How we stopped doing first calls at Tweenvest

How we used to do it

For years our pricing page had a "Book a demo" button that went straight to my Calendly. I did most of those calls myself, usually in the evening, because that's when people who invest on their own have time.

The calls were fine. Everything around them wasn't. A lot of people booked a slot three days out and then never showed up, and I had no way of knowing beforehand who was serious. More than once I spent half an hour on the stock screener with someone and found out at the end that they had €2,000 invested and weren't planning to pay for anything.

Testing it

We added Demoloop in the spring, but only for half of our traffic. The other half kept the Calendly button, because I wanted something to compare against before changing anything. On the Demoloop side the demo started right there on the page: the agent opened Tweenvest, went to whatever screens the visitor asked about and answered questions on the way.

We left it running for six weeks and tried not to look too early. In the end only one number mattered: out of everyone who landed on the page, how many started a paid plan.

What we found

On the agent side 2.1% of visitors started a paid plan. On the Calendly side it was 1.6%.

The agent also gave 2.3 times as many demos, so each demo converts a bit worse than one of my calls did. That worried me for about a day, until I thought about who was booking those calls. My calls only ever reached the people patient enough to book a slot and wait three days. The agent reaches everyone else too, including a lot of people who just want a look, and some of them end up paying. Waiting went from three days to about a minute, and about half of the demos now happen after 10pm or on weekends, when nobody here would have picked up.

The other thing I didn't see coming was how much it changed my week. During the demo the agent asks a few things (what they invest in, how they track it today, what's annoying them) and all of it ends up on the lead. Now I open a lead and already know whether it's worth a call.

When the six weeks were up we moved everyone to the agent. We kept a small "talk to us" link for people who'd rather speak to a person, and I'm happy to take those.

Then support

A few months later we put the same agent inside the app for our logged-in users. If someone can't work out how to import their portfolio from their broker, or why a company's fair value changed overnight, they ask, and the agent shows them on their own screen. Most chats end there, and the tickets that reach our team are down about 60%. People rate those chats 4.6 out of 5, about the same as when we answered everything ourselves, so they aren't just giving up. The ones that do come through arrive with the context already written down, so we don't start every ticket by asking what the problem is.