear on the site design and engineering team.
But when VistaPrint launched in Japan, the Japanese-born marketer absolutely insisted the company develop a cultural-specific site.
Holian agreed to a test. So the company launched not one but two Japanese sites -- one using the standard template and the other copying typical Japanese site design -- and split incoming traffic.
The standard template won. Looking like a Japanese site wasn't critical as long as the language and pricing were localized.
Bonus test: Personalized reorder emails.
VistaPrint sends an email to past customers featuring a graphic of their exact past order (such as the image of that customer's own business card) along with a big fat "1-click re-order" button.
Works so well that now the company is testing altering the main home page to promote reorders as well. Instead of seeing the regular home page, some returning cookied customers will see a personalized page with personalized offers.
Useful links related to this article
A few creative samples from VistaPrint tests: http://www.marketingsherpa.com/vistaprint/study.html
Responses -- the email broadcast firm VistaPrint uses: http://www.responsys.com