Implementing an ESP for digital products - Part 3: Launch
A 3-part guide with downloadable templates, playbooks, and tools.
Selecting and implementing a new ESP for your company is a massive undertaking. It’s often costly, time-consuming, and involves many different teams so is high visibility.
How do you get it right?
Read part 2 here:
Launch
Launching your first campaign is nerve-wracking. Push through. Limit risk by using a phased ramp-up. 10% of the audience one day, 20% the next, then 50%, and so on.
Do you have a dedicated IP address? You may need to warm up a new IP for sending. Figure out what emails will perform the best and prioritize those to launch first. You’re looking for clicks and engagement. Welcome and onboarding emails work great.
Calculate your average daily email sending volume and work backward to figure out your IP warming schedule. You want to reach your desired daily sending volume in 2-3 weeks, max a month.
General guide:
Start with 50-150 emails on your first day and double the sending volume daily until you reach ~15k
Add 10-15k per day until your reach around 50,000
Start adding 25-50k per day until you reach around 250k
Add 100-150k per day until you reach 1MM day
If you’re sending more than 1 million emails a day you can figure it out from here
Every day during this period you should be checking email deliverability tools to make sure you’re maintaining a strong reputation and not getting blocked. If you notice a drop in your deliverability metrics, keep your daily sending volume the same for a day or two until you notice it recovers and then can begin scaling again.
Use these tools:
https://www.mail-tester.com/
https://mxtoolbox.com/
Received a bounced message from a specific email provider and don’t know what it means? Use the SMTP Field Manual.
How are you going to measure the success of your campaigns?
Use control groups and A/B tests for an incrementality measurement approach.
Incrementality seeks to answer the question if I didn’t send this email, what number of conversions would happen anyway?
Always include a control group in any new campaign you’re launching. If you don’t have a large audience (500K+), don’t add many variants to your tests.
For campaign launches, do a 50% test group and a 50% holdout group. It’s important to get stat sig. results. You should view an inconclusive test as a bigger risk than a failed test. You learn a lot from a failed test.
For campaign iterations, do 40-45% for variant A, 40-45% for variant B, and a 10-20% holdout group. Include a control group. Your environment could have changed, so you need to verify your campaigns are still impactful.
Use a global holdout group periodically. Exclude 5-10% of your entire audience from receiving any email or notifications for a few months. Measure the impact between the groups receiving your marketing and those not, and you can understand the impact of all your work.
Track your open and click rates over time. A decline here means you need to start investigating if something is wrong or broken (deliverability, data issue, broken link, etc.).
Connect your ESP data back to your analytics platform or data warehouse, so you can track campaign performance back to revenue. Put in the work upfront to set up strong analytics.
Build dashboards where you can easily see the performance of individual campaigns and you can track them over time. Building dashboards that eliminate manual analysis will allow for better and faster learning. Work with your analytics and/or engineering teams to figure out how to automate analysis and reporting.
You should now be pretty well-armed for success with your new ESP.
A few last tips:
Don’t be afraid to take a phased approach. Your work will never be 100% done.
There is always more data to add, new tools to integrate, and mistakes to clean up.
This is just the beginning. Think big about the possibilities of your new tool, but don’t get away from the day-to-day execution.
Executing well each and every day is what makes a team great. It’s easy to dream up big and lofty ideas. Bringing them to life is what separates the good from the great.
Remember, your goal is to do great marketing, not to have a great tech stack.
Keep your customers in mind and strive to do work you’d be proud to share with the world.