Closed joethreepwood closed 9 months ago
Had a chat with Charles today and have redone the flow with some changes. I've updated the post above, and added a quick explanation of the concept.
Changes are that I've removed the repeated activation steps for each product, so that we now only check for activation of the initial product. I've also introduced a fall-back email for the first step, a final email for the end of the flow, lengthened some time delays and added a new step so we only try to upsell users who have successfully activated.
@raquelmsmith let me know if you have any strong thoughts here, otherwise I'll start drafting content and adding detail soon.
This is great! I'm excited to get more activation-focused stuff in there 🤓
For the activation emails, IMO that should be the very first thing we send after the welcome email. Currently it waits 5-7 days to send the activation email, and by then it may be too late, they've moved on to another company's product.
Before building this all out I would be (apologetically) curious about our current conversion rates on the types of emails that are being proposed here. Types I see are:
What is the reasoning behind inviting people to try out betas? That doesn't seem like a super necessary email in this flow - instead I think it could be a CTA at the end of changelog emails, where it's more relevant if we're announcing a beta for a feature/product.
Most of the existing content works well. Where it doesn't, the plan is tailor the content to the users more rather than to remove the email. Our open rate across the entire flow is 49%. CTR is 3.5%. Conversion to billing product activated
within one week is 1.7%.
For the activation emails, IMO that should be the very first thing we send after the welcome email. Currently it waits 5-7 days to send the activation email, and by then it may be too late, they've moved on to another company's product.
I softly disagree here because
It will be swayed somewhat by the fact that Onboarding 4.0 is still relatively new. Take the below with a small pinch of salt due to Xmas.
All these figures are avgs over the last 30 days unless specified. As a baseline I'd normally expect a 40-45% open rate and a 3-5% CTR.
Open rate: 60% avg CTR: 8% avg
Open rate: 64% avg CTR: Varies depending on founder (approx 20%) and engineer (approx 5%)
This gets us approx 75 clicks/subscriptions to Substack per week. I can figure out the full subscription rate here, but it's error-prone and a bit of work.
I don't think we should merge this with the community invite, as the newsletter and community exist in different places and for different audiences.
These were new in Onboarding 4.0. We felt non-technical users would greatly benefit if they knew how to create actions without code. If they have created an action they get a general advice email. If they haven't, they get advice on doing that.
Sample sizes skew things a lot here. Open rates average around 50%:
CTR is around 3%.
So, we can probably optimize this content, but the open rate seems solid.
These are currently only sent as fallback emails. If a user is subscribed to every single product, only then do they get these. In Onboarding 5 I've added them as best practices we want people to be aware of and to help break up the emails away from being salesy. We mention the betas in changelog when relevant, but I think a general encouragement here is good too but it's not a strong feeling.
Community invite Open rate: 55% CTR: 4%
Beta invite Open rate: 58% CTR: 6.7%
I feel like this is pretty solid for now, and it'll likely improve if we move it to not be a fallback.
There's a lot of these, so this is quite involved. The short version across all of them is:
Open rate: Approx 45%
CTR: Approx 1.5%
Conversion to billing product activated
(below) is around 1.5% too.
So, these aren't performing particularly well. The hope with Onboarding 5 is that we'd tailor them a bit more based on the first selected product, so that's the plan for improving these.
I've now plugged in all of the logic for the flow, so next week I'll start looking into content creation a bit more. That's likely to last a sprint or two.
@raquelmsmith as per feedback on the activation tests, I'm going to create a branch for the welcome emails here.
After 1 hour, non-engineers will get this:
While engineers will get this:
I've made some further logic tweaks.
The notable changes:
OK, new and hopefully final version.
An advantage of this approach is that all users have a more predictable email path. No matter what, we now know all users will get a total of 6 emails over the course of between 5 and 9 weeks.
I'm considering this ready to launch now. I'll give everyone 24 hours to either have a look at the content/flow that they may not have already seen (or to ask for more time), then I'll start rolling it out.
Cool thanks Joe!
Context
Part of https://github.com/PostHog/meta/issues/171 Follows https://github.com/PostHog/meta/pull/150
Apologies for the readability of this. If you feel there's a better way to review this, or if you'd like to hop on a call to feedback, LMK!
Concept
Onboarding 4 focused on introducing upsell content, to get people paying for each feature.
Now we want to focus more on activation. Get them using us. Then upsell.
At the same time, we also want to personalize the emails more to help people get started.
And we're introducing a revised in-app onboarding that lets users choose a single product to get started with. If we can harness an event from that, we can tailor email onboarding to focus on that firstly and then take them through a flow tailored to them - with subsequent upsells and emails about how products combine.
Implementation
This flow is exponentially more complex than the previous flow. Therefore, the current version just includes logic. I want to get feedback on logic first, before I write dozens of emails.
Here's the high level view of what this flow does that's new:
If users don't select an initial product they move through a condensed version of the above without the first activation check.
Here's what the full flow looks like.
Requested feedback
Everyone has opinions on email content and we are not at that point yet. This is purely to figure out what's practical, and logical. As such I'm looking specifically for feedback from @raquelmsmith @xrdt and @charlescook-ph on a handful of points: