EBITDA measures the cash profit a business generates, stripped of accounting noise that varies from owner to owner. The shift away from revenue multiples was driven by one fatal flaw of revenue: it ignores profitability. EBITDA is comparable across agencies of different sizes, tax structures, and capital structures — this comparability is why it became the standard. The normalization step strips away owner-specific items to reveal the true, transferable cash flow a new owner would actually receive.
The five add-back categories.
Compensation — the biggest single category.
Owner salary above market GM rate (market: $80K–$150K depending on agency size). Non-working family members on payroll. Excess bonuses or owner distributions above reasonable performance levels. Unused PTO or severance accruals (non-recurring). The compensation category is where buyers and sellers most often disagree — the buyer pulls the market rate down; the seller pulls it up. Documentation matters: get an independent compensation benchmark for your agency size and region before negotiating, not after.
Discretionary expenses — the personal-perk category.
Personal vehicle leases (typically $10K–$30K). Travel — personal components of business trips. Meals & entertainment above a reasonable business budget. Personal dues, subscriptions, certifications not job-critical. These are typically defensible if documented but invite scrutiny when undifferentiated from legitimate business expenses. The fix is granular records: which trip was the conference, which was the family vacation booked through the business card. Buyers test these aggressively — a vague "miscellaneous owner benefits" line will be challenged or rejected.
Facility & infrastructure — the related-party category.
Excess office space versus operational requirement. Above-market rent paid to a related-party landlord (owner personally owns the building). Redundant roles created for owner-specific reasons. Related-party rent is the most common item here — when the owner's LLC pays rent to the owner's other LLC at above-market rates, the excess is a normalization. Buyers verify with market-rent comparables. Documenting the rate gap explicitly turns this into a clean add-back rather than a contested one.
One-time / non-recurring items — the cleanup category.
Litigation settlements and unusual legal fees. One-time severance or termination costs. System implementations or one-time IT capital spend. Unusual insurance claims or losses. The category is straightforward in principle — if it won't recur under new ownership, add it back. The discipline is in proving it won't recur. A $50K legal settlement that was the third such settlement in five years is not non-recurring. A one-time IT implementation that's part of an ongoing modernization roadmap is not one-time. Buyers test the recurrence question rigorously, and over-claiming here is one of the most common over-normalization patterns.
Anomalous-year normalization — the smoothing category.
If a single year was depressed by a one-time client loss or inflated by a non-recurring contract, smooth across 2–3 years. Buyers want a normalized trend, not a one-year snapshot that may be anomalous. The discipline is symmetry: smooth down anomalously good years too, not just anomalously bad ones. A seller who only adjusts toward higher EBITDA loses credibility on the entire schedule.
The worked example.
Agency profile: $4M revenue, $850K reported EBITDA. Add back owner salary excess ($200K paid, $120K market rate): +$80K. Add back personal vehicle lease: +$18K. Add back excess facility costs: +$40K. Add back one-time legal fees: +$12K. Add back non-working family member salary: +$25K. Deduct underpaid key staff gap to market: -$15K. Deduct deferred IT infrastructure: -$8K. Normalized EBITDA = $1,002,000.
At 10× EBITDA: reported ($850K) → $8.5M vs. normalized ($1.002M) → $10.02M. The normalization alone creates $1.5M in additional value — without selling a single new policy.
The multiple spectrum, anchored to canonical bands.
Smaller SMA (under $1.5M revenue, <10% margins): 4–6× — the 4–6× distressed-or-internal band. Well-run mid-market ($2M–$5M revenue): 8–10× — the 8–10× market band. Premium agency ($5M+ revenue, 15%+ margins, >90% retention): 10–12× — the 10–12× competitive band. Exceptional top-tier (best-in-class across all metrics): 12–19× — the 12–19× PE kill-zone band for bolt-on / platform-thesis assets.
The red flags that undo a normalization schedule: over-normalization — if Normalized EBITDA is more than 15% above reported, buyers get suspicious. Most well-run agencies normalize 5–12% above reported. Vague or undocumented add-backs — "Miscellaneous owner benefits" will not hold. Tax-return conflicts — if an expense was deducted on the tax return, it cannot be added back. Speculative add-backs — savings a new owner might achieve don't count. Single-year normalization — buyers want a smoothed 2–3 year trend.
Each adjustment needs a paper trail: dollar amount with invoice or statement, rationale (owner-specific, non-recurring, etc.), evidence it occurred (P&L line item, invoice, email), and why it won't recur. Working with a CPA before the sale to formally document the normalization schedule is standard practice for sophisticated sellers.
Terminology on this shelf
- Normalized EBITDA (Pro Forma EBITDA / Adjusted EBITDA)
- EBITDA adjusted for owner-specific and non-recurring items to reflect true, transferable cash flow under professional management.
- Normalization (Recasting)
- The process of cleaning up a P&L by removing owner-specific expenses and one-time items.
- Add-Back
- An expense added back to profit because it won't recur under new ownership.
- Downward Adjustment
- Reduction to Normalized EBITDA reflecting future costs the buyer will absorb — underpaid key staff brought to market, deferred maintenance, infrastructure gaps.
- Over-Normalization
- Normalization of more than 15% above reported EBITDA — triggers buyer skepticism and rigorous re-examination of the schedule.
- Multiple Arbitrage
- PE strategy of buying agencies at a lower bolt-on multiple and combining them to achieve a higher platform multiple.