How I Bootstrapped My Analytics Business With $0
You can make thousands of dollars with publicly available data and some coding skills.
Everything running on my ThinkPad x230.
Let’s go back in time, it’s 2018, Montreal, Canada.
The Idea
I’m thinking about startup ideas, staring at my 13-inch Thinkpad screen at a Tim Horton’s coffee shop at 10:00 PM.
It’s a cold snowy day, snow flakes are slowly falling to the ground, as I’m watching outside through a window facing the street, a beautiful sight.
Being a night owl, I always loved working at night, there’s something about night time work that helps me focus.
Probably because everyone is sleeping and there’s less distraction.
So here I am thinking about starting a business without spending a dime, I challenged myself.
At that time, I knew enough of Python to scrape websites, from data extraction to cleaning and loading to a database, that was the easy part.
If my code didn’t work, I’d Google it and find the solution in a few minutes.
Fixing issues in the business world is different.
You can’t Google your way out of it and find a straight forward solution.
That’s what business is all about, and that’s why people are will to pay you a lot of money to find a solution to their unique problems.
I knew that presenting the right data to the right person would be profitable, that was my vantage point.
We all consume data, like when we look at weather forecasts before going outside, if it’s going to rain, we bring an umbrella.
That’s actionable data, information that changes our behavior, it helps us act the right way at the right time.
I was looking for a continuous source of data that could be monetized and actionable.
I wasn’t interested in building an AI or machine learning model, just a simple data reporting service.
I believed there were plenty of untapped opportunities in the DaaS(Data as a Service) domain, going straight to automated models is a bit too far fetched for what I was looking for.
Keeping it simple, not everyone needs machine learning models and it doesn’t solve everything.
At that time, I was also looking for new ways to make money, and I found out about phone flipping.
Where you’re buying used phones and reselling them for profit on sites like eBay or Craigslists.
I knew a couple guys in the US that were doing some good money doing that.
So I joined a couple phone reselling groups on Facebook, and I noticed that many were struggling when estimating phone resale prices.
One of the ways some did it was by going on eBay, searching the phone model and estimating the price from the first couple listings.
That solved the problem, but it was still tiresome if you had a lot of phones to sell.
So I thought to myself, why not automate that ?
That’s what I did and starting building bots in Python using Scrapy and Beautiful Soup, two powerful Python libraries for web scraping.
I though about all the websites that had listing data, Craigslist, eBay, Kijiji.
Should I scrape all of them or not?
I decided to go for eBay, it was simpler, their website wasn’t too complicated and they didn’t update their DOM or website layout often, which would break my bots.
Coding it
Everything was running automatically on my laptop with Debian Linux, thanks to crontab which automated everything, from web scraping to data cleaning and sending emails.
Sleepless nights and a visit to the hospital due to a panic attack right in the middle of the night.
It definitely wasn’t a walk in the park.
At one point, I was thinking about giving up:
“I can’t do that, I’ve put in hundreds of hours in that project"
I told myself.
The hardest part by far, was trying to get quality data. Separating each iPhone models (i.e cleaning and separating the iPhone X 64G from iPhone XS 256G), that took me weeks to get it right. I ended up many times with false positives/negatives.
Because if you don’t have clean data, your whole analysis becomes worthless, a few outlier data points could ruin everything.
At some point, I saw people selling 24 Carat gold iPhones for $17,000 on eBay, and that got into my dataset, which I had to update my cleaning algorithm to avoid extremely overpriced iPhones.
I also removed blacklisted, fake and iCloud locked iPhone listings by looking at the title of those listings. Listings with the keyword “fake”, “not working”, “bad esn” or “iCloud” were removed from my dataset.
But the issue with that is not everyone was transparent about the condition of their iPhone and some were not giving all the details.
So what I did was code an Interquartile Range function that removed outlier data points, which solved 90% of my dirty data problems.
The naming of my business was the least of my issues at that point, but I ended going with Phone Flipping.
MVP Release
Finally, after months of hard work, I’ve finally got something out and running without issues.
I had around 70 bots pulling data from eBay listings, then transforming, cleaning and storing it into a database.
A fully automated data pipeline.
My bots were pulling data everyday at 8:00 AM.
I was rotating my proxies and user-agents for each request. I also added random intervals between requests, to avoid getting detected eBay’s anti-bot systems.
I was sending the pricing reports every Friday morning at 8:00 AM by email.
I was using Gmail’s free SMTP server, and some python code to automate the whole thing.
That was my mail server, I really wanted to spend 0$.
On the marketing side, I did everything organically, no paid ads.
All I did is post content in Facebook Groups and have an opt-in form on a website to get emails.
Why emails ?
Because they’re super easy to get, and it’s one of the best marketing channel in ROI.
“You want free eBay iPhone pricing updates ?”
“We need an email.”
As simple as that.
There’s one YouTuber that helped me a lot, Kish Israni. He let me post my charts every week in his Facebook Group.
A lot of my first users came from his following.
So shoutout to him!
In those Facebook groups, I was helping phone resellers with price estimation and I was answering their questions.
Having a close relationship with potential customers is crucial for growth when you’re bootstrapped. Showing authenticity, empathy and understanding is key.
I focused on giving value first, I wasn’t trying to sell anything initially.
2–3 days later, after sharing free pricing sheets on multiple phone reselling Groups on Facebook.
I got dozens of subscribers messaging me.
My inbox was full of interested users wondering how I was getting my data.
In a couple months, it grew to hundreds of users on the free and paid plan.
After 1-2 months of the free plan, I decided to launch the paid tier and got 2 paying customers in the first week.
I started with $30/month, then I lowered it to $15/month.
I still remember the first person who joined the paid plan, a store manager.
He was paying me $30/month.
A surreal moment for me, seeing that $40 in my PayPal account. (Yes, I was using PayPal!)
Here’s the email my free subscribers would receive, in which I would have a call to action to join the paid tier:
I played a lot with the pricing to get it right. Should I go with a trial or not?, should I have a single plan or not? A lot of testing, trials and errors.
Before monetizing, I sent out a Google survey to get an idea of how many were interested in a paid plan.
Around 20% were.
It’s weird, because I felt like an imposter:
“Do they really like my service ?”
I kept thinking.
I couldn’t believe it.
Everything was automated, the only thing that needed some maintenance was the crawling bots that were pulling data from eBay.
As they would sometimes change and update their DOM page structure once in a while.
That was a real pain in the ass, at one point, I was staying up the whole night debugging my bots and sleeping at 8 AM.
Conclusion
So this is how I launched my first online business, my first 1$ online.
In total, I made a few thousands of dollars before shutting in down.
The things I’ve learned in the process of building my business is worth more than the money I’ve made. Especially in marketing and sales.
I ended taking down Phone Flipping, because during the pandemic my paid users stopped paying.
I was also dealing with some serious mental health issues.
The pandemic had a huge impact on me, which led me to stop everything.
It was unsustainable.
Funny enough, my Facebook Page for my business is still up.
What I’ve learned ?
Bringing the right information to the right person is monetizable.
People will pay you for automating small parts of their business.
Keep it simple, find solutions to simple problems.
Getting quality data is the hardest part, quality over quantity.
Network, network, network
Have a close relationship with your clients and have a feedback loop.