Scale rocks – especially in our era of big data. Think of all the challenges we face as data-driven and digital marketers – creating a consumer experience, infrastructure, data, analytics, talent, technology, content and strategy – all are just the “joy” of having lots of customers speaking with you! This was the perspective offered by Donald Parsons, Director of Email for Amazon in his keynote to the Mediapost Email Insider Summit this week, where I am honored to have served on the programming committee.
“The challenge of digging through it all, that is the fun of what we do,” Don said. While to the rest of us the task of managing highly targeted and customized email and digital experiences at the scale of Amazon can feel as daunting as building a Byzantine-style cathedral, Don broke down the challenge of having customized customer conversations at large scale into two big categories: Infrastructure and Messaging.
There are real problems with infrastructure as you scale, of course, and Don advises that marketers must ensure they have good internal and external partners. He advised paying particular attention to:
- Cost of storage
- Compexity of data storage
- Rapid growth of processing
- The need for specialists (e.g.: database, data hygeien, cloud marketing, etc.)
DMA would add that data governance and privacy management are also key aspects of scale. As everything gets bigger and faster, the need increases for every data-driven marketer to be the steward for the responsible use of consumer data at your organization.
When executing a messaging strategy at scale, the goal is to use the data you have to create a meaningful dialogue with customers. “Right time, right message, right channel” has become a mantra for every data-driven marketer. We’ve always talked about email marketing being a 1:1 channel, but today, we finally have the technology, automation, analytics capacity, cloud infrastructure and database solutions to handle lots of data and even big data. Amazon is in many ways the bell weather for the rest of the email industry. How many of us have been challenged by someone internally to “just do what Amazon is doing!”
Don makes it sound easy, of course. He mentioned three factors that guide his team’s messaging efforts, Pace (frequency), Cadence (timing and mix) and Conflict.
” The most obvious conflict is the the one between messages,” he said. This is a struggle for all multi-product companies. Let’s say you have five offers, and you know they are all relevant to the customer – so which do you choose or which goes first? “As you scale up,” Don says, “You can’t make any involved decisions, there is no time for a human interaction. So we’ve developed algorithms that help us decide which are based on a balance between value to customer and value to Amazon.”
It was then that Don said something pretty provocative “It’s perfectly okay to bring up in a valued conversation a message that is important to you even if it’s not important to the customer,” he said. As Homer Simpson would say, “Doh?!” Really? We should send messages that are not data-driven as part of a “conversation”?
“There is always value to send something of value, even if you don’t have anything in the behavior or the past history of that customer conversation indicating that the customer WOULD value this thing,” he said. “It’s the one time you break that ‘customer decides’ rule. If the message matters to us and we want the customer to think about it, even if they haven’t yet. So we send it.”
He quickly adds that you also have to listen to the response. “If you discover that they don’t think it’s cool, then stop,” he said. “You will find that this can be be extremely successful – it keeps a randomizing element in your data.
“It can also be very addictive,” he says. “You may have a thousand messages you want to test and you can’t do them all or even do this every week. You have to have some discipline.”
This kind of interruption in a customer conversation can only be done if you have a very strong relationship already. Customers who trust you will tolerate it, Don said.
“How do I know the right time from the wrong time or the right message from the wrong message?” Don said in closing. “Amazon always points to the one thing is the same: That the customer decides. The customer is the only who knows if they are right, if the timing is right, if the offer it right. We always model for customers that look alike and we listen carefully to the response data and all the algorithms we use to give voice to the customer,” he said.