I Built a Shop on a Street Where Nobody Walked

I uploaded my 2023 business data into Claude and asked it to show me what actually happened.

I already had theories. I wasn’t good at marketing. I needed better sales skills. The timing was off. The pricing wasn’t right.

Claude looked at the data and said: “You built a mini-department store before you had a single proven bestseller.”

Then it showed me numbers I’d been telling myself didn’t matter.

The Data I Didn’t Want To See

15 sales. Total.

After 10 months of work, designing and illustrating my collection, setting up an online store, creating pricing ladders, building spreadsheets, planning quarterly forecasts…I made 15 sales.

Let me break that down:

• 1 calendar (out of 25 produced—and that one sold at £10, half price)

• 15 tea towels (out of 25+ produced)

• 0 prints (despite planning 48 different SKUs)

Revenue: roughly £250

Production costs for those items: around £160

Total investment: £460

Net position: - £210

Claude calculated my hourly rate based on roughly 500 hours of work: £0.50 per hour.

Less than the cost of a coffee per hour.

What I Thought Was Wrong

For two years I’ve been telling myself that launch didn’t do as well as I hoped because I wasn’t good at marketing. I needed better sales skills. I should have promoted it more. Different timing. Better pricing.

Standard stuff. Fixable problems.

Keep in mind I didn’t actually look at the data until now, I just thought I knew what happened and moved on. I was ready to take what I’d learned and do better next time.

What Was Actually Wrong

The AI didn’t care about my theories.

It looked at the data and said: You had 18 email subscribers at launch. You made 6 Instagram posts in the 5 months leading up to opening your shop. 14 out of your 15 sales were to people you already knew personally.

The math was brutal:

• 10 months designing and building

• 6 social media posts

• Ratio: 200 hours of building for every 1 hour of marketing

I didn’t have a product problem. I had an audience problem.

I built a shop on a street where nobody walked.

The Pattern I Couldn’t See

Here’s what I actually did in 2023:

I spent months perfecting my illustrations. I created detailed spreadsheets tracking SKUs, costs, margins, lead times. Sourced amazing suppliers. I built a professional e-commerce operation with packaging analysis and shipping rates calculators. I planned a content calendar with newsletter schedules, Pinterest strategies, and blog posts.

Then I made 6 Instagram posts and launched.

The tea towels actually sold well—60% sell-through is strong for a first launch. But instead of seeing that signal and doubling down, I was already planning the next collection. More designs. More products. More SKUs.

I never asked: Who am I selling to?

Because the answer was: basically nobody. Just 18 people on an email list, most of them friends and family.

The Twist I Didn’t See Coming

Here’s what actually happened:

I didn’t sit down and analyse 2023’s data at the end of that year. By then I was pregnant with my second child. I thought to myself, “You launched a product line, that’s a win. The sales will improve with the next collection”.

So I spent 2024 designing collection number two. New illustrations. Better products. I worked hard to get it ready before my son was born.

I was convinced I knew what went wrong in the 2023 launch. I’d do the marketing better this time.

Collection two flopped worse than collection one.

I’d spent an entire year “learning lessons” from 2023 without ever actually looking at what went wrong. I never looked at the data.

What The AI Actually Showed Me

The uncomfortable truth wasn’t about my products or pricing.

It was about the story I had been telling myself.

I convinced myself the problem was execution: I needed to get better at marketing, improve my sales skills, and learn to promote effectively.

But the data showed a different problem: I had nobody to market to.

I had spent 10 months building a shop and 6 hours building an audience.

When you have 18 email subscribers and 14 of your 15 customers are people you know personally, one of them your mum, the problem isn’t that you’re bad at marketing.

The problem is you’re trying to market to a void.

The Question That Changed Everything

Looking at the numbers now, I realize: I never actually expected that launch to succeed.

Not really.

I was so focused on proving I could do it - design a collection, set up a shop, get products into the world, that I never stopped to define what success would look like. At the time, for me, success was having my own shop and not how many sales I had made.

15 sales felt disappointing but not catastrophic. £210 loss felt like “part of the learning process.”

I told myself it was a first launch that underperformed. Not great, but a foundation to build on.

But £0.50 per hour? That’s not an underperforming first launch…that’s a hobby I paid to have.

And I was about to do it again. In fact, I did do it again in 2024 and spent the first six months of 2025 planning a third and fourth collection. I never asked myself: What would a successful launch actually require?

What This Actually Means

If I launched the same collection today, here’s what would have to be different:

Not the products. Not the pricing. Not the “strategy.”

The audience.

I would have to spend 6 months posting consistently, building an email list, sharing the design process, and creating genuine interest before I produced anything. Pre-selling. Testing demand. Making people care.

The tea towels working (60% sell-through) proved the designs weren’t the problem. People liked the work.

They just didn’t know I existed.

I built a shop on a street where nobody walked.

The AI Framework Itself

Here’s what I actually did:

I uploaded everything into Claude:

• All my spreadsheets (costs, sales, production, planning)

• My marketing calendar (such as it was)

• Timeline of activities

• Rough estimate of time spent

Then I prompted with the following questions:

Here is all the data from my 2023 product launch. Analyse it and tell me:

• What patterns do you see when I made money versus when I didn’t?

• Which activities took most time but generated least revenue?

• What worked that I should do more of?

• What didn’t work that I need to stop doing?

It took maybe an hour. The analysis was instant.

What took longer was sitting with the answers.

Because they weren’t kind.

What I’m Testing Now

Part of rebuilding my creative practice is facing what didn’t work before.

This analysis showed me I was about to repeat the same mistake: build first, find audience later.

The discomfort of seeing that pattern? That’s the value.

This isn’t a one-time exercise. Running this kind of analysis at year-end should be standard practice—not to beat yourself up, but to actually see what happened versus what you told yourself happened.

Because the gap between those two things? That’s where your next year’s strategy lives.

I’m five weeks into a 12-week Substack experiment, documenting how I’m rebuilding a creative business that works within the constraints of parenting two small children. This week I tested my first workflow: using AI to analyze what actually happened in my business versus what I told myself happened. If you want to try this yourself, the process is simple: upload your data, ask hard questions, sit with uncomfortable answers.

This post originally appeared in my Substack newsletter. Subscribe for weekly updates.