Russell Lewis (Adobe) and Daniel Stubs (Conde Nast) spoke at Adobe Summit 2015 about how to combine a variety of tools for a personalized experience.
A testing solution and personalization engine are only as effective as the quality of data fed to them. Combine the powerful data aggregation and audience mining capabilities of Adobe AudienceManager and Adobe Analytics with Adobe Target real-time testing and automation, and quickly validate the best strategies for personalizing content to your most profitable audiences. In this session, Conde Nast shares how it uses Adobe Analytics Premium and Adobe Target Premium with Adobe AudienceManager to test and deploy propensity models, look-alike modeling, and machine learning for personalization across 20 properties.
The Pre-Adobe State is the human driven approach to marketing. It’s very manual and personalization is limited to the relationship.
The modern state is to automate personalization based on up to date data.
Note: You should at least peruse the key examples here. They had some great content.
- Start with the data you know. After that you can go buy data.
- Prioritize your objectives
- Evolve your automation. Use the tools you have for that first.
Conde Nast Info
It’s a 100-year-old company with 21 brands from fashion to technology. It’s more than just a magazine company. The Scene, for examples, is a video hub. Conde Nast has a strong editorial culture. Their role is understanding people and what they want.
When first developing their digital practice, the editor in chiefs and digital editors pushed back. They were thinking it was only about telling them what to say. It took some time to overcome the challenge. Daniel positioned his team as the voice of the consumer to give the editors key information on channels, devices, what’s being read, etc. Daniels team is about analytics and audience development.
Conde Nast’s relationship with Adobe extends back to 2012 with Creative cloud and web analytics. Then bought Audience Manager, Target, Dynamic Tag Manager, Data Work Bench, Adobe Target Creative, and Automated Personalization. Adobe’s stack helps Conde Nast better understand their users and bring value to their users.
Just a few years ago, the primary interaction was through a magazine. It’s a good business. Everyone pays for the content. The advertisers pay as well. But then…… the web took off and advertising moved towards digital channels. In order to maintain ad revenue, they had to shift towards the digital channel.
Big question: what to put on the web site? All magazine articles, some, paywall? No login?
Then everything changed again with smartphones and tablets being 50% of all interactions. Conde Nast’s biggest challenge isn’t the mobile phone interface, it’s the fact that facebook, and others were the channel. It’s a part of the facebook or twitter experience. That’s the big challenge.
The good: reaching more people every day than ever before. Personalization is an opportunity now with many different reader segments.
Abigail is a reader on the site via “Side door traffic” There are a number of goals
- Sign up for newsletter
- Buy the magazine
- Follow on social
- View and click the ad
- Buy the product
- Read a second story
Abigail wants very different things:
- email her bos
- call a friend
- make a reservation
- read her facebook read
She’ll give you 15 seconds to make an impact.
Hence the personalization problem. What to prioritize given everything you can do and everything different people at the company want?
Insight: Loyalty matters. It’s the top metric. Loyal users give you 20x more pages than average. They are 50% more likely to engage with ads. They are 8x more likely to convert to a subscription
Loyalty lends itself to a customer journey approach. You can’t create a loyal user out of thin air: You follow key steps:
- Edit / Audience Development are the internal stakeholders
- Engagement – enjoy the experience. Share and comment on an article
- Edit and product management are the internal stakeholders
- Loyalty or repeat visitations
- Edit / Subscription Marketing
- Covnersion – paid subscriber
- Monetization – target ad experience. We know things about you so we can create a better more relevant experience.
1. What type of headline attracts people? (headline testing using Adobe Target)
Vogue.com: Headline A: This is all you need for valentines day. B: Llama, Pajamas, Male models, What more do you need for a snow day. B won by a mile. They learned that alliteration works.
2. Engagement: personalizing content. compile a list of similar articles. It used to be a manually curated list. Adobe used Adobe Recommendations (part of Adobe Target premium suite) It makes recommendations based on user behavior.
Result: The machine won. With testing they found a 26.28% lift in monetizable inventory.
Note: this is data Conde Nast already had.
3. Loyalty: does a targeted newsletter interstitial increase email acquisition?
Started with channel analysis. facebook vs twitter vs email vs direct. Found that email groups were very small but highly engaged. They spoke to the editors and found that they didn’t really care about email. It was managed by interns and didn’t get much attention.
Solution: move the signup module out of the bottom of the page to front and center. This got more sign ups but while angering people.
Second solution: Look at who signed up. Use that data to create a segment of the type of user and then target the interstitial .
Result: 300-400 new sign-ups (a 10X increase)
4. Conversion: How do we personalize magazine cross sells
How do you pair the right impulse offer to the right consumer? The past method used a direct mail method. They tested the impulse offer from the direct mail and found that impulse changed when moving online. this increased impulse buy by 11%
How does Conde Nast get info:
- Browsing data from Adobe Analytics
- Subscriber data
- Third party data
Adobe Audience Manager takes that info to create a unified user profile and setup the segment for additional personalization. Conde takes key information like name and address to get more info on subscribers visiting the site. Live Ramp does this and then sends it to Audience Manager. (really cool and really scary….)
This really helps when you don’t have a registration wall.
Goal: create a persona for glamour addicts to inform test copy and creative design
How: Used Audience Manager to define the segments and then test it out on what worked for those types of users (A/B testing) They also found which types of user the testing worked least well (mobile was one segment)
Another example: used a propensity model to help New Yorker define their paywall. they found that while conversion was low. the 5 articles for free model still gave 250,000 subscribers.
Want to dynamically assemble pages based on what they know about you. This includes;
- The banner on most shared – don’t show what’s already been read
- subscription marketing offers – stop targeting the current subscribers for example
- Advertising: help Cadillac understand who in the subscriber based is likely to buy a car so they can get better ROI