How Youtube 5x’d our Activation Rate

April 30, 2024

Our Youtube channel went from video storage to driving 70% of our sign ups in 6 months.

How? A ton of grinding and some luck – but it taught us one key insights about activation that changed the way I think about growth.

Activation starts before onboarding.

When we figured this out – our monthly sign ups quadrupled, our activated user base jumped from hundreds to tens of thousands, and our self-serve revenue went from a side thought to doing half our overall revenue.

Let me walk you through it.

But first – some stats:

  1. Our Youtube monthly views grew from a consistent 5k each month in Feb to over 50k by December. This translated to a 5x increase in monthly sign ups for Voiceflow.

2. Our product activation rate increased dramatically. Going from <3% to a peak of 17%. Driving more conversions than we had ever seen in the business and creating a thriving community outside of the core app.

3. Our Youtube subscriber count grew from 4000 with an average of 100 new subscribers each month to 10,000 with an average of 800 new subscribers each month. Allowing us to start compounding the traffic we were getting.

Lets walk you through how that happened.

Last year we had a pretty big challenge. We had been focused on enterprise for the last 2 years and no longer had a strong PLG motion, but needed to change that quickly.

Voiceflow is a platform to build mid-high complexity AI assistant. Previously this market was limited to enterprise, but with ChatGPT becoming more mainstream and stronger AI models (LLMs) becoming available, the market for AI assistants was expanding from something that only big enterprise companies had to a commodity that every SMB and Midmarket customer would want.

We needed to find a way to get more people launching production AI assistants on Voiceflow in a self-serve way. But Voiceflow is a fairly complex tool to use and learn.

On the enterprise side our activation motion was fairly high-touch (I built our Enterprise success team). We would spend a few weeks doing office hours and demos and guide customers through getting their first MVP stood up on the platform. It would usually take 2-3 months for a customer to fully get it and transition from beginner to expert.

This was okay for large enterprise deals with a lot of skin in the game but would not scale. We were already feeling the impacts of it on our self-serve user base.

To give some tangible numbers, our self-serve activation rate was really low. We had tons of sign ups but the number of people who were able to understand how to properly use Voiceflow and launch to production in 30 days was <3%. (note: I’m a disciple of Lauryn Isford at Airtable – that a truly meaningful activation metric is likely going to be in the single digits, and it should constantly be shifting down the funnel).

The main cause of this is that Voiceflow is a platform. Its the best tool available for building a mid-high complexity AI assistant with your team, but takes a while to master and has a steep learning curve. Similar to Webflow for websites or Figma for design. There are plenty of different ways to get started and plenty of different things you can build, but getting to your ideal end goal is often highly customized.

If we could move this metric even up to 5%, it would cause a pretty significant impact downstream.

Where did we start?

When I moved over to lead our Growth side in February to tackle this problem, there wasn’t really a clear point to start from.

The obvious one was to see how we could create a smoother onboarding experience. Replicate what we had created for our high-touch enterprise clients in a low-touch way.

But that didn’t really work out.

My team spent a ton of time tinkering with pop ups in the onboarding flow telling users how to get started, defined milestones, and aligned our email flows to them.

We looked at our data to figure out if there was some inflection point that we could drive people too and tried ways to optimize that in the app with notifications, re-activation loops, and all the things that you read about.

Some of these helped move it slightly upwards – but honestly nothing was really moving the needle for us. Our goal was to make big game changing improvements, not do micro-optimization to an existing flow.

Thats when one of our experiments showed us the one key learning that was holding us back from having a significant impact.

Activation starts before onboarding.

When you’re building a platform that can do many things, the hardest part is aligning your users and product with one end goal. People come to your product with an idea in mind but don’t have a roadmap to get there.

People saw Voiceflow somewhere online and had different ideas of what they wanted to build and no roadmap to get there. They wanted a discord assistant that could moderate their community, a personal assistant that could manage their day, a support assistant that managed all their customer communication. But all they had to figure it out were generic tutorials and tooltips focused on features not outcomes.

This resulted in most users coming to the product, playing around a little, and then leaving because they couldn’t figure out how to get to their end goal.

If we wanted to make a real impact on activation – that was where we needed to start. Align people on a use case, and provide them the step by step resources to get there quickly.

(Huge shoutout to Reforge, Tom Willerer, + Hannah & Curt from Openview for pointing me in the right direction here!)

Reforge has this excellent chart on the 4 fits of activation, where it looks at Audience Fit, Promise Fit, Intent Fit, and Knowledge Fit.

A key part of what we were missing is the promise fit. Its aligning a users expectation of what they can achieve with the core value of the product. Then giving them the resources to get there.

Thats where we started playing around with templates, demos, and tutorials.

We took our efforts to pre-sign up and tried to show people what they could build with Voiceflow and made a ton of different templates, long form step-by-step guides, and code repos to help them get started.

This was the first thing we did that made a significant difference and started compounding growth in our activation rate.

Here’s how we broke it down 👇🏾

The proposed experiment:

If we showcase really specific use cases with Voiceflow (ie. turn a PDF into a research assistant in 3 minutes, an AI discord bot for moderators in 5 minutes) and pair them with hyper specific resources – will that improve our activation rate for those customers?

We did the following:

✅ Created short demos (a few minutes) and tutorials for specific Voiceflow use cases.

✅ Distribute them on our personal LinkedIns, using our network to get some initial views.

✅ Created templates showing exactly how to build it.

At first it felt a little gimmicky. These are pretty simple use cases and you can do so much more on the platform. I felt like we weren’t really showing off all the bells and whistles and were missing the more complex stuff that really separated us out (spoiler: no one cares about them at the beginning)

But once we were consistently pushing them out – these posts started getting some traction. First a handful or users, then 10s, then low hundreds.

It wasn’t anything dramatic – but the real impact was after sign up. Users who signed up from our tutorials had a 10x activation rate compared to others (yea you read that right – 10x).

So we doubled down on distribution. Lets find ways to get more people watching these videos and figure out where they get stuck. Then rinse and repeat.

Taking It to YouTube, Starting a Discord Community and Refining the Process:

Everything we made, we also hosted on YouTube so it could be discovered easier and over time.

We already had a decent following of 4000 subscribers from our OG days as an Alexa skill builder, but all of our content was geared at existing users with generic tutorials, feature demos, and webinar records.

Youtube was an educational channel for us, not a discovery channel and we were treating it that way (ie. just reposting videos on there from our docs, product releases, etc).

At first we were just letting the videos we posted be discovered organically. Getting a few hundred views max. But then we started being more intentional.

We did keyword research via SEMrush and Keywords Everywhere to see what were emerging topics that we could build tutorials on/competitive keywords that our Youtube videos could appear in the top search for.

(Shoutout to Steph Smith’s Doing Content Right ebook for bootstrapping my SEO knowledge – great resource for brand building and finding trends)

We saw our view count start to creep up close to the 1k mark and eventually pass it. But we didn’t stop there.

The green line represents views from new users vs subscribers. After we started optimizing our videos we had a steady increase in the number of new users watching our videos.

Whenever someone signed up from a tutorial, we targeted them with a personalized invite to our discord, and offered 15-minute calls with a ‘product expert’ (aka. me) to understand where gaps were and help them.

I think at the peak I was doing like 20-30 calls a week talking to users and figuring out what they wanted to build/where the gaps were. Didn’t matter who they were, we were talking to them. This helped us learn at an extremely fast rate where all the hurdles were.

We also started our discord community around that time as a way to talk to these users and get them started faster (our discord community also went from 100s to 1000s – but more on that in another article). That worked really well. It let us talk to users and answer their questions async with loom videos. We started to see the same questions pop up over and over again which made it very obvious where the gaps where.


So we worked double time to create and pump out new, longer form videos and tutorials. We thought we needed to do catchy 5 minute clips – but it turns out once someone decided to use Voiceflow for a specific person they actually wanted long form 30+ minute tutorials filled with how to debug and work through problems.

A big wave of Growth:

As we started activating more users something really unexpected happened – we started seeing videos of Voiceflow pop up all over YouTube (not created by us)! We actually started getting more videos created by community members than we were creating ourselves.

It turns out that some of our new users turned into champions. They loved what they built and started making their own Youtube tutorials to help other people learn how to use Voiceflow too.

It was surreal to see and triggered a growth loop. New users found us on Youtube, joined discord, got activated, and created content. Leading to more people finding us and so on!

Preparation meets timing ✨

We started seeing this impact our in June/July. Not only did we start getting more sign ups, but our overall activation rate started going up and we had our first significant increase in self-serve MRR in 1.5 years.

Then we started to catch a trend that pushed us up dramatically. As GPT became more robust, more businesses started wanting AI chatbots. This expanded our market from primarily enterprise (you needed to have a big team to create a powerful assistant) to everyone from SMBs to MidMarket.

With this rise in demand came a number of brand new agencies that were looking to service these businesses. These were smaller existing agencies in the digital marketing space that started seeing demand for AI chatbots from their clients. Mostly for lead-gen and support use cases.

We started seeing a flood of them come in as ‘AI Automation Agencies’ started to become a trending search term.

We did as many interviews as we could with the new users joining our community in waves and kept hearing the same set of names and communities that were driving them to Voiceflow.

Two of our community members mentioning that they found us through the same discord community

So we doubled down.


One of the big factors driving this new growth of agencies were a set of Youtube influencers who had agencies and were walking others through tooling and how to adapt and serve this new market.

We started doing a bit of digging and found a few of them that were still starting out but had a serious vision for the future of AI assistants and really aligned with our values of building serious businesses for the long term.

We decided to take a big bet on Liam Ottley and struck a partnership agreement with him and his team. We shared the same vision for the future of AI assistants and found that Liam not only had his own agency servicing large clients, but had strong conviction and a unique perspective on the future of AI assistants for Businesses aswell as the skillset that agency owners would need to gain to adapt to this market.

This allowed us to double down on the momentum and inspired other Youtubers in this niche to start adopting Voicefow as the gold standard for building AI assistants.

The result? A 5x increase in our MRR, turning it from a dead channel to our newest revenue driver and a production user base that is compounding every month.

What now?

The last 8 months were a whirlwind but Its not all sunshine and rainbows, while we have way more people entering our top of funnel with the goal of launching a production assistant we still have similar challenges later down the funnel with this level of growth our churn rates went up and our community grew way past what we could handle.

While its clearer than it was before – users still come across the spectrum of use cases and technical skill. For us to continue to grow we need to adapt to new challenges.

  1. How can we learn which users and use cases are the stickiest and narrow our top of funnel to them
  2. How can we attract the niche set of users that want/need a complex AI assistant that aligns strongly with our value prop
  3. How can we better support our user journey from they 1st production interaction to their 10,000th.

We learned alot with the large number of new users we gained and our focus for the next 8months is to continue narrowing down our Audience Fit and Promise Fit for the core use case of medium-high complexity AI Customer Experience. This is where we’ve seen the strongest fit in customers that have come to Voiceflow and launched production assistants.

Leading a growth team is a bit like working in waves. You try a number of different experiments, double down on the ones that work, find ways to build momentum, then learn, slow down, and repeat the process with a bit more focus each time (and pray that luck/timing is on your side).

Always stressful, but never a dull moment.

If you want to learn more about this journey and the future of AI assistants – hit follow on LinkedIn and check out our Youtube channel @voiceflow!

Key Takeaways

  1. Alignment: We showed what you could build on Voiceflow to align expectations.
  2. Guidance: Videos + templates = step-by-step guide.
  3. Optimization: We spent a ton of time talking to users and making it easier to go from sign-up to a live app.
  4. Partnerships: Find out how your customers are finding out about you and double down on people or complimentary businesses that are sending them your way.