The Saudi Scale Playbook

Most groups plan expansion as a list of sites; the groups that keep their margins plan it as a sequence. No stage is funded until the previous gate is passed. Skipping a stage only moves the failure somewhere more expensive.

1. 1

   Stage 1: Prove the unit

   Does the format make money reliably, without heroics?

   Gate to pass Gate 1: ≥ 3 comparable units within plan for 2 consecutive quarters; reference payback ≤ 30 months

   Tool Unit Proof & Capacity Check

   Gate 1 →

2. 2

   Stage 2: Choose the site

   Will the site pay back, and will the network be better off?

   Gate to pass Gate 2: the Doctrine's opening gate

   Tool Opening Gate Calculator

   Gate 2 →

3. 3

   Stage 3: Open and ramp

   Is the branch tracking its approved model?

   Gate to pass Gate 3: ramp index on plan at day 90; service standard held at peak

   Tool 90-day plan (§6)

   Gate 3 →

4. 4

   Stage 4: Densify the city

   How many branches can this trade area carry?

   Gate to pass Gate 4: S_b of the next branch still positive; measured cannibalization within plan

   Tool Opening Gate Calculator, per trade area

   Gate 4 →

5. 5

   Stage 5: Replicate to a second city

   Does the system travel without the founder's attention?

   Gate to pass Gate 5: first-wave GRI at or above the group floor

   Tool Replication Index Calculator

   Gate 5 →

6. 6

   Stage 6: Industrialise (30+)

   Can head office, channels and supply grow more slowly than revenue?

   Gate to pass Gates 6–8: G&A dilution < 1 and a bench of ready managers; every channel passes its floor; any central kitchen pays back within threshold

   Tool Scale Overhead Check, Channel Contribution (Doctrine), Central Kitchen Payback

   Gate 6 – 8 →

7. 7

   Stage 7: Enter the GCC

   Will a proven Saudi system work under different rules?

   Gate to pass Gate 9: Stage 5 passed in ≥ 2 Saudi cities; entry mode chosen; regulatory path verified; pilot and kill criteria set before entry

   Tool City Prioritization Board, Replication Index

   Gate 9 →

Three rules govern the sequence:

1. Gates are passed with evidence, not conviction. Each gate names its metric and its data.
2. Stages overlap; gates do not. A group can densify Riyadh while testing a second city, but no city reaches Stage 5 before Gate 4 has shown how dense a trade area can get.
3. "Not yet" is an outcome, not a failure. The capital returns to the next-riyal list (Doctrine §6) and competes again.

Each gate follows the same pattern: objective, what must be true, metrics, decision rule, typical failure, tool.

A city score tells you where to look, not where to sign. The model ranks candidate cities, in the Kingdom or the GCC, for a site study and a demand test; each must then pass Gates 2 and 5 on its own numbers. It ships with no city scores: scores that are not yours allocate your capital on someone else's assumptions.

Score each factor 1–5, always oriented so 5 is favourable (low rent = 5; heavy competition = 1). Tag each input as measured, estimated or assumed.

City score = Σ (weight × score) ÷ Σ weights

The board sets the weights to total 100, and the result falls between 1 and 5. Three safeguards apply:

1. Knockouts. Any factor scored 1 removes the city from this cycle. A market you cannot supply is not a market.
2. Assumption share. If more than a third of the weight rests on assumed inputs, run a demand test before ranking the city.
3. Flip test. Move any weight by ±10 points. If the top two cities swap places, they are tied: test both.

Workforce is a design decision, not a compliance afterthought. Every opening changes the headcount base, and so the group's nationalization position. The Saudization–Automation balance model plans hiring and automation together.

Variables (all yours)

• N_req: headcount the operation needs, including planned openings
• r: your required ratio, from Qiwa
• S: current Saudi employees
• W_s and W_e: average monthly cost of a Saudi and an expatriate employee in the roles concerned
• L: monthly government and statutory costs of an expatriate role
• A: roles automation can remove at standard
• C_auto: monthly automation cost (amortized capex, maintenance and software)

• Required Saudi employees= ceiling(r × planned headcount).
• New hires= required − S, if positive.
• Monthly cost of a plan= Saudis × W_s + Expatriates × (W_e + L) + C_auto.

Compare two plans:

• Plan H hires to the ratio on N_req.
• Plan A automates A roles, then hires to the ratio on N_req − A.

Automation lowers the base only by the headcount it actually removes. Check how the counting rules treat each role type before you rely on that effect.

Otherwise, hire and invest in the role.

• Compliance is a gate item. No opening is approved if its staffing plan would take the entity below its target band.
• Hire ahead of the site. Build your measured hiring lead time into the opening plan. Train bench managers in live branches before the lease starts running.
• Design careers, not ratio roles. A path from team member to shift lead to branch manager to area manager keeps the people the group trains. Roles that do not change the operation fail both the ratio and the P&L.
• Check support programmes. The Human Resources Development Fund (HRDF) and HRSD run training and wage-support programmes. Their terms change, so model them only once they are confirmed in writing.
• Re-run the model at every wave.

The ownership model is a capital decision and a replication decision. Franchising does not solve a replication problem; it exports it with your brand on the door.

1. GRI below the floor: company-owned only. Fix the system before anyone else runs it.
2. Company-owned is the default wherever a site passes the gate and the group has the capital and the bench. It keeps the contribution and the data.
3. Franchise when capital or bench is the binding constraint, the GRI is at the floor and the system is documented. The franchisee must also pass the payback bands: franchisee payback = capex ÷ (net monthly contribution − royalties − fees). A franchise that cannot pay back its franchisee damages the brand.
4. Choose a joint venture when the partner brings a scarce asset, such as sites, licences, a delivery network or an audience. Value that asset as you would a bridge (Doctrine §6). If the partner brings only capital, price that capital against the alternatives before giving up control.
5. Choose a master franchise for GCC markets that are small relative to the management attention they need, or where local rules favour local operators.
6. Verify the legal frame. Franchising in the Kingdom is regulated. Confirm disclosure and registration requirements with the Ministry of Commerce, and with the equivalent authority in each GCC market.

A new branch has two jobs: reach its approved model at standard, and produce clean data for the next decision. Gate 3 is judged on day 90, but its data is set up before day 1.

• Site approved through the opening gate, with any lever plan attached.
• Licences confirmed as issued, not "in process"; inspections scheduled.
• Manager from the bench; team trained in a live branch; post-opening nationalization ratio checked (§4).
• Weekly ramp plan approved for weeks 1–13: sales, contribution and orders by channel.
• Sister-branch like-for-like baseline recorded.
• Supply, par levels, point of sale, platform listings and direct channel tested.
• Catchment partnership list ready: offices, residential compounds, nearby businesses.

1. Ramp. What is the ramp index? Defaults: ≥ 0.9 on plan; 0.75–0.9 recovery; < 0.75 escalate.
2. Contribution. Is contribution after rent where the model placed it for month 3?
3. Capacity. Is peak U within standard, and is service time holding?
4. Cannibalization. How do sister branches compare with baseline at month 3? The measured rate goes into this trade area's next S_b.
5. Replication. Did the numbers hold after the launch team left? This feeds the month-6 GRI.

Before approving the next five branches, the board sees one pack. It shows where the group stands, what the last openings actually did, and how each new site passes the gate. Everything else goes to the appendix.

Resolution wording

"The board approves sites [ ] and [ ] against the models in this pack. It approves site [ ] subject to the attached lever plan, and records site [ ] as not yet. Each approved site will be reviewed at day 90 against the thresholds on page 8. No further openings in [trade area] will be tabled until measured cannibalization at month 6 has been reported."

Keep out: follower counts, impressions, sites without models, revenue growth without like-for-like. If vanity metrics outnumber governing ones, cut the pack before the meeting (Doctrine §8).

Figures as reported by the client groups, for the periods stated where stated. None was measured against a control group. “Across his career” figures are not attributed to any one group.

The groups that scale well are not the fastest openers. They know, at every stage, which gate they stand at and what it takes to pass it. They sign sites that leave the network better off. They open a second city to test the system, not the brand. And they say "not yet" early, while it is still cheap.

Revenue is designed, not discovered.

Start with the Branch Gate on your own numbers: five numbers about the site and a three-word verdict, sign, sign with conditions, or not yet. Then every engagement starts with a 30-minute Scale Review. If there is a fit, the next step is a fixed-scope Branch Expansion Diagnostic. It covers:

• a payback and cannibalization model for your branches and the next 5–10 sites;
• a Geographic Replication Index score;
• capital-allocation rules;
• a board memo built on the pack in §7.

Groups that want continuing support move to a retained Revenue Engineering Advisory.

Try the Branch Gate on your numbers

Book a 30-min Scale Review