A group was opening a community space — a coffee shop, a bar, a makerspace, and a small gallery, all under one roof. They needed to prove to the city it would be worth a $2M grant, and then actually run the place once it opened. I built two analyses: one that projected what the space would be worth to the neighborhood, and one that tracked individual customers the way a website tracks users. The grant cleared. The tools still run the business today.
The thesis, in one line: treat a building like a product. Count every visit, follow every regular, and the math of why someone walks in — and whether they come back — stops being a guess.— Starting point
The team was turning two vacant buildings into a community space with four businesses inside. To unlock a $2M city grant, they had to show — on paper, before opening day — that the space would bring real jobs, visitors, and tax revenue to the neighborhood.
Visitors per year. Jobs created. Sales tax generated. A five-year revenue projection. The kind of numbers a grant committee can defend in a public meeting.
A good bakery knows its regulars. A good website knows its users. I gave the operators the same thing: who walks in, how often they return, what they spend, and which ones are worth chasing for a membership.
Both analyses live in Jupyter notebooks — working documents that mix written explanation with live calculations, so every number on the page can be traced back to the assumption that produced it. Nothing is hidden in a spreadsheet.
No prior data experience required. Technical bits stay in boxes; the story is in the paragraphs around them.
The grant committee doesn't care about vibes. They want to know, in dollars and jobs, what this place will do for the neighborhood. To answer honestly, we build the number up from its smallest pieces — the size of each room, the price of a latte — instead of picking a hopeful total and working backwards.
Before any projection, we list what we actually know. How many square feet is the coffee shop? How many seats does the bar have? What does the average customer spend? These aren't guesses — they come from the lease, the floor plan, and comparable venues nearby.
Every chart later in this analysis is built from the four rows shown below. If the coffee shop ends up smaller than planned, we change one number here and every downstream result updates automatically. That's the whole point: no hidden math.
Operating assumptions per venue. Edit here; all sections downstream re-derive.
Next, we simulate an entire year of daily visits. We build in the patterns every real venue has: coffee peaks on weekend mornings, the bar fills up Thursday through Saturday night, the makerspace stays small and steady. We also factor in a slow opening ramp — no one is at capacity on day one.
The chart below is the result. Each line is one venue over 12 months, smoothed so the shape is easy to read. It's not a prediction of one specific Tuesday — it's a reasonable picture of a normal year.
Not everyone who sees an ad walks in. Not everyone who walks in buys something. Not everyone who buys something comes back. The chart below tracks that dropout, step by step — what marketers call a funnel.
Read it top to bottom: 500,000 people see an ad on their phone; 75,000 click through or visit the website; 19,600 actually walk in the door; and by the end of the year, about 4,939 have become repeat visitors. Each drop-off is a lever the operator can pull — a better sign out front, a free first drink, a reason to return next week.
Multiply visits by average ticket and you get direct revenue: about $1,035,920 in Year 1, split across the four venues. That's the money changing hands at the register.
But the full economic picture is bigger. When the coffee shop buys beans from a local roaster, and the roaster pays a driver, and the driver buys lunch — that chain of spending is called the multiplier effect. The U.S. government publishes official multipliers for exactly this purpose (they call it RIMS-II). Applied to our direct revenue, the total economic output comes out to $1.86M, supporting roughly 27 jobs and returning $52,719 in sales tax to the city in the first year alone.
| Revenue Stream | Annual | Share |
|---|---|---|
| Coffee Shop | $262,080 | 25% |
| Bar & Beverage | $337,040 | 33% |
| Events & Private Rentals | $336,000 | 32% |
| Makerspace (lease) | $90,000 | 9% |
| Art Gallery (commission) | $10,800 | 1% |
| Direct Revenue | $1,035,920 | 100% |
| × 1.8 output multiplier (RIMS-II) | $1,864,656 | |
| Direct FTE · supported (×1.4) | 19 · 26.6 | |
| Sales tax to City of Denver | $52,719 |
Most business plans show a single line going up and to the right. Grant committees have seen a thousand of those, and they don't trust any of them. So instead we show three futures side by side: a cautious one (slow growth), a realistic one (the base case), and an optimistic one.
Even in the cautious version, the space clears the grant's requirements. That's the point of showing the range — the pitch doesn't depend on everything going right.
Why the grant cleared — committees don't reward the biggest number. They reward the number they can defend. Every assumption was written down, every multiplier was borrowed from a federal source, and the downside case was presented alongside the upside. The one-page summary fit on a single sheet; the supporting math ran fifty more.
Part 1 gave the grant committee the big numbers. Part 2 is for the people running the place once it opens. Instead of totals — 75,000 visitors, a million in revenue — we zoom in on individual customers. The goal: know your regulars the way a neighborhood bartender does, but at scale and with data.
The trick to all of this is a moment most people don't notice: the free WiFi sign-in. When a customer joins the WiFi, they give an email address. That email is the thread that ties everything together — the same email might show up on a bar tab two weeks later, or on a membership signup two months later.
To show how the analysis would work once the space was open, I simulated 5,000 fictional customers, each with a realistic mix of where they heard about the place, which venue they prefer, and how engaged they are. The table below is a peek at that roster.
| user_id | source | zip | engagement | affinity |
|---|---|---|---|---|
| 1 | social_ad | 80202 | 0.24 | coffee |
| 2 | organic | 80205 | 0.61 | both |
| 3 | referral | 80206 | 0.08 | bar |
| 4 | search_ad | 80204 | 0.33 | coffee |
| 5 | event | 80211 | 0.77 | both |
| … | … | … | … | … |
| 5000 | organic | 80203 | 0.19 | coffee |
Same idea as the funnel in Part 1 — but this time we're watching individual customers instead of crowds. The good news: 88% of first-time visitors come back at least once. That's a healthy sign that the space itself works.
The harder news: only about 1% actually become paying members. That sounds small, and it is — but it's also the most valuable step on the whole chart, because (as we'll see next) members are worth many times more than casual visitors. This funnel is the operator's to-do list.
“Customer lifetime value” is a fancy term for a simple idea: how much money does one customer spend with us over the course of a year? Add up their coffees, their drinks, their event tickets, and their membership fees. That's their value.
The chart below is the most important picture in the entire project. A person who just drops in a few times is worth about $60 a year. A person who signs up for a membership is worth $450 to $550+ — roughly seven and a half times more. The entire strategy for running the space boils down to one sentence: move people from the short bar on the left to the taller bars on the right.
Once the customer view was running, three patterns jumped out immediately. Each one was specific enough to change how the team spent time and money in week one.
Multi-venue users (coffee and bar) have 5.7× the LTV of single-venue users. The membership ladder is designed to pull single-venue visitors across.
Event attendees and organic walk-ins convert at roughly the same rate as paid — but their LTV is $3–5 higher. Community events aren't marketing spend; they're the best marketing spend.
Users who make it to their third visit become 6× more likely to convert. The first-90-days playbook is designed around pushing people past that threshold.
All the charts up to this point are averages. To make it real, here's what one specific customer's year actually looked like, pulled from their record in the system. She saw an Instagram ad in January, came in for a latte, came back a week later with a laptop, wandered over to an art opening at the bar, and eventually bought a coffee-shop membership on Valentine's Day.
By the end of the year she'd spent $477. One person. One story. Multiply by a few thousand customers and you have a business.
None of this analysis is possible without one small trick: recognizing that the person on the WiFi, the person at the register, and the person on the membership list are often the same person. The diagram below shows how those three systems get stitched together with a single shared email address — and how the stitched data then flows into the two analyses you just read.
The one-sentence takeaway — we stopped counting foot traffic and started tracking people. Who came in, how often they returned, what they spent, whether they stuck around. Once you can see the difference between a $60 customer and a $450 customer, the whole job becomes obvious: help more of the first group turn into the second.
What I learned — the instinct in grant work is to make the numbers look as big as possible. The discipline is the opposite: show the downside, cite your sources, and let the real number speak. The same discipline that convinces a committee also keeps the business honest with itself six months later.