Planning Email For When It’s Seen, Not When It’s Sent
Columnist Jose Cebrian envisions a future in which email is truly relevant and acts on the data at the time of open — but to get to that point, you must jump the hurdles along the way.
The future of email will be live content. First-party data will be the key differentiator in driving results, so the latency of today’s production process will need to change.
The advent of real-time content in email over the last few years is changing how marketers can drive relevancy and increase revenue. But to truly leverage its power will require fundamental changes — not only in how we plan what content to show email recipients, but also in the way we look at the data that drive those content decisions.
Consider the following email scenarios from the recipient’s perspective:
The points junkie — When emails show outdated rewards points data, it indicates the company is slow, and the customer doesn’t feel rewarded. If the customer sees her latest points balance, status and offers when the email is opened, even if it is a month later than the send, it indicates that the company is well-run and connected. And proving that the information is relevant to the customer encourages closer inspection of the email.
The concert goer — Show the latest concerts in the recipient’s area based on location data, past purchase history and ticket availability at the time of open, instead of showing what was available at the time of send. Many “big” events sell out early, which creates wasted space and a bad experience in your email.
If you can show your latest available inventory based on what customers are likely to want to attend, they will have a reason to click through and explore.
Your regular shopper — Normally you send an email with the products you want to promote. But what if the email is so successful that some of the products run out? Instead of a poor experience, you leverage the most up-to-date inventory data and the opener’s location to determine what is most relevant and available in the person’s area.
The time shifter — You send an email that has several components, each promoting something specific that was relevant at the time of send. It could be a credit card offer or a special deal. But when the person opens the email, what if he or she has already taken that offer, or it has now expired?
Rather than repeat — or possibly cannibalize — the offer, you can serve a next-best offer based on information at the time of open to maximize the value of that message.
None of these scenarios can be accommodated with today’s most common method of sending email to a large group with static content — the dreaded “batch and blast.”
As marketers utilize more manually written, dynamic content, determining what to show whom based on data values, we start to approximate the desired end state. But even that falls short, because the data used to drive the content are often days, if not weeks old, depending upon how well a company’s systems are connected.
The complementary tactic used mostly by retailers is to employ recommendation engines that can dynamically populate a section of an email with recommended products based on an individual’s past purchases and the most common purchases of people like that individual. Some of these engines generate the content at the time of open and would count as real-time content.
You Need First-Party Data To Make This Work
To be generally applicable to many industries and do more than retail product recommendations, we need to leverage real-time content in newer, more connected ways.
We must make email act more like websites and landing pages, responding to a mix of first-party (your company’s) data and environmental data, such as time of open, device type and IP-based geolocation.
The First Step: Get Your Data Together
You don’t need all of your data, but this is no trivial step. You need the information that is important to your customer and also the information that is important to you for making content decisions.
These are data values that speak to products a customer has bought or is likely to buy, modeled next best product (in any industry), last transaction date, points tier and values, home location, and so on. You do not have to get it all together to start. Get some of it and prove the value. Improve from there.
Access The Data At The Time Of Open
The second step is to be able to access the data for content decisions at the time of open. The content served may be a data value such as points balance, or it can be recommended based on rules written to leverage data values or machine learning.
For example, if we learn that a person just booked a hotel in Miami yesterday, but she opens an email sent three days ago, then the email may serve content to show what she can do while in Miami at the property or in the city, instead of an ad to book in Miami.
The question of where to store these data is an important one, because there can be huge spikes in the volume of requests against a data store when an email is sent. Then there is a surge in opens that would call on the data store to determine a value on which to make a decision.
The answer will most likely be in a cloud environment that can handle the spikes without your having to buy more physical hardware. That cloud environment might be accessed through your email service provider (ESP), your own infrastructure, your marketing service provider, or the real-time content email provider if different from your ESP. That determination will be based on cost, availability, sensitivity of data and speed.
It should be noted that you can make the data less sensitive by using an unidentifiable key passed from the email to the server at the time of open. That key is then matched against a key in the data store to pull the correct values on which to make a decision. There is no need to provide name or any other truly sensitive data.
Once the data are available, you need to make decisions about what to show.
What content do we want to show and to whom, based on what? In this sense, it is very similar to how we write rules for dynamic content in email today.
The difference in the future is the element of timing. Today, when you write a dynamic rule, you are taking a finite amount of content relevant at the time of send. In the scenarios we depict, the email may be viewed several days or weeks from the time of send, which affects how you plan the content.
Don’t be fooled by the myriad technology companies pitching plug-and-play solutions. The technologies out there are impressive and will continue to be even more so, but they all require technical support and service to get up and running.
More than that, though, without guidance — a strategy on what to say when and a plan for how to measure it — any initiative will fall short. You want to improve business results, not implement technology for its own sake.
While the vision of the future is exciting — email that is truly relevant and which acts on very recent data at the time of open — getting there is not without its challenges. Just because you can show updated content at each open doesn’t mean you should.
You may not want someone to go back to your email to revisit something they saw earlier and find that it’s no longer there. Updating email content has to be done appropriately, to provide information that is valuable to the recipient — e.g., a better offer than what was there before or an updated points balance.
Getting the data together and accessible is no easy thing, but it is being done for websites and advertising. Whether email drives the effort or benefits from another initiative, leading companies in the future will need to access up-to-date information about a customer across several touchpoints, including email.
Measurement is going to be difficult until ESPs and messaging platforms truly integrate this type of technology. Today, real-time content is generally powered by a separate platform from the ESP. Dynamic content within one campaign is already hard to measure today.
Given that content blocks can be leveraged across time and segments that shift, more focus will have to be placed upon how these get measured and reported in an easy-to-understand way.
Many industries, especially the regulated ones like pharma and financial services, already have a protracted legal process around email and marketing in general. And some have an archival requirement that will be difficult to account for. That process often serves as a deterrent to moving heavily into dynamic content; adding a time factor will complicate it even further. However, websites and display advertising content approvals will pave the way for this approach.
Today, real-time content in email can be costly. As with email CPMs before it, pricing will come down as more players enter the market and volume increases. As the costs come down, the absolute ROI threshold will come down, and this will be applicable to more scenarios.
As always, start small. Test a simple set of content blocks like tiles in the rails to see what challenges arise. Then move on to bringing in more challenging items like up-to-date points balances or site visit data and build from there.
One can imagine a further evolution in which marketers will plan content to be seen across all of their messages in time parts. In essence, the ESP or email technology becomes an ad server that serves the best content based on time and the information available on the individual.
For example, if a retailer sends an email every day, people are unlikely to open that email every day — they may open only one, and it may not be on the day sent. If retailers can plan by time part and by first-party segment, they will be able to promote the most important or relevant items at that time — across ALL or only some of their emails. The same is true for a media company that wants to promote a particular show or movie.
So, as we move toward the future, the emphasis will be on the speed of data. The faster the data convergence and availability, the more relevant the content can become. As this happens and ESPs integrate real-time email technology, we will be able to move toward managing content on a timeline, not just sending messages.
One can even imagine a scenario in which emails will go back to being primarily image-based, so all of its components can be changed remotely. In essence, the message becomes simply a shell, and its contents are served based on the mix of first- and third-party data available at the time of open.