Getting serious about marketing after hitting $500k ARR

David Bressler built an AI-powered data analytics platform called Formula Bot before AI was cool. It went viral and then he grew it organically to $500k ARR.

And now, he's getting serious about marketing.

Here's David on how he's doing it. 👇

Building on paternity leave

I spent 15 years working in the data analytics space before launching Formula Bot while on paternity leave with my second child. Formula Bot is an AI-powered data analytics platform built to make working with data easier and faster for everyone, regardless of technical or analytical background.

I grew the platform as a side project while working full-time as a Senior Director of Analytics at a media analytics agency. Once Formula Bot's income surpassed my full-time salary, I made the leap to focus on it full-time.

Formula Bot enables users to upload files like Excel or PDFs, or connect third-party data sources such as Google Sheets and Google Analytics. Users can then chat with their data to generate charts, tables, insights, advanced models, and more.

In addition to quantitative analysis, the platform also supports enrichment tasks like generating sentiment from customer reviews or classifying and cleaning messy data.

MVP

Formula Bot today is a completely different product from what it was at launch. It started as a simple Excel formula generator — users would input text instructions, and it would return the corresponding formula.

I thought of the idea because I had been working in Excel throughout my career and became very good at writing Excel formulas. I became the ‘go-to-guy’ for Excel help at work so the first idea I had when testing out AI in its early stages was to see if it was good at Excel. It was okay, but I figured with my background, I could make it good enough to be a product.

All the stars aligned while I was on paternity leave. Around that time, OpenAI’s Playground and the no-code platform Bubble.io were gaining traction - especially on X. Within a week, I built the MVP — just an input box, a button, and an output box powered by a single API call. That was it.

At the time, my competitive advantage was formula accuracy. LLM’s quickly evolved, so my focus then went elsewhere.

A low-code stack

Starting off, the MVP tech stack was simple - Bubble.io and OpenAI.

Now, it’s a lot more robust - Bubble.io powers the majority of the application in the backend, but also integrates with the following:

• OpenAI (still)

• Brevo (email management)

• AWS S3 (file storage)

• Render/Github (executing robust Python scripts)

• N8N (retrieves data from external data source APIs like Google Analytics)

• Gemini (performs certain AI tasks)

Currently, I’m focused on rearchitecting the entire application to be more scalable. As a result, there’s been a shift from no-code to code. It has gone from 100% no-code to 70% (current) and will soon be 30% in the coming months.

Try hosting "office hours"

For over a year, I hosted an “Office hours” where any user (free or paid) could book time to chat with me. I’d typically take 3-5 calls a week, where users would provide me with feedback and show me their workflow and use cases. Those calls helped develop the product into what it is today.

Moving to a subscription model

Despite being so minimal, the app took off. Influencers shared it across social media, generating millions of impressions and bringing tens of thousands of users to the site. I hadn’t expected that kind of response.

There was no paywall, no login system, no API limits - just a free tool making one API call. When it went viral, I was excited… until I saw my OpenAI bill. I was spending thousands of dollars a week. I had a choice: shut it down or monetize it.

Before introducing a subscription model, I was scrappy, making money through banner ad placement and affiliate marketing, directing users to Excel tutorial affiliate sites.

I did that for a few months, until I was ready to relaunch. It was no longer a ‘free for all’ MVP. This time, it was a full-fledged product with subscription tiers, new features, API limits, login, etc.

Today, Formula Bot generates 99.9% of its revenue through its subscription plans that range from $15 - $35 per month. I still get 0.01% of revenue from affiliate partners that I no longer market, but I’m still receiving commissions from sales I drove over 2 years ago!

Formula Bot has over 1M total users (email addresses), and last year, we had $500k ARR.

Going viral on reddit

The first place I went to share Formula Bot was the Excel subreddit, naturally. Again, this was before ChatGPT launched, so when I shared a free tool to generate Excel formulas to thousands of Excel nerds like myself, it went instantly viral. Then, someone recommended I share it on the /InternetIsBeauitful subreddit, where it generated 1M+ views.

I’ve found that this subreddit, specifically, is a breeding ground for social media influencers. If a post goes viral here, it will be picked up the next day by influencers on TikTok. If you can get a lot of upvotes here, it’s bound to go viral on multiple other channels.

Of course, if I were to do the same thing today, it would never go viral. V1's functionality was a novelty then; now it's a commodity. Plus, the post would get banned in a second for being promotional.

There was actually a YC backed company called Tersho that launched their own Excel AI tool on Reddit the exact same week as me. It did the same thing as Formula Bot, but it was better. Cleaner UI, more accurate, more features. One went viral (mine) and the other one did not. The company has since pivoted into gen AI powerpoint slides.

This really was such a pivotal moment for my startup. Had I launched any later, my post wouldn’t have gained so much traction and I wouldn’t have experienced the viral snowball effect.

Eventually, when ChatGPT launched, AI was everywhere and Formula Bot grew with it. I was getting tens of thousands of signups each month, riding the wave of early AI hype. For a while, Formula Bot was one of those “AI tools that feel illegal to know about."

But then the buzz faded and so did my growth. I had gotten used to a steady stream of organic traffic driven by word of mouth and influencers sharing it for free. Signups and new customers flatlined for months. That’s when I hit a turning point: If I wanted this to grow, I had to take marketing seriously because no one else was going to do it for me.

The mechanics of going viral

I sometimes wonder why my app went viral and the other one didn’t, despite them even having a better product at the time.

I chalk it up to a couple of reasons:

1. Generic name that literally explains what the app does without visiting the website. Tersho versus Excel Formula Bot.

2. Free with no sign up required. When I launched, I didn’t require a login (because I didn’t know how to set it up). I also didn’t have an API limit, so users were able to generate thousands of Excel formulas at no cost.

In the short term, this sucked because I quickly racked up thousands of dollars in OpenAI API costs, but in the long term, it was the best case scenario, given that my site became synonymous with Excel AI.

Taking marketing seriously

Here, you can see very stagnant traffic from mid 2023 to the end of 2024. But to start the new year, I made more of an effort to focus on marketing and the trendline speaks for itself.

Year to date, MRR has grown 20% driven by increased marketing. That includes SEO, PPC and influencer marketing — the channel that got me started, though this time it’s paid.

Many VC-backed companies are OK not being profitable via paid ads. That’s not an option for me. I run Google Ads, but it’s very minimal and to the point where I break-even in 7 months. 

I’ve had a lot of success through SEO, specifically for free tools. Build tools (AI or not) that compliment your core product. For example, some of my top SEO landing pages are: PDF to Excel converter, SQL query generator, Excel formula generator, and Excel spreadsheet generator.

Additionally, over time, I’ve made the application less fragmented. I removed 30% of the buttons/sections on the website and consolidated nearly all of the features within a single chat interface. The chat interface makes use of function calling to execute different tools and functions that were previously in different locations. This consolidation and simplicity helped improve the conversion rate by 45%. Less is more!

And displaying educational YouTube videos all throughout the site helps educate people on how to perform certain actions from within the chat via a single prompt, instead of having to click 10 buttons.

"Don't do what I did"

It’s okay to look over your shoulder and see what your competitors are doing, but don’t drive yourself crazy over it. Don’t do what I did.

Most of your users don’t know your competitors. If they offer something that you don’t or if they have more competitive pricing, don’t fixate on it. Just focus on what you can control. There will always be something to make your product better, just market what you have and price accordingly.

If your product is a 1 trick pony or MVP, price it accordingly. When I launched, a monthly subscription was $2.99. It was literally just a single API call with a fancy prompt. Over time, I added more features and I increased the price accordingly.

Run paid ads earlier to test the waters and validate your product, especially if you don’t have an audience on social media.

Once you feel good about it, continue to develop your product and then focus on organic marketing — especially in a competitive market, like AI data analytics.

What's next?

I’ve recently shifted my focus back to product to rebuild the backend to allow for more features and to be more scalable. Once this is built out, these features will unlock a new way for me to create a lot of fun and entertaining marketing content.

As I continue to remain bootstrapped, I’ll never be able to keep up with the other companies, but I’m okay with that. I don’t need a big piece of the market, a small piece is more than enough to live the lifestyle I’m living now - it’s a calm lifestyle: quality family time, few vacations a year, ride bikes to drop off and pick up my kids, work out several times a week, low stress.

You can follow along on my X. Or check out Formula Bot, our X, or our YouTube.