Do you know what’s painful? Spending about three weeks (or more) taking an email from conception to deployment …… only to realize later that there was a mistake that kept getting delivered to your client’s inbox.
I know this feeling all too well – I’ve made a few mistakes myself over the years. While none of those mistakes resulted in an apology email, they still haunt me. Want an example? How about this subject line on behalf of Home Depot.” Lightning’s New Low Price.”
Lesson: Waiting until the approval is complete and then manually entering a different subject line exponentially increases your risk of failure. (Off-topic: How cool would it be if Home Depot actually sold lightning bolts?)
But many times, it’s more than just a contextual spelling or grammatical error getting through. Sometimes it’s a user experience on a landing page, a broken link, or even just a backend issue.
At Grammarly, we’re committed to learning from our mistakes – especially grammatical ones. That’s why we’ve built a bunch of active safety nets to help us keep a close eye on the quality of our emails.
Want to learn more about these programs? We’ll tell you below.
#1 Aggressively integrate proofreading into your workflow.
This is number one for a reason. It’s your first, middle, and last line of defense. At Grammarly, we use both our product and a human copy editor to be doubly sure nothing slips through.
To give you a better sense of just how careful we are, check out our high-level workflow:
- QA (Quality Assurance)
We rely on both software and a human editor for a few reasons. For one, we’re firm believers in “eating our own dog food.” Our team of world-class linguists are constantly iterating our product to ensure it’s something we can get behind. But we’re also firm believers in the power of context, which is why we have a full-time or freelance proofreader run through all of our email copy to ensure that no stone is left unturned.
The outcome? Quality content that feels really human.
#2 Create efficient and actionable customer feedback loops.
After you send, keep your eyes and ears out for qualitative customer feedback. These days, it’s fairly easy to set up a loop via Slack (or Hipchat, etc.) that makes real-time user feedback actionable for your organization. Here’s how we do it:
- Scenario A: Deploy Emails > User Engages > User Provides Comments on Social Media (positive or negative)
- How we solve for it: Our social media team helps us monitor company mentions on Twitter for several hours after a large campaign goes live.
If a user is having a buggy experience with the product landing page, for example, then we work with our social media manager right away to develop a fast, helpful response. If the problem appears to be a corner-case or temporary issue, we alert the product team right away.
- Scenario B: Deploy Emails > User Engages > User Replies Directly to Our Email > Reply Automatically Creates a Zendesk Ticket > Customer Support Team Manages Tickets
- How we solve for it: The secret sauce for feasible automation here was ensuring that any replies to our marketing campaigns automatically create Zendesk tickets. Our support team autonomously handles the cases they can solve themselves or have a known resolution for. Anything qualitative is shared via a #Marketing-Replies Slack channel that the email team monitors along with the support team. This process facilitates not only quick resolutions but also assistance with copywriting for responses that may need extra nuance.
The biggest piece of advice here is to look for feedback and the heartbeat of every single large send. This feedback may not affect your team directly, but by tracking it and relaying it to appropriate teams, you’ll increase your value — and that should do wonders to strengthen the company perception of the email team. You’ll also learn about phrases that may not resonate with your audience so that you can refine your style over time.
#3 Balance autonomy and redundancy
The most important part of your final approval process is a system of checks and balances.
Try to keep it as lean but potent as possible. Have one main stakeholder checking all the variables, and specialists ensuring quality for their niches. Here’s what this process might look like across teams:
- Coder and/or Designer: Checks email rendering across clients (we use Litmus), implements all links, tracking, alt tags, etc., and loads the HTML into the email platform.
- QA Team: Tests all the above variables across possible use cases in production (i.e., they deploy from the email platform).
- Main Project Manager: Checks all of the above before deployment — at least at a high level — and ensures final versions are sent to any necessary stakeholders (your boss, legal team, proofreader, etc.). The project manager is likely to be your email marketing person.
- Proofreader: Gives final approval of creative. Ideally, the final test send should not be the first time your proofreader is seeing the copy.
One additional piece of advice: Don’t put the final QA check solely on the project manager. And if you are the project manager, don’t cut corners in the interest of efficiency. Nowadays, with the ever-shifting variables to optimize for in email marketing, there are just too many things that can go wrong.
#4 Choose a cadence for revisiting and auditing automated emails
This one’s short and sweet: Don’t set it and forget it. Revisit your life cycle creative like it’s your core product. Shoot to revisit your life cycle at a cadence that makes sense and use a good decision framework for business impact and UX.
At a minimum, don’t go longer than a year without revisiting live creative.
#5 Review samples thoroughly for dynamic emails
Another short one: If you’re sending out personalized emails with a lot of dynamic content, then make sure you preview multiple variations ahead of deployment.
Grammarly’s weekly statistics emails are 1:1 and have over 25 unique data slots. There is no way we can preview all variations during our QA. However, we do have a test list of employees that we use to eyeball all of these touchpoints before hitting send.
#6 Don’t let a robot do all the writing
There are a lot of cool third-party technologies in the email world today. And while we’ve seen our fair share of demons, we try not to rely too much on automation for copywriting.
This isn’t to say you shouldn’t perform meaningful experiments to identify new ways to optimize your email copy. But allowing machine learning to take the wheel entirely could jeopardize the quality and consistency of your email strategy. After all, these automated tools usually optimize individual email campaigns but aren’t able to think about the entire customer journey.
Evaluate these tools with extreme caution. While having a robot spit out 100 subject lines for you sounds great, you should always check to make sure that the outcome reflects your brand’s message and voice.
But don’t let this perspective scare you away completely, though. There is likely a hybrid process that still allows for steps one through five above to be baked in for quality control. You should vet these opportunities yourself and make an informed decision. But if you start selling lightning instead of lighting, don’t say I didn’t warn you.