The morning we opened site #2, I stood in the new café at 6:45am, watching the first customers queue outside, and thought we've done it. The pastries looked right. The coffee smelled right. My new manager was smiling. I'd parked my car around the back and walked in like I owned the place, because for the first time in years I felt like I genuinely did.

At 9:12am my phone rang. The espresso machine at site #1 - 14 miles away in the original café - had thrown an error code none of the staff had seen, the queue was out the door, and the closest engineer was in Reading. I was standing in a brand-new café with no jobs to do and a fully functional team, while my original site, the one I'd run for over a decade, quietly fell to bits without me.

That was the moment I understood what I'd actually signed up for.

The myth of doubling

Everyone tells you about the turnover. Open another site, double your turnover. And in a literal sense, yes - if site #2 trades at the same rate as site #1, the till sums look very nice on paper.

What they don't tell you is the workload. The workload doesn't double. It triples. You now have two of everything that used to be one - two leases, two landlords, two suppliers (often), two teams, two compliance regimes, two deep-cleans, two boiler services, two sets of small daily fires. And you've added an entirely new third job that didn't exist before: the bit in the middle, where you have to coordinate the two and notice when they drift.

Margin doesn't follow turnover either. The bigger numbers attract bigger costs - more admin, more software seats, more accountancy, more management time, more breakages because more people are using more equipment more hours a week. Profit doesn't double. In year one it usually goes down, often quite a lot. If you haven't budgeted for that, you're in trouble before you've sold your first flat white.

What I got away with on site one

Site #1 ran on me. I didn't see it that way at the time - I genuinely thought we had systems. We had a recipe folder. We had a rota. We had a cleaning rota stuck on the back of the cellar door.

But the systems were a thin layer over a much bigger thing, which was me. I was the costing system. I was the supplier escalation route. I was the trainer, the troubleshooter, the recipe arbiter, the person who knew which kettle had a dodgy switch. The recipes in the folder were aspirational; the real recipe was whatever I'd shown the last person who joined.

The first café runs on the founder's head. The second one can't. The day you open site #2, every shortcut you got away with on site #1 becomes visible - because suddenly there's a place you physically cannot be, and the systems either hold or they don't.

This, by the way, is the central argument of building a café that runs without you, and I'd read that one before signing a second lease, not after.

The five things to write down before you sign

If I were doing it again, I'd refuse to sign the second lease until these five things existed on paper. Not "in my head, ready to type up". Actually written down, printed, used by someone who isn't me, on site #1, for at least three months.

1. Recipes and costings

Every recipe, with weights, methods, and a current cost per unit. The reason this matters more than people think is that suppliers vary by town. My Witney butcher and my Burford butcher are different people, charging different prices, in different pack sizes. Same recipe, two different food costs. If you don't have the recipes formalised, you can't compare site margins, and if you can't compare site margins, you have no idea which site is actually working.

This is the failure that became CostingBrik. The recipe spreadsheet broke in week three. I was maintaining two columns in Excel for two suppliers, and they kept getting out of sync, and one Sunday I priced a sandwich at the new site using the old site's flour cost. That sandwich made nothing for six weeks before I noticed.

2. Cleaning rotas and closing checklists

The cleaning at site #1 happened because I noticed when it didn't. That's not a system. A system is a printed checklist, signed off nightly, that someone else can audit on Tuesday morning. Closing routines especially. The drift on a closing checklist - the small bits that get skipped because the closer is tired - is the exact thing that gets you a four-star hygiene rating instead of a five.

I wrote about this in closing checklists and staff accountability after the first time I opened up at site #2 and found the milk fridge hadn't been wiped in days. The fix was unglamorous and worked.

3. Cash handling and till cashing-up

Two tills. Two cash floats. Two end-of-day procedures. Two opportunities for the count to be £14 short and nobody to know why. Document the process exactly: who counts, who checks, where the bag goes, who banks. If you've never been robbed by a member of staff you trusted, brilliant - keep it that way by removing the temptation that comes from a sloppy process.

4. Stocktake process

A stocktake is only useful if it's done the same way both times you compare it. Two sites doing two different stocktakes give you two unreliable numbers and no usable variance. Pick a method, write it down, train both teams the same way, do it on the same day of the week. Boring, essential.

5. The "what to do if X" decision tree

This is the one I had nothing for, and the one that hurt most. Machine breaks. Staff member doesn't show. Fire alarm goes off mid-service. Customer claims food poisoning. Card terminal won't connect. Suspect cash discrepancy. Power cut.

For each of those, who decides what, and who calls whom? At site #1 the answer was always "ring Ed". At site #2, when I was 14 miles away in a different café with a queue, "ring Ed" was a terrible answer and everyone knew it.

The manager hire

The most important hire of my life was my first second-site manager. I got it wrong.

I hired someone who would have been a brilliant me. Strong on craft, charismatic, knew the food, customers loved her. What she wasn't was a manager - someone who builds rotas without prompting, runs a stocktake on a Sunday because it's stocktake Sunday, has uncomfortable conversations with underperforming staff without being asked, and reads a P&L without flinching.

The owner-spec and the hireable-manager-spec are completely different jobs. I'd been doing the owner job and I wrote the JD as if I was hiring an owner. What I needed was a manager. The gap between the two cost me about nine months and a lot of money before I admitted it.

The second time around I hired for management instinct first and trained the craft on. That worked. It also meant accepting that the new site wouldn't feel exactly like the old site. Which leads us to the next problem.

Cash flow gets weirder, not smoother

I had assumed that two sites would smooth things out. More turnover, broader base, less exposure to a quiet week.

The opposite is true at first. You now have two rents, two business rates bills, two utility deposits, two POS subscriptions, two card terminal rentals, two insurance policies, two waste contracts, two pest contracts, two deep-clean cycles, two refit reserves. Most of those are fixed monthly outflows that don't care how busy you were last Tuesday.

Worse: the seasonal dips compound, they don't cancel. A wet July hits both sites. A quiet half-term hits both sites. The "diversification" you thought you were buying turns out to be much smaller than you assumed, because both cafés sit in the same regional weather, the same regional school calendar, the same regional consumer mood.

The week we genuinely nearly went under, in year two, was the third week of a wet August during a rail strike. Both sites were quiet. Both sets of fixed costs went out of the bank on the 1st of September anyway. I spent a Sunday morning at the kitchen table moving money between accounts and felt physically sick.

After that I built a 13-week rolling cash flow forecast and it stopped being optional. Not "I'll do one when I have time". A live document, updated weekly, that I look at every Monday morning, with both sites and head-office costs in one view. The first time you spot a problem six weeks before it arrives, you understand why people bang on about cash flow forecasts.

Stock and ingredient pricing across two sites

The spreadsheet does not survive contact with two sites. We tried. Ours had four tabs - one per site for ingredients, one per site for recipes - and within six weeks they were lying to us. Different versions. Different supplier prices. Different VAT treatments. A supplier change at one site that nobody back-filled to the other.

The cost-per-unit on a sausage roll at site #2 was, at one point, forty pence different from site #1 because of an out-of-date butter price. The site that looked more profitable wasn't, and the site that looked less profitable was being unfairly judged. Decisions made on bad data are worse than decisions made on no data.

This is genuinely how CostingBrik happened. Not as a clever startup idea - as a thing I built because I couldn't get a straight answer to what does our hummus actually cost us this month, at both sites.

Same or different? Just decide.

Here's a question I didn't think hard enough about before opening: do I want both sites to feel the same, or to feel local?

There's no right answer to that. Costa wants every Costa to feel like every other Costa. Plenty of brilliant independents take the opposite view - each site has its own personality, its own micro-menu, its own playlist. Both work.

What doesn't work is not deciding. Because then site #2 drifts. Maybe the new manager prefers a different milk. Maybe the local customers ask for a sausage bap and it creeps onto the board. Maybe the music is louder. Six months later you have a half-cousin of site #1, and you can't tell whether the differences are deliberate brand decisions or accidental drift, and your suppliers and your costings and your training all have to handle both versions of every product.

Pick a position. Write it down. Tell your managers. Revisit annually.

The site you love most

A rueful one, this. Six months in, it will be obvious - to you, to your staff, and to your regulars - which of the two sites you love most. The other one will feel it, and it will hate you a little.

Usually you love site #1 more. It's the one you built. You know its quirks, the customers, the staff who've been with you for years. Site #2 is the harder, less rewarding one - it's the one with the unfamiliar problems and the manager you're still figuring out.

So you spend more time at site #1. The team at site #2 notices. Standards drift. You then visit site #2 to "fix it", which feels to that team like you only show up to tell them off. They get defensive. Performance dips. You visit even more, in even worse spirits. And on it goes.

The fix is mechanical, not emotional. Block your week. Same days at each site, every week. Run a P&L per site every month and read both at the same desk in the same hour. Make sure both managers see you give the same time and attention to their numbers. The team will trust the rota of your attention more than they'll trust your sentiment.

What I'd tell my younger self in three sentences

One: Sign nothing on site #2 until site #1 has run for three months without you in the building, on documented systems, and the numbers held. If site #1 can't survive a quarter without you, site #2 will eat both.

Two: Hire the manager twelve weeks before you need them, on a salary you can't quite afford, and use those twelve weeks to have them rebuild your site #1 systems with you. Cheap manager, hired in a panic the week before opening, is the single most expensive mistake I've made.

Three: Build the 13-week cash flow forecast before you sign the lease, populate it with realistic openings (not your best-case dreams), and if it goes red in any week of year one, pause - don't push. The ego of opening on time has cost more independents their business than anything else I've watched.

Where Brikly fits

I won't be precious about it: every Brik I've built started with a thing that broke when we went from one site to two.

CostingBrik exists because the recipe spreadsheet broke. Two suppliers, two pack sizes, two VAT treatments, and a hummus that I couldn't price honestly. StaffBrik exists because the rota WhatsApp groups broke - one group per site, then a third for "managers", then a fourth for "kitchen", and within a fortnight nobody knew which group a shift swap was meant to land in. PulseBrik exists because I couldn't see both sites' P&L on the same page in the same week - by the time my accountant sent the monthlies, the moment to act on them was gone.

I'm not trying to sell you anything in this post. I'm trying to be honest that the product is downstream of these specific failures, not upstream. The reason I think modular software, paid for as you go, is the right model for independents is precisely this: the things that break when you scale break in a specific order, and you should be able to add the tool that fixes the next break, when you hit it, not pre-pay for a suite you don't yet need.

Site #2 doesn't double the work. It exposes every shortcut you got away with on site #1. That's not a reason not to do it. It's a reason to spend the year before opening as if site #2 already exists - building the systems, hiring the manager, formalising the recipes, running the cash flow - and then signing the lease.

I wish I had.

Ed O'Brien has run Hunters Cake Company for 17 years across cafés in Witney, Burford, and a bakery in Carterton, Oxfordshire. He's building Brikly - modular tools that give independent café owners the same data the big chains have, without the big chain price tag.