The Series B is closed. $8M at a $45M pre-money valuation. Fifteen thousand subscribers across Perth, Melbourne, and Brisbane. Thirty-five people and growing. The mandate from the board: go national within eighteen months. Three cities to eight or more. The money is in the bank. The question is how.
Charlotte and Diane are in the GreenBox meeting room on a Monday morning in late January. It’s the first week of the new year, the office is still half-empty, and the whiteboard from the last planning session hasn’t been erased. Someone – probably Priya – has drawn a small planning onion in the corner with a sticky note that says “2027.”
Charlotte has a spreadsheet open. Diane has her leather satchel, a flat white, and the expression she gets when she’s about to say something Charlotte won’t like.
“Adelaide is the first expansion,” Charlotte says. “We need to run the full discovery cycle. Event Storm the local supply chain, map assumptions about the Adelaide market, interview ten to fifteen potential subscribers. The Brisbane expansion took four months. If we do it properly, Adelaide should be similar.”
Diane takes a sip of her flat white. “Four months.”
“Yes.”
“For one city.”
“If we do it properly.”
“Charlotte, we need to open five cities in eighteen months. If each one takes four months of discovery, we’ll open three. And we’ll run out of money before the fourth.”
Charlotte’s jaw tightens. She’s heard this before – the speed-versus-rigour argument. She heard it at the meal kit company, where they moved fast and the product didn’t survive contact with the second market. She heard it at the SaaS platform, where they ran a twelve-week discovery cycle for each new region and the board accused them of moving too slowly. The tension never resolves. It just finds new shapes.
“The Brisbane expansion worked because we did the work,” Charlotte says. “We Event Stormed the supply chain. We mapped assumptions. We ran JTBD interviews with Brisbane subscribers before we launched. We found six things we would have got wrong if we’d just copied Perth.”
“I know,” Diane says. “I’ve read your notes. You did excellent work. I’m not saying skip it. I’m saying: you discovered Perth, Melbourne, and Brisbane. What did you learn that’s transferable? And what’s actually local?”
Charlotte pauses. The question isn’t wrong. It’s the question she’d ask if she were advising someone else. The difference is that she’s inside GreenBox now, not coaching from the outside, and the instinct to protect the process is stronger than the instinct to adapt it.
“Some of it transfers,” Charlotte says, carefully. “The subscription mechanics. The matching engine. Customer support. Onboarding flows. Those are the same everywhere.”
“Good. What doesn’t?”
“Farm partners. Delivery logistics. Customer demographics – income levels, household size, cooking habits. Seasonal produce availability. Local competitors.”
Diane nods. “So you need discovery for the things that are local. Not for the things you’ve already discovered. How long does that take?”
Charlotte considers it honestly. The full Brisbane discovery cycle was four months. But half of that was re-discovering things they already knew – how subscription businesses acquire customers, what the onboarding experience should feel like, how the matching engine works. The other half was genuinely local: which farms, what produce, what distances, what the customer base looks like.
“If we only discover what’s local,” Charlotte says, “maybe nine weeks. Ten at the outside.”
“Can you do it in nine?”
“If we have the right people on the ground. And if we have a playbook.”
The playbook idea
The word hangs in the air. Playbook. It’s not a word Charlotte would normally use – it sounds corporate, procedural, the kind of thing a franchise operation publishes in a three-ring binder. But Diane isn’t talking about a franchise manual. She’s talking about something more specific.
“When I scaled Sunridge,” Diane says, “we opened in forty markets in three years. The first five were chaos. We reinvented everything every time. The sixth market was where we built the playbook. Not because we were smart. Because we were exhausted.”
She opens her notebook – the illegible one, the one that somehow contains the operating history of a dozen businesses in handwriting only Diane can read.
“A playbook isn’t a script. It’s a separation of concerns.” She catches herself and smiles. “Tom would be proud of me. I’m using his language.”
Charlotte laughs, despite herself. “What do you mean by separation of concerns?”
“I mean: some things are the same in every city. You’ve figured those out. Document them. Lock them down. Don’t re-discover them. Other things are different in every city. Those need discovery – real discovery, the kind you’re good at. The playbook tells you which is which, so you spend your discovery time on the things that actually vary.”
It’s a clean framing. Charlotte recognises the principle – it’s the same logic that made the bounded contexts work in the engineering architecture. Separate the things that change independently. Define the boundaries. Let each part evolve at its own pace.
“The bounded contexts were Tom’s idea,” Charlotte says. “He separated the subscription system from the supply matching from the delivery logistics. Each one could change without breaking the others.”
“Same principle,” Diane says. “The playbook separates the universal from the local. The universal gets documented and replicated. The local gets discovered fresh.”
They spend the rest of the morning building the first draft. Charlotte contributes the structure – clear phases, defined outputs, explicit decision points. Diane contributes the market-entry speed – parallel workstreams, time-boxed activities, a bias toward action when the evidence is sufficient.
Universal vs local
The first task is classification. What has GreenBox learned in three cities that will be the same in every city? And what will be different?
They pull in Tom, Sam, and Jas for the afternoon. Maya is on a call with the board, but she’s delegated this – she told Charlotte and Diane on Friday: “You two design the playbook. Bring me the first draft. I’ll push back on anything that compromises the farm relationships.”
Charlotte writes two columns on the whiteboard.
Universal – the same in every city:
- Subscription system (tiers, pricing structure, sign-up flow)
- Customer onboarding (welcome email sequence, first-box experience)
- Matching engine (the decision table framework – not the specific rules, but the structure)
- Customer support processes (ticket categories, response times, escalation paths)
- Financial reporting (margin tracking, churn analysis, per-box unit economics)
- Brand voice and marketing templates
- Engineering deployment process
- Team structure (squad model, roles, rituals)
Local – different in every city:
- Farm partners (who grows what, within what radius, at what scale)
- Delivery logistics (routes, distances, courier partnerships, cold chain)
- Customer demographics (income distribution, household size, dietary patterns)
- Produce availability (what’s in season, when, in what quantities)
- Decision table rules (substitution logic specific to local produce)
- Local competitors (who else is doing produce boxes, at what price point)
- Regulatory requirements (state-level food safety, labelling, permits)
Sam adds one that nobody else thought of. “Customer acquisition channels. In Perth, we grew through farmers’ markets and word of mouth. In Melbourne, Anika used corporate partnerships. In Brisbane, it was social media. Each city has different channels that work.”
Jas adds another. “Brand localisation. Not the voice – the voice is the same. But the imagery, the farm stories, the social media content. Brisbane subscribers respond to different farm stories than Perth subscribers. The produce is different. The farms are different. The landscape is different.”
Tom, who has been listening quietly, raises the point that Charlotte knows is coming. “The deployment process isn’t as universal as you think. Right now, we deploy to four environments – Perth, Melbourne, Brisbane, and staging. Each deployment takes about forty-five minutes. I have three people who can do it. Adding Adelaide means a fifth environment. Adding four more cities after that means nine environments. We’re deploying to four environments manually and it takes three hours for a full release cycle.”
Charlotte writes it on the whiteboard under a new heading: Constraints – things that will break at scale.
Tom continues. “I’m not raising this to block the expansion. I’m raising it because if we put it in the playbook under ‘universal,’ we’re promising something we can’t currently deliver. The deploy pipeline needs work before it can handle nine cities.”
“When does it break?” Diane asks.
“It’s already groaning. Five cities is uncomfortable. Six is dangerous. Eight is impossible without automation.”
“How long to fix it?”
Tom considers. “Three months to build a proper CI/CD pipeline. Six months to do it well. But I can’t start until I have a platform engineer, and we haven’t hired one yet.”
Charlotte captures this as a dependency. The playbook can describe the ideal deployment process, but the actual capability won’t be there for city five or six. That’s a constraint the board needs to understand.
The three phases
Charlotte proposes a structure. Every city expansion follows three phases.
Phase 1: Market Assessment (1 week). Answer one question: should we launch in this city? Demographic analysis, competitor mapping, ten JTBD interviews with potential subscribers, unit economics estimate. Exit criteria: a one-page market brief that recommends go or no-go.
“One week feels fast for JTBD interviews,” Charlotte says.
“These aren’t the deep discovery interviews from Series 1,” Diane says. “You already know the job. ‘I want dinner sorted.’ You’re confirming it exists in this market, not discovering it from scratch.”
Charlotte nods slowly. The JTBD work from eighteen months ago defined the job. The market assessment interviews aren’t exploratory – they’re confirmatory. Different purpose, different rigour, different time investment.
Phase 2: Partner Onboarding (4 weeks). The local discovery. Farm visits, a supply chain Event Storm, city-specific decision tables, logistics setup, Example Mapping for local business rules. Exit criteria: a functioning supply chain that can serve 500 subscribers.
“Four weeks is tight,” Sam says. “Brisbane took six.”
“Brisbane was the second city,” Diane says. “We didn’t know what we were looking for. Now we do.”
Phase 3: Soft Launch (4 weeks). Two hundred subscribers, capped. Daily monitoring, weekly retros, decision table adjustments from real data. Exit criteria: four consecutive weeks at 95%+ delivery reliability, 90%+ substitution accuracy, unit economics within 10% of projections.
“Nine weeks total,” Diane says. “Plus a week of buffer. Call it ten weeks per city. We can run two in parallel.”
Tom writes something on a sticky note and puts it on the whiteboard: Pipeline capacity: 4 cities max before automation required.
The Adelaide test
Adelaide is first because it’s the safest bet. Smaller city, lower cost, fewer competitors. If the playbook works in Adelaide, they can trust it for Sydney. If it fails in Adelaide, they can fix it before Sydney.
Maya joins the conversation at the end of the afternoon. She’s been on the board call – the new investor (Cerulean) is keen on the national expansion timeline and wants monthly progress reports. Maya looks at the whiteboard, reads the three phases, and asks one question.
“Who does Phase 2 in Adelaide? Dave can’t be in two places at once.”
It’s the question Charlotte and Diane hadn’t addressed. Dave Morrison – the Margaret River farmer who’s been GreenBox’s anchor since the handshake in month one – is the person who evaluates potential farm partners. He visits the farms. He looks at the soil, the infrastructure, the operation. He talks to the farmers in a language they understand because he is one. His son Ben has been helping with the Melbourne and Brisbane farm networks, but Ben doesn’t have thirty years of experience reading a farm in twenty minutes.
“Dave does the first two cities himself,” Diane says. “Adelaide and Sydney. After that, we need someone who can do what Dave does but in a city Dave hasn’t visited.”
“That’s a new role,” Sam says. “Farm Partnership Manager. Or something.”
“That’s the playbook working,” Charlotte says. “We’re discovering what the playbook needs. A role we didn’t know existed until we tried to replicate the process.”
Maya nods. She texts Dave: Can you come to Adelaide in February? We’re doing the first expansion under the new playbook. Need your eyes on the farms.
Dave replies twenty minutes later: What’s a playbook?
Maya: I’ll explain on the drive.
Dave: Is it like that time you made me put sticky notes on a wall?
Maya: A bit.
Dave: Helen says I can go Tuesday week.
The first week in Adelaide
Dave and Maya fly to Adelaide on a Tuesday. It’s February, still hot, and Adelaide has the dry heat that reminds Maya of Margaret River summers. They rent a car – Dave insists on driving because he doesn’t trust Maya’s city driving on country roads – and spend three days visiting farms.
The market assessment is already done. Sam ran it the week before, working from Perth. The demographic data looks good: Adelaide’s median household income supports the $20-25 price point. There are two existing produce box services but both are small and neither sources locally – they buy from the Central Market wholesale. The JTBD interviews confirmed the job exists: ten interviews, eight positive responses, two who said they already get boxes from the Central Market operators.
Now it’s Phase 2. Farm visits.
The first farm is forty minutes south of the city. Mixed vegetables – lettuce, tomatoes, zucchini, Asian greens. The farmer is a woman named Kath, early fifties, who inherited the farm from her parents and has been running it with her daughter. She’s curious about GreenBox but cautious.
Dave does what Dave does. He walks the rows. He looks at the irrigation. He asks about the soil. He crouches down and picks up a handful of dirt and rubs it between his fingers. Kath watches him with the expression of someone who recognises her own kind.
“Your lettuces are good,” Dave says. “How many heads a week in summer?”
“Four hundred. Maybe five hundred if we push it.”
“And winter?”
“Two hundred. We switch to brassicas. Broccoli, cauliflower, cabbage.”
Dave nods. He looks at Maya. “She’s real.”
That’s Dave’s assessment framework. It isn’t written in any document. It isn’t codified in any decision table. It’s thirty years of farming compressed into a gut check: is this person real? Do they know their land? Will they deliver what they promise?
Charlotte, back in Perth, would want to formalise this. Turn Dave’s intuition into criteria. And she’s not wrong – the playbook needs something more portable than “Dave says she’s real.” But for Adelaide, for the first test, Dave’s judgement is the standard.
Over three days, they visit eleven farms. Dave says yes to seven. Four are too small, too far, or – in one case – too enthusiastic. “Bloke’s promising more than his acreage can deliver,” Dave tells Maya on the drive back. “I’ve seen it before. Eager to sign and then short on volume by month three.”
Maya takes notes. This is playbook material – the criteria for evaluating farm partners. She’s capturing Dave’s thinking so that the next expansion, in a city Dave hasn’t visited, can draw on it.
By Friday, Adelaide has seven partner farms. Not fourteen, like Perth. But enough for a soft launch.
The Event Storm that almost wasn’t
Charlotte flies to Adelaide for the supply chain Event Storm. She’s planned a full-day session – the local team (two new hires, fresh out of onboarding), the farm partners, Sam (remote from Perth), and Maya.
The Event Storm starts at nine. By eleven, something is wrong. Not with the technique – Charlotte runs a tight session and the wall fills with orange sticky notes. The problem is that the domain events look almost identical to Perth’s. Farmer harvests produce. GreenBox places order. Farm packs boxes. Courier collects. Customer receives delivery.
Charlotte stops the session. “This is the Perth supply chain on an Adelaide wall. We’re not discovering anything new.”
Diane, who is observing from the back of the room, says: “That’s because the supply chain is universal. The events are the same in every city. The details – which farms, which courier, which routes – are different. You’re Event Storming the wrong thing.”
Charlotte bristles. She takes a breath. “What should we be Event Storming?”
“The exceptions. The things that don’t work the same way here. Where does Adelaide’s supply chain diverge from Perth’s?”
Kath, the farm partner, raises her hand. “The distances are different. My farm is forty minutes south. The next farm you’ve signed is fifty minutes north. In Perth, Dave told me all his farms are within thirty minutes of the city.”
That’s the divergence. Adelaide’s farms are more spread out. The logistics model that works in Perth – a single courier doing a circuit of fourteen farms in a morning – won’t work when the farms are scattered across a hundred-kilometre radius.
The Event Storm pivots. Instead of mapping the whole supply chain, they map the logistics domain – the sequence of events from farm harvest to customer door, with specific attention to the points where Adelaide’s geography changes the story. They discover three things:
First, Adelaide needs a hub. In Perth, the courier collects directly from farms. In Adelaide, the distances make direct collection uneconomic. They need a central collection point – a hub – where farms deliver and couriers consolidate. It adds a step and a cost, but it makes the logistics viable.
Second, the delivery window is different. Perth delivers on Thursday. Adelaide’s courier partner can’t do Thursday – they’re committed to another client. Wednesday or Friday. The team chooses Friday, which means the packing schedule shifts by a day, which means the farm ordering schedule shifts by a day.
Third, the substitution rules need local calibration. Perth’s decision tables assume year-round availability of certain produce – tomatoes, lettuce, zucchini – because Perth’s climate supports it. Adelaide’s climate doesn’t. Three weeks of the year when Perth has tomatoes, Adelaide doesn’t. The substitution engine needs to be parameterised by region.
“This is what the Event Storm was for,” Charlotte says at the end of the day. “Not the happy path. The divergences.”
Diane nods. “The playbook says: Event Storm the exceptions, not the whole domain. That’s what we learned today.”
Charlotte writes it on the whiteboard: Phase 2 Event Storm: local divergences only. Don’t rediscover the happy path.
The pipeline groans
Tom is watching from Perth. Adelaide needs a fourth set of decision table rules, a fourth delivery schedule configuration, and a fourth environment for the deploy pipeline.
The decision tables are the easy part. The framework that Maya and Anika built is parameterised – plug in the local produce data, the rules engine generates the substitution logic. Adelaide’s tables take Priya half a day. The architecture from the DDD work is paying off exactly as intended.
The deploy pipeline is the hard part. Tom deploys to Perth, Melbourne, Brisbane, and staging. Each deployment takes forty-five minutes: pull code, run tests, build the package, SSH into the production server, stop the application, deploy, migrate the database, restart, manually check the key pages load. Three hours for a full release. Only Tom, Priya, and Kai can do it. One in five deployments has an issue – a migration that needs manual intervention, a configuration variable that differs between environments.
“Adelaide is the fourth production environment,” Tom tells Charlotte. “Every release day, I spend the morning deploying. Five cities will take four hours. Eight cities – six hours. A full working day.”
“That’s not sustainable.”
“No. I need a CI/CD pipeline and a platform engineer to build it. Three months after hire.”
Tom writes the job description that evening. It’s the first time he’s written a job description for a role he can’t do himself. He can build a CI/CD pipeline – he’s built simpler ones before. But he can’t build it while running the engineering team, supporting four city expansions, and managing a growing squad.
He texts Sarah: Writing a job description for someone who does the thing I wish I could do.
Sarah: Because you want to do it yourself.
Tom: Obviously.
Sarah: Write the job description for what the company needs. Not for what you miss.
Tom deletes the draft and starts again.
What the playbook looks like
By the end of February, the Adelaide soft launch is running. Two hundred subscribers, seven farms, one logistics hub, Friday deliveries. The playbook has been tested and refined.
Charlotte and Diane sit down together to write the final version. It’s the first time they’ve co-authored anything, and the process is revealing. Charlotte writes in structured sections with headings, subheadings, and bullet points. Diane writes in paragraphs that read like advice from someone who’s done it before.
The playbook they produce has both qualities.
The GreenBox City Expansion Playbook
Purpose: A repeatable process for launching GreenBox in a new city. Separates universal elements (replicate) from local elements (discover).
Phase 1: Market Assessment (1 week)
- Confirm the job-to-be-done exists in this market (10 JTBD confirmation interviews)
- Map local competition and pricing
- Identify candidate farms within 100km
- Estimate unit economics
- Go/no-go recommendation
Phase 2: Partner Onboarding (4 weeks)
- Farm visits and partner selection (led by farm partnership lead)
- Event Storm: local supply chain divergences only
- Build city-specific decision table rules
- Establish logistics model (direct collection or hub-and-spoke, depending on geography)
- Example Map city-specific business rules
- Delivery schedule and courier partnerships
Phase 3: Soft Launch (4 weeks)
- 200-subscriber cap, scale in 50-person increments
- Daily monitoring of delivery reliability and substitution accuracy
- Weekly retros with local team
- Adjust decision tables based on real data
- Confirm unit economics
- Exit criteria: four consecutive weeks at 95%+ delivery reliability
Dependencies:
- Farm partnership lead available (Dave for first two cities, then new hire)
- Engineering: deploy pipeline supports new environment (capacity limit: 5 cities with current manual process)
- Hiring: local operations coordinator, local delivery lead
What we learned in Adelaide:
- Hub-and-spoke logistics model needed when farm density is low
- Event Storm the divergences, not the whole domain
- Substitution engine must be parameterised by region
- Delivery day flexibility is a city-level variable, not a company-level constant
“It’s not perfect,” Charlotte says.
“It doesn’t need to be perfect,” Diane says. “It needs to be useful. Perfection is the enemy of the second city.”
Charlotte looks at her. “That sounds like something from a motivational poster.”
Diane grins. “It was on the wall at Sunridge. I hated it. But I also used it every time someone wanted to spend another month refining a process that was already working.”
Charlotte shakes her head, but she’s smiling. “I’ll accept ‘useful and improvable.’”
“That’s what every good playbook is. A living document that gets better every time you use it.”
The constraint nobody is ignoring
Tom sends the board an email – his first direct communication with the board, not routed through Maya. Patricia Osei told him to do it, at the board meeting where he presented the technology update.
The email is short. Tom doesn’t waste words in writing any more than he does in conversation.
Subject: Engineering constraint on expansion timeline
The current deployment process supports five city environments with significant manual effort. Beyond five cities, the process becomes a bottleneck that will affect delivery reliability and engineering velocity. We need a CI/CD pipeline and a platform engineer to build it. I’ve written the job description. Hiring should start immediately. Estimated time to capability: three months after hire.
This does not block Adelaide or Sydney. It blocks city six.
Patricia replies within an hour: Thank you for the clarity. This is exactly what I need in a board communication. Short, specific, actionable. Let’s discuss hiring timeline at the next board meeting.
The Cerulean investor replies the next day: Agree. Platform investment is in the use-of-funds plan. Prioritise this hire.
Maya forwards Tom’s email to Charlotte with one line: He’s getting good at this.
Charlotte replies: He’s always been good at this. He just didn’t know who to tell.
Adelaide is working. The playbook is written. The first test confirmed that three years of discovery – Event Storming, Assumption Mapping, JTBD, Decision Tables, Example Mapping – created knowledge that scales. The playbook doesn’t replace discovery. It focuses discovery on the things that are actually unknown.
But a playbook for cities doesn’t solve the other half of going national. GreenBox has thirty-five people. It needs eighty. That’s a new hire every two weeks for eighteen months. The culture that worked when everyone knew everyone – when Maya could pull a chair up to anyone’s desk and talk about the farm network or the matching engine or what Mrs Patterson said in last week’s interview – doesn’t work when half the company has never met Maya.
Hiring at Scale (coming 5 January) is next. And it turns out that hiring well at speed is its own kind of discovery problem.