Diligence for Multi-Location Platforms: What to Model Pre-Close

After thirty-plus pre-close diligence engagements on multi-location service platforms — dental DSOs, HVAC roll-ups, childcare academies, fitness — I can tell you the predictive difference between the deals that worked and the deals that didn’t isn’t top-line growth assumptions. It’s how well the model treats branch-level heterogeneity, the second derivative of performance most funds underweight because it’s easier to build a top-down growth number than to understand why branch 7 under-earns branch 3 by 400 bps.

A good pre-close model for a multi-location platform answers four questions honestly. Most models answer one well and the rest poorly. Here are all four.

Question 1: Is the top quartile the real operating benchmark, or an outlier?

Every multi-location target has at least one branch that makes the group look great. The question is whether that branch is a reproducible operating model or a local anomaly — a flagship with a charismatic GM, a grandfathered lease, or a market position that doesn’t replicate.

The test: segment the top-quartile branch performance by variables you can control (service mix, labor composition, pricing structure) and by variables you can’t (market size, demographics, lease economics). If the top-quartile performance is dominated by controllable drivers, it’s a real benchmark. If it’s dominated by uncontrollables, your thesis that the bottom quartile can converge to it is weaker than the deck implies.

Question 2: How brittle is the labor model?

In service platforms, labor is 50–75% of cost. Diligence models that express labor as a flat % of revenue miss the variability that actually drives misses post-close. The right expression: labor as step-function against utilization, with explicit handling of overtime thresholds and benefit-load curves.

Three specific tests:

• Overtime sensitivity. What happens to EBITDA if overtime rises from current 4% to 7% of base? (Answer: at typical prime-cost ratios, roughly −250 bps of EBITDA margin.)
• Step thresholds. Where in the utilization curve does the model add a new FTE? The deal-case assumes utilization rises 300 bps; does that trip any step-cost thresholds?
• Wage inflation. What’s the model’s wage escalator? If your market is tight, 2.5% isn’t the right number — 4–5% is.

Question 3: Do the covenants survive the downside scenario?

Most platform acquisitions are lendered with senior debt at 3.5× to 4.5× EBITDA with a DSCR covenant at 1.20 to 1.35 and a leverage covenant stepping down annually. Your model probably shows comfortable compliance in the base case. That’s not the question.

The question is: in a downside where revenue drops 12% and labor inflation runs 4.5%, does DSCR hold above the covenant floor? What about minimum liquidity? What about the step-down covenant in year two?

If the answer is “no” in any downside scenario with plausible probability, you need either different debt structuring pre-close (looser covenants, larger cushion) or different cushion in the model.

Question 4: What’s the working-capital rhythm you’re inheriting?

Multi-location service platforms usually have some form of unearned-revenue mechanic (memberships, deposits, prepaid services) that makes the GAAP EBITDA look better than the cash EBITDA in a growth period and worse in a shrink period. Diligence models that treat GAAP EBITDA as distributable cash will consistently over-estimate post-close cash flow.

Three adjustments that matter:

• Unearned revenue reversal. If growth stalls, the deferred revenue balance stops growing — EBITDA flattens and cash flattens harder.
• AR aging. If the platform has grown rapidly, AR has grown with it. What happens to cash if you normalize AR days back to the pre-growth rate? For many platforms, $300k–$800k of one-time cash release.
• Cap-ex catch-up. Has the platform under-invested in maintenance cap-ex during the growth period? If so, budget for a catch-up in year one, which the lender won’t.

How Aziell supports diligence work

While Aziell is primarily a post-close operating platform, we have a diligence mode where a target’s books can be imported into the driver-based model in a read-only sandbox. This lets you segment branch-level performance the way a post-close operator would, stress the labor model, and run downside scenarios against real covenants. The value is less the diligence artifact itself (your QofE provider delivers that) and more the continuity — the same driver model you use to evaluate the target pre-close becomes the operating plan post-close, with no rebuild.

For the full post-close operating framework, see value creation plans that actually move EBITDA.

The pattern across the losers

Every platform deal our desk has seen go sideways within two years traced back to one of these four questions being answered wishfully at close. Top quartile turned out to be a flagship. The labor model was too variable. The covenant cushion was 60 bps. The working-capital rhythm inverted. Any one of these is survivable; two together usually is not.

Ninety minutes spent hardening the pre-close model against these four questions is the single highest-leverage diligence activity a fund can do on a multi-location platform.