Spotted: Email personalization at scale

We’re noticing more data-driven marketing teams moving (back) to sending plain text (instead of the beautifully design & templated messages). They appear to be from a human, not a brand. They appear more personal.

Brands beautifully templated emails are like glossy flyers in your post. But if I remove the gloss, does that disguise my generic, sent-to-all message? No!

Last week, our friends (and customers) at Drift announced their plain text email product. And whilst we follow and fully support their argument, it doesn't go far enough for us.

Plain text templates do not equal personalization. No alone.

Marketing teams (in particular) still need to engage many people at scale. Whatever the style and templating, email is still the primary and preferred channel to communicate with leads and customers. Which means they need to be able to send one-to-many email at scale.

Email isn’t dead. It’s still the #1 owned channel for nurturing leads into customers into raving lunatic fans… and it’s MOST startups’ MOST under-leveraged channel.

Susan SuMarketing at Stripe

Emailing one-to-one, you have the context of the recipient in mind. So you write personally. When emailing one-to-many, you need to build this context from the data you have and make sure it is available, complete and accurate within your email tool. (Then wrap that all up in a plain text template).

So you have a data problem. Not a marketing problem. Not a design problem.

The challenge is assembling relevant & reliable data, writing an effective email that leverages that data, then sending that to the right people at the right moment for them.

From observing teams at Drift, HubSpot, Mention, and more, here’s how they break down the data needs of personal email into different tactics.

Tactical Takeaway #1: Shortlist criteria for personalizing your email before creating segments & writing messages

Remember, the idea of personal email is to gather the context of your recipients first. But, when sending to many people at once you need to gather the different categories of context. The best teams start with the end-in-mind and gather this context before they even build segments or write content.

There are three questions to answer:

  1. What are the objectives of the email?
  2. What are the critera for variations?
  3. Who are specific, named people who match different criteria who are going to receive this email?

Note: objectives, not objective. Criteria, not criterion. If you’re sending to different people, you may well have different sets of objectives (signup, share, reply etc.) to consider. You may (if sending a one-to-one email) ask more than one thing to a person. Since you're sending to many people, you can afford to be a little greedy in the objectives you want to accomplish.

For example, with a new SaaS product or feature launch, there are a number of useful factors which might determine who should get an email, and what an email should say. Consider what you might email an influencer on ProductHunt vs. your current sales opportunities vs your biggest & best customer vs. your worst customer, and so on.

Objectives:

  • View product page
  • Signup & try it
  • Share it on social
  • Upvote on ProductHunt
  • Comment on ProductHunt

Variations:

  • Are they an existing user? A beta user? A churned user?
  • Are they on ProductHunt? Twitter? Are they active?
  • Which timezone are they in? Could they be useful for early upvoting & sharing?
  • Weekday (how acceptable is it to email someone on a Friday evening their time)
  • Are they an in-real-life friend? Do you "know" them?
  • What’s their native or preferred language?
  • What’s their psychographic profile?
  • What’s their job title, role, and seniority?
  • Have they engaged in product-focused email campaigns like this before? (‘cause it isn’t to everyone's taste)

This list is deliberately longer to give you some ideas, but aiming for 1-2 objectives and 3-4 significant variations is a good starting point.

The final point is to name specific people who fit into different variations. Ideally, these are people who you personally know already. These individuals will help you calibrate and personalize the emails you write.

Voila! You now know what you are personalizing your email for.

Tactical Takeaway #2: Enrich your email tool with the data you’ve outlined

The context of all the objectives & variations above needs to be in your email tool.

For each criteria you’ve outlined above, find a reliable data source and integration method you can use to integrate it through Hull (or otherwise). You need to ensure you have reliable, accurate, accessible data for your objectives and variations before investing in creating segments and crafting messages.

This involves finding a reliable data source and an integration method. This can extend beyond one-click integrations and manually uploading CSVs — see what customer data integration without integrations we’ve spotted.

Your email tool might limit you to the type, formats, or volume of data that can be uploaded. As you create more and more personal email, you may need to condense many of your traits into one global “signals” attribute instead of one attribute for each factor you’re ever personalizing emails with.

Variation Data source Integration method
Timezone Last page view (web analytics) Integrations
User status Product database (user.email “has any value”) SQL Importer
ProductHunt user Custom data enrichment & scraping for profile Processor
IRL friend Gmail contacts Google Sheets importer
Language Data enrichment (Clearbit Location) Integrations
Psychographic profile Custom data enrichment (Crystal DISC scores) Processor
Job title, role & seniority Data enrichment (Clearbit Job Role) Integrations
Engaged with product launches Email opens & clicks (email tool) Integrations & Webhooks

Voila! You now have the data you need in your email tool of choice

Tactical Takeaway #3: Create three types of segments

There are three sets of rules you need to create to manage who receives what email.

  1. Everyone who will be receiving the email (as a segment)
  2. The variations within the email (for managing dynamic content in the templates)
  3. Suppression (as a segment)

The first segment is a simple rule for everyone is receiving this specific email. For instance, anyone who is subscribed to your email newsletter.

The second is to manage your variations within your personal email. This data needs to be created in a way your email tool of choice can ingest and use to segment its list and power templating within your emails. Not all tools can search within segments to power templating, however. From Hull, there are three options here to review based on your email tool:

  • Segments
  • Traits as an array
  • Group of individual traits

Finally, you need to create and manage suppression lists. Suppression lists are vital to throttling the amount of email you send any one individual, preventing sends to spam or bouncing emails, and managing unsubscribes globally. By creating and updating the rules for your global suppression lists centrally in Hull, you can ensure every team and tool that is sending email is in sync.

Voila! You now know who you can send your email to.

The only way to win customers in today’s world is to be real, be authentic, and be a human. That’s what helps you cut through the noise.

Dave GerhardtVP Marketing at Drift

Tactical Takeaway #4: Write to individuals, send to lookalikes

Now you’re ready to write the emails. At this stage, many sales and marketing professionals fall into a common trap: Simply because you can capture a wealth of data on a person or company doesn’t mean it is all useful or valuable within your email — this triggers the brand-destroying “creepy” feeling.

To avoid this, you need to write as you’d write to a person you know. Go back to your shortlist of named individuals you started with and use that to calibrate your intuition on what the right level of personalization is.

Writing the email

Start by getting the outline in place. Write a straightforward, “impersonal” email first:

  • Salutation
  • Introduction
  • Body
  • Call-to-action
  • Sign off

Use this as the basis for your variations. This saves you rewriting (and restructuring) emails for each individual variation. Start by writing your outline from the list of objectives for the email you created earlier. As a personal email, this should be conversational in tone (like you would in your own emails), not wooden.

Now, you need to personalize the email. From your list of named, known individuals that match variations, you will have personal context with each of them. Take your sets of criteria for variation you created earlier and start writing variations. Swap words, sentences, paragraphs, links & images for different segments.

Implementing the email

In your email tool, try to avoid creating many separate emails for a single email send. We’ve seen copying campaigns can become cumbersome in certain tools, creates silos & obscurity in your campaign metrics, and can cause many emails to be sent to the same person (where multiple emails go out and suppression lists aren’t updating fast enough — “communication clash”).

Instead, try to manage the complexity of variations within your email templates. Sales engagement tools like Salesloft are built around this idea, though it’s often an SDR who adds the personal touches individually.

For email personalization at scale, we need to abstract ourselves away from the 1:1 final touches and create rules to power dynamic content. Some general rules to follow:

  • If you’re using dynamic variables like {contact.firstname} or {company.name}, then make sure your email system & template has a fallback that works.
  • Check the values of data that are inserted into sentences are clean. The company name ”Hull” is cleaner than “Hull Inc.” in the context of a personal email.
  • Check capitalization on names of values which are inserted into sentences. Your email’s templating engine might enable you to auto-capitalize certain words

If your templating engine allows it, try to move beyond simply substituting words for values of data. Instead, use data to change a complete paragraph or length of an email (particularly important for psychographic segmentation, for instance with Crystal data).

# Word replacement example
As a {{ customer.technology_CRM }} user, I'm sure data transformation is a pain-in-the-ass.

# The above example renders as this
As a Salesforce user, I'm sure data transformation is a pain-in-the-ass.

By using conditional “If/then” logic, you can swap the entire sentence. This gives you the space to write much more naturally which feels much more personal.

# Conditional sentence swap
{% if customer.technology_CRM == "Salesforce" %}
  I'm sure you're sick of Salesforce APEX too
{% else %}
  I'm sure data transformation from your CRM is a pain-in-the-ass.
{% endif %}

Different tools let you manage dynamic content blocks in different ways.

We notice Customer.io is a preferred email tool amongst data-driven SaaS teams. Their Liquid templating enables almost limitless flexibility within email templating — it condenses the complexity into one email send. With Hull's Customer.io integration to sync the data you need, you can create deeply personal emails quickly.

You can also manage similar dynamic content in Mailchimp with conditional merge tag blocks. We see a number companies use Mailchimp as an email newsletter solution, which means sending lots of email regularly - a prime candidate for email personalization! Their merge tag templating isn’t as neat as Liquid (Ruby) which can make it trickier to write, read & test. Beware! You can also sync data to Mailchimp from Hull.

In other tools like HubSpot, you might not be able to manage sentence-level swapping, but you can swap words and create conditional blocks of “smart content” to replace sets of paragraphs & images. This is a little more cumbersome, but you can achieve a similar result. Sync the data you need through Hull’s HubSpot integration.

Testing the email

You should review each major variation of the email you’re about to send. Go back to your original set of named individuals who are included on the list and review each email:

Does this look like an email I would send to them 1:1? If not, why not?

Most modern email tools enable you to preview and send test emails with different individuals who are receiving the message. Here's a sample preview inside Customer.io - where I can tab between this view and "Layout & Preview".

If you like the copy you're previewing there, you can subscribe to Hull's newsletter for more "Spotted" playbooks, guides & best practices, and product news.

If you’re starting out with hyper-personalized email, it may take a few iterations to learn how to fluently use your data through your email tool to craft the desired email. Go back and rewrite until you have a set of emails that work. Write to an individual, then send to lookalikes.

Voila! You have an email that is perfectly personalized to all your recipients

Don’t get caught losing control of messaging. It is still your job to optimize the messages and you can’t optimize what you can’t measure.

Jordie Van Rijn

Tactical Takeaway #5: Orchestrate your email engagement

Personal email? Personal emails!

Now you can create and send highly personalized, individual emails at scale, you can see how to make many emails work together in a way that’s truly personal. This involves a combination of smarter suppression and developing email nurture strategies.

Some teams caution against managing many multiple threads open with any one person — marketing, sales, product, support, and other might all have legitimate reasons to email a customer today. To manage this, teams setup suppression lists to block sending email to an individual if they’ve been sent an email recently.

However, this also throttles high priority circumstances. The result is teams have to work around your suppression rules and send anyway — negating the value of the suppression list in the first place.

This also misses the context a person might have with an individual. If suppression is set to 24 hour intervals, then sales and marketing might be able to send a daily lead nurturing email — for many people, this is too much.

It’s okay to have multiple threads with someone if you have a relationship there and someone else (not something else) is adding voice there and the emails aren’t all on top of each other. For instance, an automated email to update credit card could be supplemented a day or two later with an email from an account manager.

Flip your thinking around.

Instead of "who can I send my email to?" think "what should we email this person today? If anything?". It's about scheduling emails that fit each person's individual customer journey, not building segments to blast out messages.

To create this, we notice teams combine a global suppression rule (e.g. Email sent less than 1 day ago Subscribed: False ) with nurturing triggers. The are rules based on recent activity, recent emails sent, and the primary goals for nurturing that particular user. For example:

  • Nurture: Profile completion
  • Nurture: Salesforce Integration
  • Nurture: User invite
  • Nurture: G2Crowd Review
  • Nurture: Upsell opportunity

These nurture rules and segments can then be synced to email tools for drip campaigns, or to individual sales reps for personal email outreach. When the individual is "eligible" for another email (according to your rules of suppression and nurture) they will be enrolled.

When creating drip email campaigns, more often we see shorter, simpler, highly specific drip flows working better than longer, generic drip sequences when orchestrating with many other types of email. This reduces the chance of emails overlapping and means drip campaigns themselves can be more personal and responsive.

For instance, instead of a long, generic onboarding email series, break it up into many short, simple drip flows based on specific objectives (like your personal emails — see Tactical Takeaway #1) and previous user action:

  • Account creation
  • Adding integrations
  • Configuring XYZ feature
  • Inviting collaborators

Voila! You’re now able to manage your email campaigns around the customer

Best fit criteria for email personalization at scale

Everyone who is sending email ought to be trying to send more personal messages. However, the data-intensive setup and templates are only worth it if you're expecting to beat the value of improving your emails in general. However, we see a few criteria where this strategy has been particularly effective.

1. Email (as a channel) is already working for you

You have a substantial opt-in email list through newsletter subscribers, product signups, user invitations, and so on - at least 10,000 "real" contacts. You're continuing to see subscriber growth month-over-month. Your email sends are already driving conversions and revenue.

2. Emails (individually) are underperforming

Your email open rate is less than 20%. Your email click-through rate is less than 20%. Your unsubscribe rate is more than 1%. You have plenty of people interested in hearing from you, but the emails today aren't hitting the mark.

4. Your emails are sent by more than one source

Marketing automation. Product "transactional" emails. Sales reps. Support reps. Survey tools. Email newsletters. The collection of tools which can communicate with your customers is unwieldy.

Results we've seen

Since so many emails reaches so many aspects of the customer lifecycle, it's hard to attribute personal email to have fundamental changes through organizations. However, with respect to email sends themselves:

  • Tripled open rates
  • 4X click rate (clicks from the total list of people sent the emails)
  • Significantly fewer unsubscribes (partly by sending less email overall)
  • Increase in email forwarding

We send personal emails about smart, data-driven growth

Subscribe for more "Spotted" playbooks, case studies & best practices from the teams we work with at Hull.

What you should do now

  1. Request a custom demo - and see how to unify & sync all your tools, teams & data (like we did for all these companies), or book a demo with a product expert.

  2. If you'd like to learn our best practice for customer data integration, read our free Guide to Getting Started with Customer Data Integration.

  3. If you enjoyed this article, perhaps your team will too? Why not share it with the links below.

Ed Fry

Prev 'Ed of Growth at Hull, working on all things content, acquisition & conversion. Conference speaker, flight hacker, prev. employee #1 at inbound.org (acq. HubSpot). Now at Behind The Growth

If you've questions or ideas, I'd love to geek out together on Twitter or LinkedIn. 👇