Behind the scenes of launching our own software company

The 3 Challenges and how we (tried) solving them

Hi there reader!

Today I want to show you the behind-the-scenes of growing our in-house software named Betterpic, which we launched today on Producthunt.

The good, the bad, and the ugly with our real metrics.

In case you missed it, here is why we bought the business 2 months ago.

So let’s start with what we call our Monthly Performance Bingo of February 2024:

Betterpic Monthly Performance Bingo

In short, we are growing, improving key performance metrics, and burning cash (yikesss).

So let me walk you through 3 key challenges we faced and how we are solving them along the way.

Table of Contents

Challenge 1: Aligning the teams

When starting, the main challenge was ensuring that our entire Saasmic and Betterpic team was aligned.

Although I tend to believe that we have an amazing team, I noticed comments from my team in late December 2023 that we were doing too much.

This went from selling our own SaaS Financial Template to scaling our in-house podcast. It just wasn’t healthy.

So we decided to do what many people are afraid to do (including myself), kill our darlings and refocus the company.

In December 2023, we held a meeting where we announced one of the following Objective Key Results (OKR) to align all teams:

And for the first month, for some reason, I thought setting up OKRs didn’t make sense for us. Mostly as in January, we were still in the middle of killing the things we weren’t supposed to do anymore.

Until step-by-step you noticed people dropping things, getting excited about growing a certain channel, or hyper-focusing on refund rates.

A nice example is when Thibaut started calculating KPIs to reach the 50k objective by the end of the year.

Or hosting our reflection workshop on the past 3-6 months where we received post-its like ‘loved how well Betterpic integrated within the team’.

Part of the Saasmic & Betterpic Team in Barcelona

Now suddenly all those nights dreaming up OKRs, killing projects, and making sure we get drunk together somehow made sense.

Suddenly these targets were becoming a lot more tangible.

Challenge 2: Low Website to Purchase Rate

One of the challenges faced was the feedback we received on our website:

“it looks like you are going to scam me” or “It looks like a school project” weren’t uncommon things we heard when showing the website to people.

I came to a point where I noticed myself talking about it but mentioning that we are currently rebranding it before people would even search for the website.

I’ll let you be the judge of it as this is how it looked like:

This is also reflected in our sales figures when we looked at our website visit-to-purchase ratio which we have set up in Mixpanel.

Yup, only 0,63% of our website visitors convert into sales.

These metrics would never be able to get us into profitability, especially given that most of our users are still taking the lowest package of 25$.

We decided to do something about it and spend an entire month rebranding every single screen you could find.

What started with “let’s just change a few colors" ended up taking a full month to rebrand every single screen we had.

As our CTO, Miguel mentioned in a conversation: ‘If we killed the legs, arms, and the body, can we still call it a rebranding or is it just a new company’?

So here is a small walk-through of the video made for our Producthunt Launch.

Do you like the new Betterpic Brand?

Login or Subscribe to participate in polls.

Challenge 3: High refund rates

My personal favorite is refund rates. Not because I like to see them but because I cringe every single time I see them passing in my Slack. On the other hand, I know they are just part of a business that not every single customer will be happy with their orders.

We know this because the NPS on our headshots is at the core of what we do and it’s not perfect yet:

And they also make us laugh from time to time:

I remember saying to our CTO, that we’ll start with implementing Helpscout to automate and centralize our tickets, but I want you to handle them for the first months.

Not because I think that’s his role, but because I wanted to make sure that our development team can see exactly what is going wrong with our headshot orders.

Until one day I noticed we were very quick in replying to the people requesting a refund and noticed this in our Helpscout:

The guy even automated our refund requests with an exact response based on the data we have about our customers through ChatGPT. I guess engineers will remain engineers.

But he took it a step further when it comes to giving people a complimentary order. He even created a custom GPT to give free refund orders (I assume he was tired of us asking him).

Now of course the biggest refund rate drop didn’t come from being able to answer quicker when people had a problem.

The biggest refund drop will come from:

1. Increasing the quality of our in-house AI model to deliver higher-quality pictures. (done but needs always-on improvement)
2. Helping people upload better-quality pictures through our AI assistant (done)

3. Better managing the expectations of what people will receive (finishing in March 2024)

4. Allowing users to adapt their received pictures with Manuel Edits and AI edits (finishing in April 2024).

And well, many more late-night coding, coffees, and drinks to create the best AI headshot software out there.

So here it is, a small update on how it’s going, the challenges we are facing, and an open book on our metrics.

Hopefully, if you read up to this point, you are either crazy or you liked the content. I’ll let you be the judge of it.

See you in our next update.


Join the conversation

or to participate.