Lessons learned from the Machine Learning in the Real World launch

Lessons learned from the Machine Learning in the Real World launch


Hi there, it's Sophia here.

I am chatting with you from sunny San Diego today. I'm here to visit some family, chill out a little bit and as promised, I have lots of things to talk about from the launch I just finished up.

As anyone knows, launches are a huge undertaking that involve multiples of emails, landing pages, and kind of everything in between that stitches it together.

And so I know I usually talk about onboarding, but I think that since I hadn't done launch copy in such a long time, it's really important to just kind of remember why we need to keep those skills and how they relate to onboarding, in order to learn because if we're only reading onboarding emails, we don't necessarily get the full breadth of why we send email in the first place.

So I kind of wanted to share with you some lessons learned through the launch.


I was working on a launch called Machine Learning in the Real World.

My business partner is a machine learning consultant. He really wanted to start training other people to do the work like him, so that they could bring value to their organizations and, we had an idea and we weren't really sure how to bring it abou.

Luckily after, you know, 90 days of validated learning and iterating, we have a great cohort that we're starting out next week, and I know that the launches will improve with time as we continually refine the offer and really nail down what people are trying to get with machine learning, which at this point, we have a guess that they want to drive value, be the smartest person in the room.

But...we don't know.

It's kind of one of those things when you're starting a business, when you're doing a startup, you have your gut intuition, which is a valid way to make decisions. But then there's also you need data to see if people are either interested or committed.

And so here's a couple of lessons I wouldn't say that I learned, but I remembered and reminding all of you today to just talk about.

Lesson #1: If you have an idea for anything and that means anything, start building a list.

If you can't get people to move and sign up for an email list, you're not conveying the value well or your idea just really isn't going to fly.

So, after 90 days we were able to grow our list by 248%.

We did a book giveaway, some free workshops and of course got some students to convert into the paid version of the workshop for anyone who wanted to go further, that was beyond the free workshop.

And so this is what we call a seed launch. That's the Internet marketing term.

If you want to know a little bit more about it, you can check out Jeff Walker's Product Launch Formula. I think he has some free content that you can check out where he kind of goes into detail what the seed launch is.

(Note: I'm not an affiliate for Jeff Walker, I just got a lot of value from his content!)

And it's really just meant for business owners who really want to validate learning before they kind of go in deep and create a product.

They really just want to get insight from what customers are really looking for, as opposed to anything else, it's just really important to do that validated learning even if it costs a bit of money in the short term, in order to pay off in the long term as opposed to spending all this time creating product and then having nobody buy it. That's just really tragic. 

So, bottom line, if you have an idea for anything, just start building a list.

Lesson #2: Re-engagement campaigns work super well.

We had an old list that had been lying around, it had been 18 months old.

It was from a webinar that we had done and we weren't really sure about whether or not those leads were dead or not.

Nowadays with GDPR, there's like an extended or a very limited amount of time that you could hold this information, especially for subscribers from Europe and we were working, we didn't have a huge European clientele on that list. 

So we figured out that, you know, it was probably a good idea to just re-engage this list from other past campaigns we had done.

And so we were able to recover up to 27% of the people, and that's including people who had come into the campaign and then unsubscribed again, as a result of the launch emails going out.

And granted, we were blasting them with a lot of email.

Some people would say unsubscribes are a really bad thing when I would say that, you know, at least they're doing something because if they're opening email and then they're not taking clicks or doing anything, then you haven't done your job to convey your value so that people can make that decision right away to say, "Is this valuable to me," or "Is this something that is not really my priority right now?"

And so, you know, I wouldn't take it personally. I think there are moments when you see unsubscribers and you're like, oh my God, what's going on?

But for us it was a really good thing and you know, without re-engagement, we wouldn't have rebuilt up our list so quickly.

So if you have old contacts, I would highly recommend getting them on a list.

Depending on your email provider, you can do that in a various amount of ways to get them onto a new campaign or at least have them to take one action to say, yes, I'm still interested.

[We were using] ConvertKit, but it could vary for any other ESP you're using. I'm sure, if you're a SaaS company, you have other ways of doing that behaviorally.

But that's just a really basic run-down way of saying that re-engagement campaigns work well.

And I do have an article that I wrote, in order for you to kind of get a sense of what I think re-engagement campaigns are.

Lesson #3: Don't rely blindly on swipe files or first-hand experience. A/B test everything. 

And the last thing that I want to say is the subject lines that we used. During a course of a launch, especially when you're still trying to validate an idea, it's really difficult to understand what your persona is.

And when I say that, you know, we love to create personas. I think they're personally fun.

You kind of base them on people you know, you start combining characteristics and think about the car they drive and what they like to eat. Some of that stuff really doesn't matter.

And some of you could argue, okay, don't use personas, use jobs to be done, but in this case we're still learning all about our persona because some people aren't really willing to share, but you can test what motivates them and what doesn't motivate them based on the subject lines that you provide in your email.

So, I was using all sorts of copywriter tricks. If you've worked in email as long as I have, you have all sorts of tactics in your arsenal in order to guarantee opens and click-throughs.

And some of my swipe stuff didn't actually work. It worked in the past for other types of personas, but for this particular persona, software engineers and analysts, they didn't work.

So it's a lesson to me in order to not necessarily rest on my laurels and say, you know, I know, I'm going to use my guts to say that these types of subject lines worked in the past and they work because you'd never actually know because you have different cohorts of people, different users, that sort of thing.

So, the subject line that actually won and got the most opens was something as clearcut as...

"We're open, Machine Learning in the Real World Registrations."

I kid you not. And this is after gems like, 

  • "Why employers are dying to spend money on you,"
  • "Busy this summer?" "17 hours left, bonus added" or,
  • "Will machine learning take away your job?"

My theory is that this particular persona that we were going after, which is software programmers and analysts, they just prefer it to be straight.

Don't do salesy, don't hype, don't use open loops.

But just kind of say it like it is because they're engineers so they don't really have time for it or maybe they've been sold to so many times and now they're kind of immune to it.

So their bullshit reader's kind of going up because you know, engineer personality types and everything's very black and white. Everyone's very skeptical. If it's too good to be true, it probably is.

And so that's kind of something that you have to keep in mind when you're creating launch campaigns or even onboarding, just that you're always learning things about your persona and that learning never stops.

So, you might say, you know, the subject line worked six months ago and I would argue A/B test everything even if you think, you know, in your gut which one you win. You will be surprised.


So those are my kind of three lessons that I took away from Machine Learning in the Real World. So we talked about swipe files and subject lines. We talked about re-engagement campaigns and then we talked about building your list in the first place.

If you have any sort of idea, that's the first thing I would recommend doing before you start writing a line of code or talking to customers, just set up a landing page, try to capture some interest with saying the value and just kind of seeing what people will do because that's the best way to learn and that's the best way really to start up any idea, whether you're an established startup and you're trying to launch a new product line, or you're trying to go out on your own, these are lessons that we need to always keep in mind as email marketers.

So with that,  did I miss anything?

Is there anything else you want to know about the launch?

I am super excited for this cohort to start next week...

I'll be sharing some more lessons learned of that. Not necessarily email, but you know, we're in the business of using email to make a better businesses. 

Why you won't win your dream house with an offer letter (and what to do instead)

Why you won't win your dream house with an offer letter (and what to do instead)