project_log: 028

Yo.

Last week, I realised I'd spent 5 months waking up at 6am to build the wrong solution for my side project, Ballpark.

Every morning from 6am to 8am, I've been building an estimator that plugs directly into your existing website. I got the idea from a friends site I'd built a few months prior. We created a simple form with a calculation in the backend and a nice little automation flow:

1. Client fills in form.

2. We run a calculation based on their answers.

3. Send them an email with the estimate.

4. Store their details in notion.

5. Ping slack for hasty follow up.

Simple stuff that worked perfectly.

But I convinced myself it wasn't good enough. "This is too simple for a product. Any technical person could set this up in a few hours. How do I make this better and more innovative?" AI was my answer.

I told myself that estimates needed to be more complex. That projects had more nuance. Estimates needed to account for things like business type, current challenges, and employee count. Even though these things rarely affected how I actually priced my web development services.

My brilliant solution was to use AI to find similarities between past projects and client inputs. Based on this, AI could "guess" what an estimate should be. I told my friends and family about my idea to make myself feel smart and technical. People praised me because they didn't understand it and thought it was cool.

Problem is, this estimate is your first touchpoint with new clients.

How detrimental would it be to to let AI "GUESS" what that touch point included? In reality, your first touchpoint needs to be concrete and predictable.

Let me show you my stupidity with an example:

Let's say you're a web designer. You're using Ballpark to send automated estimates because you understand that speed is the game. You know that reducing the time between client enquiry and delivery is what wins work and allows you to charge premium rates.

Client fills in their details and your estimator sends out $5K as the estimate. Client is happy with the estimate and want's to proceed. You check your Ballpark dashboard and are shocked. You'd usually charge $10K for this project.

You reach out shivering in your boots telling them the price is actually going to be double. You lose the relationship and the client because they're anchored to the $5K price.

You lost a client because of something I built.

Last week, after multiple users shared their concerns, I realised the mistake I'd made. I'd chased the shiny object of AI. I'd forced it into a process that didn't make sense. I let my ego of sounding technical cloud my ability to solve a problem well.

The simple solution that works is almost always better than the complex one that sounds impressive. Especially when real money, clients, and relationships are on the line.

Don't overcomplicate shit.

I bring this up because I'm currently rebuilding the estimation engine of Ballpark. No more AI guessing games, just predictable estimates that you control completely. If you want, you can join the waitlist here and I'll be in touch when your invite is ready.

That's all for this week.

Keep it simple. Talk soon.

Will