As @Shay mentioned I will be taking over this thread. I wanted to post an update and ask a couple of questions to the group here. Here is the basic functionality we will be supporting in the Beta:
A simple 50-50 split between 2 variants
Ability to start and end tests on the same automated email (without cloning and republishing).
Ability to see a history of the tests run on that automated email.
Manually choosing a winner based on open/ click data.
When digging into this project we realized that AB testing for Automated emails differs from regular 1-off batch emails. The main difference is that Automated emails have the potential to last forever. Thus we needed to support the ability to start and end multiple tests on an Automated email. Tracking those changes and tests over time. Here are some questions I had for the group:
Q1: How do you all keep track of the different experiments and tests you have run today with Batch Email? Or with other tools (website pages, CTAs, etc.)?
Q2: Are their other metrics that are important to picking a winning variant other than Open & Click information? What are they and how do you solve for that today?
Q3: What are the other important pieces of functionality that are missing from above? And which are the most important to you? (ex/ configurable percentages, multivariant, calculations to pick a winner, ability to have HS pick a winner, etc.)
We are getting very close to having a Beta. Once we have it I will post a form in this thread in order for you all to join. We will then be quickly iterating and adding more advanced functionality.
Q1: We use excel to track tests and how they did. Would love something built-in.
Q2: We would love to narrow down click information to only track winning variation based on certain links clicked. For example, some emails may have a much higher click rate, but when you look at the clicks they are on the header, social media icons, and other things that mess up our results. We would love to have metrics just on the cta/links we care or are testing.
Q3: We would love to test more than 2 emails. For instance, in Marketo, you can test multiple things by either having a randomizer (currently workaround is assigning to round-robin owner). Marketo also you can test up to 5 subject lines, which is super nice because often times we have more than two variables.
Thanks for keeping this moving - really excited to get this functionality rolling. Just to confirm - are you just testing beta-ing automated emails against each other or will you be able to beta test full workflows? Regarding your questions:
Q1: How do you all keep track of the different experiments and tests you have run today with Batch Email? Or with other tools (website pages, CTAs, etc.)? We tend to track testing as one-offs either through batching email or by running an automated test a certain way for a period of time and switching to a challenger layout and seeing if performance increases (obviously not a true test but often the best we can manage without true A-B automated email or workflow testing).
Q2: Are their other metrics that are important to picking a winning variant other than Open & Click information? What are they and how do you solve for that today? For us, while front-end metrics like opens and clicks are good initial barometers, we ultimately want to look at conversion information, which is often tied to actions on our site. So in a perfect world, we would be able to test two workflows against each other with the same goals and see which are performing better at converting on that workflow goal - this would allow a more holistic /accurate look at performance vs. just comparing open/click rates for a singular or automated email.
Q3: What are the other important pieces of functionality that are missing from above? And which are the most important to you? (ex/ configurable percentages, multivariant, calculations to pick a winner, ability to have HS pick a winner, etc.) All of these would be great features, additionally see above re: full workflow testing with customizable
@jaleonard19 Thanks for the update! In response to your questions...
Q1: How do you all keep track of the different experiments and tests you have run today with Batch Email? Or with other tools (website pages, CTAs, etc.)?I pull all of the metrics I need out of HubSpot and Google Analytics into excel. I find this to be the easiest way for me, and then the data is at that point is "static" and won't change when I revisit it. I can also easily combine results from multiple tests of the same thing (as I never only test something once), and see what the overall results are of what I was testing, in addition to each individual test.
Q2: Are their other metrics that are important to picking a winning variant other than Open & Click information? What are they and how do you solve for that today? Usually one of those metrics is the main KPI, as you can't control what happens after like on your website, but it can depend on what is being tested. In any case, we're definitely looking at conversions and revenue as well, which we get from Google Analytics as that is our main source across all channels when it comes to conversions and revenue.
Q3: What are the other important pieces of functionality that are missing from above? And which are the most important to you? (ex/ configurable percentages, multivariant, calculations to pick a winner, ability to have HS pick a winner, etc.) I would say all of those examples would be great to have, as well as the ability to test the workflow itself, so we're able to test other things than just an email in a workflow, like delays (i.e. test 3 days against 2 days) for example.
Hi, just looking into email workflows A/B tetsing it would be helpfull to know when HS are looking to launch this tool its works in campaign monitor and Mail Chimp and rtaher than using other tools it would be good use wiithin the CRM tool.
Upvoting this idea. I would love to see a solution to test emails on workflows.
For the moment I have to create seperate test lists to test the idea. This is ok for simple drip campaigns but can bee too complicated for more intuitive smart ones.... it's be great to have simplified process inplemented here.