Rollout governance for a restaurant group: gates, budgets and recovering from a failed GC

What goes wrong when a group builds several sites at once

In Cost and Length of Time for Full Build Out?, read through controller-supplied text, owners report projects ranging from about five months to two years. One owner fired a GC mid-project and managed the rest themselves. Another says hiring a GC that worked only on restaurants and hospitality made all the difference. In what goes wrong that costs you the most money?, seen as search excerpts, multi-unit owners name a weak development budget, delays while costs are already running, and bad managers. In This restaurant was supposed to be finished this week, also excerpts, one commenter describes three years of rent on a second location that never opened.

These are individual accounts, not a measured failure rate. They point to a governance problem rather than a single bad contractor. When a group has several sites in progress, the risks multiply in ways a single-site owner never sees. Openings overlap and stretch management. Budget overruns at one site are quietly covered by another. A failed contractor at one site can leave rent running for months while the team is busy elsewhere. The tools below make each site's status and exposure visible early enough to act.

Stage gates for every site

A stage gate is a decision point where a site may proceed only if specified evidence exists. The value is consistency. The fourth site goes through the same checks as the first, even when the team is tired and the landlord is pushing. Keep gates short and evidence-based. The question at each gate is 'what document shows this is true?', not 'does everyone feel good about it?'

A development budget that covers more than construction

Construction is usually the largest line, but the overruns that hurt groups are often elsewhere. Pre-opening payroll runs longer when the opening slips. Rent starts on the lease's commencement date, whether or not you are open. Equipment ordered for a site that stalls ties up cash. Build a development budget for each site with every category below, approved at G1 and re-forecast monthly.

Keep sites' budgets separate. A savings at one site should not silently fund an overrun at another. Moving money between sites should be an explicit decision recorded on the dashboard, so the group knows its true position.

• Hard construction cost, with the contract and approved change orders
• Soft costs: design, engineering, permits, expediting, testing, legal
• Furniture, fixtures and equipment, including owner-furnished kitchen equipment
• Pre-opening costs: training payroll, recruiting, opening inventory, marketing
• Carrying costs from lease commencement to opening: rent, common-area charges, utilities, loan interest
• Construction contingency for scope and hidden conditions
• Schedule reserve: a set number of weeks of carrying cost for delay

Two reserves: construction contingency and schedule reserve

A single contingency line hides the difference between a cost problem and a time problem. Construction contingency pays for scope: a hidden condition, a design gap, a required change. Schedule reserve pays for time: weeks of rent, payroll and interest while the opening slips. Size the schedule reserve in weeks of carrying cost, so everyone can see what a delay of a given length does to the budget.

Neither reserve should be a borrowed percentage. Size the construction contingency from the site's risk register, as described in our hidden-conditions article. Size the schedule reserve from the realistic range of permit, supply and construction durations in that jurisdiction and building type. Track both on the dashboard as amounts remaining, not just amounts budgeted.

The portfolio dashboard

One page, updated weekly, covering every site in progress. The most useful single number is forecast opening date compared with the baseline, multiplied by weekly carrying cost. That turns a slipping schedule into money the whole leadership team understands. Add a rule for overlap. In this example the group allows no more than one site in G5 or G6 at the same time, because its trainers and opening managers cannot be in two places at once. Set your own limit based on the people you actually have.

The hypothetical dashboard below shows three sites. Site 1 is on track. Site 3 is early in construction with a modest slip. Site 2 is the problem: its forecast has slipped 10 weeks because its GC has stopped performing. That is $55,000 of carrying exposure, more than its $44,000 schedule reserve can absorb.

Replacing a general contractor mid-project: the playbook

Termination is a legal step with consequences, so follow the contract's procedure. That usually means written notice of default, a cure period and then termination, as our delay article describes. Have a construction attorney in the property's jurisdiction run it. If there is a performance bond, notify the surety exactly as the bond requires, because the surety may have the right to arrange completion. The operational work below runs in parallel and should start as soon as the notice to cure goes out.

• Secure the site: change access codes and locks where the contract and lease allow, and confirm builder's risk and liability coverage continue
• Document the work in place: dated photos and video of every room, plus an inventory of stored materials and owner equipment
• Collect the record: permits, inspection reports, drawings, submittals, change orders, pay applications and waivers
• Map lien exposure: list subs and suppliers, amounts they say are unpaid, and the waivers you hold
• Ask the building department how a change of contractor of record is handled for open permits; do not assume
• Decide whether to engage existing subs directly, which depends on their subcontract terms and on what they are owed
• Have the replacement GC audit the work in place before pricing completion, and accept that it will price unknown risk

Hypothetical worked example: Site 2's GC fails at 70% complete

All figures are hypothetical. Site 2 has a $520,000 contract with 10% retainage. Certified work stands at 70%, or $364,000. With $36,400 retained, the owner has paid $327,600. The unpaid contract balance is $156,000 of uncompleted work plus $36,400 of retainage, a total of $192,400. After the contract's notice and cure steps, the owner terminates. A replacement GC audits the site and prices completion at $230,000, higher than the $156,000 remaining because it must take over unfamiliar work and correct some defects. Two subs report $18,000 unpaid by the original GC. The $230,000 quote covers only completion and corrections; it excludes those $18,000 of sub claims, which are counted separately below, and we assume no work is paid for twice. On counsel's advice, the owner resolves those claims in exchange for waivers.

Cash out: $327,600 already paid + $230,000 completion + $18,000 sub claims = $575,600, which is $55,600 over the original $520,000 contract. If the contract allows completion costs to be charged against the unpaid balance, that $55,600 becomes a claim against the original GC or its surety, subject to the contract and local law. The 10-week slip at $5,500 a week adds $55,000 of carrying cost, so total exposure is $110,600. Site 2's own reserves cover $89,000: $45,000 of construction contingency and $44,000 of schedule reserve. The remaining $21,600 must come from a group-level reserve or another decision, recorded on the dashboard. Whether any delay cost is recoverable depends on the contract's delay and liquidated-damages terms.

The opening gate is more than construction sign-off

Gate G6 exists because building, health and fire approvals are not the whole permission to operate. In DC, for example, the Department of Buildings' Certificate of Occupancy page describes Certifi submission with an Eating Establishment Questionnaire and menu, and an application approval that typically takes about seven business days. That is not a full closeout schedule. DLCP's Food Services licensing page (DC) requires a Certificate of Occupancy, a DC Health food manager's certificate, the DC Health inspection report, tax registration and Clean Hands certification, and corporate registration where applicable, with applications made through BOSS. Alcohol needs a separate application through DC's Alcoholic Beverage and Cannabis Administration; see its alcohol applications page. Ask the agency about timing rather than assuming it. Each new jurisdiction in a rollout needs its own map of agencies and sequence, built at G2, not discovered at G6.

Closing the loop after each opening

Gate G7 is where a group gets better or repeats itself. Finish what G5 left provisional: the remaining punch items, final manuals, as-builts, final waivers and retainage, filed alongside the final inspection records gathered at G6. Within about 90 days of opening, compare final costs with the G1 budget by category, opening date with baseline, and the reserve used with the reserve planned. Read the change-order log, the commissioning deficiency log, the punch list and early warranty calls for patterns. Record each lesson as a specific change: a spec book revision, a new investigation at G1, a contract clause, or a different bidder list. Then check at the next site's G0 that the change was actually applied. A lesson that never reaches the next site's documents has not been learned.