Your business is worth a multiple of its normalised earnings, and that multiple is set mostly by size and transferability — businesses under $500K in seller’s discretionary earnings have traded at roughly 2.0× for four consecutive years, while companies with $5M–$50M of EBITDA reached about 6.5×. That gap is why owners who apply a headline multiple from a much larger company routinely overestimate their own by two to three times.

Best for: owners of profitable businesses roughly $500K–$50M in enterprise value who have a number in their head and have never tested it against comparable transaction data. Skip if: you have a current third-party valuation from a credentialed appraiser, or you’re pre-profit — a pre-profit company is valued on entirely different logic and none of this applies.

What is my business actually worth?

Normalised earnings multiplied by a market multiple, adjusted for the specific risks a buyer sees in your business. That’s the whole model. The difficulty isn’t the equation — it’s that both inputs are usually wrong when an owner estimates them, because earnings get inflated by add-backs that won’t survive scrutiny and the multiple gets borrowed from companies several times larger.

A defensible answer is a range, not a number, and it comes with the comparable transactions that produced it.

Why do business owners overestimate what their business is worth?

Four mechanisms, and most owners are running at least two. They anchor on revenue rather than earnings. They borrow a multiple from a much larger company or a venture-funded competitor. They count add-backs a buyer’s accountant will reject. And they reason backwards from what they need — a retirement number is not a valuation input.

The size effect is the most damaging because it’s invisible. An owner reading that their industry “trades at 6×” rarely learns that the figure describes companies ten times their size.

Should I use SDE or EBITDA?

SDE — seller’s discretionary earnings — applies to owner-operated businesses, typically smaller ones. It adds back your salary, benefits and discretionary spending, because a buyer stepping into your role receives that full economic benefit. EBITDA applies once a paid management team runs the business without you, because it reflects earnings after a market-rate manager is already paid.

They are not interchangeable, and their multiples are not interchangeable either. Applying an EBITDA multiple to an SDE figure is the most common valuation error made by owners.

What are current business valuation multiples?

By size, using transaction data rather than asking prices:

Industry, growth rate and earnings quality move you within a band. Size determines which band. That ordering is the part most owners have backwards.

Figures verified 13 August 2026 against the sources listed at the foot of this guide. Multiples move with credit conditions and deal volume — re-check before relying on them.

Why should I never let AI calculate my valuation?

Because language models produce arithmetic errors with total confidence, and a valuation is a decision-grade number. A model that adds a column wrong will present the wrong total in the same authoritative tone as the right one, and nothing in the output signals which you received.

Every prompt in this guide runs in formula mode. The model produces the equation, names each input, and states where that input comes from. You put the numbers in a spreadsheet. This is slower by about four minutes and it removes an entire category of silent failure.

FORMULA MODE — MANDATORY
Do not perform any arithmetic. Do not state a computed result.
Output only:
  (a) the formula, written in plain algebra
  (b) each input, named, with its unit
  (c) where each input comes from, and how confident you are
  (d) the spreadsheet formula I should type
If you catch yourself producing a number, stop and give me the
equation instead.

What factors suppress a business valuation?

Buyers discount for anything that makes the earnings less likely to continue without you. In rough order of impact: owner dependency, customer concentration, unreviewed or inconsistent financial records, undocumented processes, thin management depth, short or terminable customer contracts, supplier concentration, and declining or volatile margins.

The useful property of that list is that most of it is fixable in 12–24 months, and fixing it moves the multiple — which compounds across every dollar of earnings, unlike revenue growth, which only adds at the current multiple.

Prompt 1 — The Earnings Normalisation Build (start here)

You cannot pick a multiple until you know what you’re multiplying. This prompt determines which base applies and builds it in formula mode.

You are a transaction accountant preparing a normalised
earnings figure for a business sale. You are not the owner's
advocate.

BUSINESS
  Revenue, last 3 years:      [YEAR: $, YEAR: $, YEAR: $]
  Net income, last 3 years:   [$, $, $]
  Owner's W-2 salary:         [$]
  Number of working owners:   [N]
  Hours/week owner works:     [N]
  What the owner actually
    does day to day:          [DESCRIBE — sales? production?
                               admin? relationships?]
  Is there a general manager
    who could replace them?   [YES/NO — and their salary]
  Interest expense:           [$]
  Depreciation & amortisation:[$]
  Taxes:                      [$]

PROPOSED ADD-BACKS
[LIST EACH: description, annual amount, your justification.
 Include the ones you are unsure about.]

STEP 1 — Tell me which earnings base applies: SDE or EBITDA.
State the reasoning in two sentences. If the business sits at
the boundary, say so and explain what would move it.

STEP 2 — Build the normalisation as a FORMULA, not a total.
List every line, its sign, and its amount as a named variable.

STEP 3 — For each proposed add-back, classify it:
   ACCEPT / PARTIAL / REJECT
   plus the evidence a buyer's accountant would require.

STEP 4 — If EBITDA applies, state the market-rate manager
salary that must be deducted, and where that figure comes from.

FORMULA MODE — MANDATORY
Do not perform any arithmetic. Do not state a computed result.
Give me the formula, the named inputs, and the exact
spreadsheet formula to type. If you catch yourself producing
a number, stop and give me the equation instead.

Prompt 2 — The Comparable Multiple Prompt

Help me identify a defensible multiple range for my business
using comparable transaction data.

MY BUSINESS
  Industry / NAICS:          [DETAIL]
  Earnings base and amount:  [SDE or EBITDA] of [$]
  Revenue:                   [$]
  Growth rate, 3-yr average: [%]
  Recurring revenue share:   [%]
  Geography:                 [DETAIL]

Do the following:

1. Identify which SIZE BAND my earnings place me in, and state
   the multiple range published for that band. Name the source
   and the reporting period for each figure you give me.

2. Tell me explicitly where I should NOT look for comparables
   and why — public company multiples, venture-funded
   competitors, asking prices rather than closed transactions.

3. Give me the industry adjustment, if any, relative to the
   all-industry band figure. If you don't have a sourced
   industry figure, say so rather than estimating.

4. State the three characteristics of my business most likely
   to move me above or below the band midpoint.

5. List the specific databases or reports I should consult for
   real closed-transaction comps, and what each covers.

RULES
  - Cite a source and a period for every multiple you state.
  - If you are uncertain about a figure, say "unverified" next
    to it. Do not smooth over gaps with plausible numbers.
  - Do not compute a valuation. Range only, formula later.

Why the citation rule matters: valuation multiples are exactly the kind of figure models generate fluently and incorrectly. Requiring a named source and period per figure converts unverifiable output into a checkable list.

Prompt 3 — The Factor Audit (what’s suppressing your multiple)

Score my business against the factors that move a valuation
multiple within its size band. Be harsh. Owners consistently
score themselves two points high.

ANSWER HONESTLY:
  If I disappeared for 90 days, what breaks?     [DESCRIBE]
  Largest customer as % of revenue:              [%]
  Top 5 customers as % of revenue:               [%]
  Recurring or contracted revenue share:         [%]
  Average customer tenure:                       [YEARS]
  Financial statements are:                      [TAX RETURNS /
     BOOKKEEPER-PREPARED / REVIEWED / AUDITED]
  Documented processes exist for:                [DESCRIBE]
  Second-in-command who could run it:            [YES/NO — who]
  Revenue trend, 3 years:                        [%, %, %]
  Gross margin trend:                            [%, %, %]
  Employee turnover:                             [%]
  Supplier concentration:                        [DESCRIBE]
  Customer contracts are:                        [AT WILL /
     ANNUAL / MULTI-YEAR — and notice periods]
  Industry direction:                            [YOUR VIEW]

For EACH of the 14 factors:
  1. Score 1–5, where 3 is market-normal for my size band
  2. State the specific evidence a buyer would ask for
  3. Say whether this factor moves me above band, below band,
     or is neutral
  4. If below: what would it take to reach a 3, and roughly
     how long

Then:
  - Rank the factors by how much each is currently costing me,
    highest first
  - Name the three worth fixing before a sale, and the ones
    that are structural and must simply be disclosed and priced

FORMULA MODE — do not compute an adjusted multiple or a
valuation. Give me the factor scores and the DIRECTION and
SIZE of adjustment each implies, expressed as a range.

Prompt 4 — The Range Builder

Build my valuation range as a spreadsheet model. Do not
calculate it.

INPUTS I HAVE
  Normalised earnings (from Prompt 1):  [$] as [SDE / EBITDA]
  Size band multiple range (Prompt 2):  [LOW]x to [HIGH]x
  Factor adjustment direction (Prompt 3):[ABOVE / BELOW / NEUTRAL]
  Factor adjustment size:               [ESTIMATE FROM PROMPT 3]

Produce:

1. The formula for LOW, MID and HIGH enterprise value, written
   in plain algebra with named variables.

2. The exact spreadsheet formulas I should type into cells,
   with a suggested cell layout.

3. A sensitivity table structure: earnings on one axis,
   multiple on the other, showing me which cells to fill.
   Do not fill them in.

4. The three inputs where a small error causes the largest
   swing in the output, so I know what to verify hardest.

5. A plain-language sentence I can use to describe the range,
   with the caveats that belong in it.

FORMULA MODE — MANDATORY. No arithmetic. No computed totals.
If I have given you numbers, use them as named variables, not
as things to multiply.

Prompt 5 — The Overestimation Audit

Adversarial. Run this against whatever number you arrived at, including the one you had in your head before you started.

I believe my business is worth [MY NUMBER].

Here is how I got there: [YOUR REASONING]
Here is the underlying data: [PASTE PROMPTS 1-3 OUTPUT]

Argue that I am wrong and the number is too high. Build the
strongest honest case against my figure.

Specifically check for:
  1. Am I applying an EBITDA multiple to an SDE figure, or
     vice versa?
  2. Am I using a multiple from companies larger than mine?
     By how much?
  3. Which of my add-backs would not survive a buyer's
     quality-of-earnings review?
  4. Am I anchoring on revenue rather than earnings anywhere
     in my reasoning?
  5. Am I citing asking prices rather than closed transactions?
  6. Have I confused enterprise value with what I would
     actually receive?
  7. Is any part of my number driven by what I need rather
     than what the market pays?

For each error found, state the direction and rough magnitude
of the correction — as a percentage or a multiple adjustment,
NOT as a recalculated valuation.

Then: what is the strongest honest case that my number is
RIGHT? If there isn't one, say so plainly.

Do not soften this. An owner who hears an inflated number
from an AI and a real number from the market six months later
has lost six months.

Prompt 6 — The Proceeds Bridge

The gap between the headline number and your bank balance. Most owners have never modelled this and it is routinely a 30–50% reduction.

Build me the bridge from enterprise value to net proceeds in
my pocket. Formula only — do not calculate.

WHAT I KNOW
  Enterprise value estimate:        [$]
  Outstanding debt:                 [$]
  Excess cash on balance sheet:     [$]
  Expected deal structure:          [ALL CASH / EARNOUT /
                                     SELLER NOTE / ROLLOVER]
  Advisory arrangement:             [BROKER % / INVESTMENT
                                     BANK FEE / NONE YET]
  Entity type:                      [S-CORP / C-CORP / LLC /
                                     SOLE PROP]
  State:                            [STATE]
  Years held:                       [N]
  Approximate tax basis:            [$ OR "unknown"]

Lay out every line of the bridge in order, as named variables:

  Enterprise value
    − debt repaid at close
    + excess cash retained (if any)
    = equity value
    − advisory and legal fees
    − escrow / holdback (state typical % ranges and cite
      a source, or mark unverified)
    − earnout at risk (portion not payable at close)
    = cash at close, pre-tax
    − federal tax
    − state tax
    = net proceeds

For each line:
  - The formula
  - Where the input comes from
  - Whether it is negotiable, and by roughly how much

On taxes specifically: explain WHICH questions determine my
rate — asset vs stock sale, purchase price allocation,
ordinary income vs capital gain treatment, state residency,
entity type — rather than estimating a rate. Do not state a
tax percentage.

FORMULA MODE — MANDATORY. This is exactly the calculation
where an arithmetic error changes a life decision.

Prompt 7 — The 24-Month Value Plan

Turn my factor scores into a work plan ranked by value
created per unit of effort.

FACTOR SCORES: [PASTE PROMPT 3 OUTPUT]
Current normalised earnings: [$]
Target exit window: [MONTHS]

For each factor scoring below 3:

1. The specific work required to reach a 3, described as
   tasks rather than outcomes
2. Realistic elapsed time, accounting for the fact that some
   fixes need time to demonstrate — a documented process is
   worth little until it has run for a year
3. Cost, in money and in owner attention
4. The multiple impact, expressed as a RANGE and as a
   direction — not a computed valuation increase
5. Whether the fix is credible if started late. Some things
   read as genuine improvement; others read as staging.

Then produce:
  - A sequenced 24-month plan, since some fixes depend on
    others being done first
  - The three highest-return actions
  - Anything that CANNOT be fixed in my window and must be
    disclosed and priced instead
  - The one thing most owners in my position do that adds
    effort without adding value

FORMULA MODE — express all value impact as multiple ranges
and directions. Do not compute a projected valuation.

What separates a strong valuation prompt from a weak one?

Real example: input → output

Illustrative composite. Figures constructed from common patterns, not a real transaction.

Step 1 — Prompt 1 output (excerpt). The model declines to use EBITDA and explains why:

SDE = NET_INCOME
    + OWNER_SALARY
    + OWNER_PAYROLL_TAX
    + INTEREST
    + DEPRECIATION
    + AMORTISATION
    + ADDBACKS_ACCEPTED
    − ADDBACKS_REJECTED

Spreadsheet: =SUM(B2:B8)-B9

Add-back rulings:
  Vehicle (personal use portion)   ACCEPT   partial, 60%
  Owner health insurance           ACCEPT
  Spouse on payroll, no role       ACCEPT
  "One-time" legal, 3 yrs running  REJECT — recurring
  Trade show travel                REJECT — operating cost
  Home office                      PARTIAL — needs support

Owner computes in a spreadsheet: SDE = $780,000.

Step 2 — Prompt 2 output (excerpt):

Step 3 — Prompt 3 factor audit, three lowest scores:

Step 4 — Prompt 4 range formula. The model produces the equation; the owner computes:

EV_LOW  = SDE × MULT_LOW
EV_MID  = SDE × MULT_MID
EV_HIGH = SDE × MULT_HIGH

Band range:        2.5x – 3.3x
Factor adjustment: BELOW midpoint (three factors at 1–2)
Adjusted range:    2.4x – 2.9x

Owner's spreadsheet result:  $1.87M – $2.26M
Owner's original belief:     $4.00M
Gap:                         roughly 45% too high

Step 5 — Prompt 6, the proceeds bridge. This is where the conversation usually changes:

PROCEEDS BRIDGE — formula structure

  ENTERPRISE_VALUE            (midpoint ≈ $2.07M)
− DEBT_REPAID                 (equipment notes, line of credit)
+ EXCESS_CASH_RETAINED        (per working capital peg)
= EQUITY_VALUE
− ADVISORY_FEES               (broker %; negotiable)
− LEGAL_AND_ACCOUNTING
− ESCROW_HOLDBACK             (unverified — confirm % with advisor)
− EARNOUT_AT_RISK             (portion not paid at close)
= CASH_AT_CLOSE_PRETAX
− FEDERAL_TAX                 (rate depends on asset vs stock
                               sale, allocation, entity type,
                               basis — do not assume)
− STATE_TAX                   (residency-dependent)
= NET_PROCEEDS

Owner had been planning around $4.0M.
After running the bridge, the planning number moved to a
range roughly 35–45% below cash at close — a different
retirement conversation entirely.

What this cost him: about ninety minutes. What it saved: a listing at an unachievable price, six months of no offers, and a stale-listing discount when he eventually repriced.

Level-up: the Three-Buyer Panel

The section competing guides won’t have. Your business does not have one value — it has three, because the three buyer types apply different logic, and knowing which one you’re built for determines both your price and your preparation.

Value my business from three different buyer perspectives.
They should reach different numbers and disagree about why.

MY BUSINESS: [PASTE PROMPTS 1-3 OUTPUT]

PANEL

1. THE STRATEGIC BUYER — a competitor or adjacent company.
   Values synergy: my customers, my geography, my staff,
   cost overlap they can remove. Pays the most when the fit
   is real. Blind spot: overpays for strategic logic that
   doesn't survive integration, and may not close.

2. THE PRIVATE EQUITY BUYER — platform or add-on.
   Values transferable earnings, management depth, and a
   growth story that supports an exit in 5 years. Will not
   buy a job. Blind spot: structures heavily, so headline
   price and cash at close diverge.

3. THE INDIVIDUAL BUYER — an operator buying themselves a
   business, likely with SBA financing. Constrained by what
   a lender will advance and by needing a salary from day
   one. Blind spot: cannot pay above what debt service and
   their own income requirement allow, regardless of merit.

ROUND 1 — Each states: would they buy this at all, and if so
what earnings base and multiple logic would they apply? Under
150 words each.

ROUND 2 — Each names the single factor about my business that
most affects THEIR number specifically, and explains why the
other two buyers would weight it differently.

ROUND 3 — Each states what I would need to change to become
their ideal target — and what that change would cost me with
the other two.

ROUND 4 — Joint output:
   · Which buyer type is my natural buyer today, and why
   · Which would pay the most if I made specific changes
   · Which changes serve all three, and which force a choice
   · Where my current preparation is aimed at the wrong buyer

Then, in your own voice: which buyer should I be building for,
given my stated timeline of [MONTHS]?

FORMULA MODE — express all values as multiple ranges and
earnings bases. Do not compute enterprise values.

Why this matters more than it sounds: preparation aimed at the wrong buyer is wasted. Building management depth is essential for a PE buyer and largely irrelevant to a strategic acquirer who plans to fold your operations into theirs. Owners routinely spend two years optimising for a buyer who was never going to be interested.

How do I run this end to end?

1. Prompt 1 — establish the earnings base and build normalised earnings. Compute in a spreadsheet.
2. Prompt 2 — find your size band and its sourced multiple range. Check every citation.
3. Prompt 3 — score the 14 factors honestly. Have someone who works in the business score it too, separately, and compare.
4. Prompt 4 — build the range in a spreadsheet from the formula.
5. Prompt 5 — attack your own number. This is the step that saves the six months.
6. Prompt 6 — model the proceeds bridge before you plan around anything.
7. Prompt 7 and the Panel — convert findings into a sequenced plan aimed at the right buyer.
8. Take the model to a credentialed appraiser, a CPA and an M&A advisor. You will arrive with a defensible range and better questions than most sellers ever ask.

Which model should I use for each prompt?

Model-version note (August 2026): assignments validated 13 August 2026. Enable extended reasoning for Prompts 3, 5 and the panel. Formula mode degrades over long conversations — models begin computing again after roughly fifteen exchanges. Start a fresh session for Prompts 4 and 6, and re-paste the formula-mode clause if you see a number appear where an equation belongs.

Frequently asked questions

What is my business worth?

Normalised earnings multiplied by a market multiple set mainly by your size and transferability, then adjusted for specific risk factors. For owner-operated businesses that means SDE; for businesses run by a paid management team it means EBITDA. A defensible answer is always a range supported by comparable closed transactions, not a single number.

What are business valuation multiples in 2026?

By size band: businesses under $500K in SDE have traded at approximately 2.0×, unchanged in Q1 of 2023, 2024, 2025 and 2026 per IBBA Market Pulse data. The $1M–$2M SDE band sits around 3.0–3.3×. In the lower middle market, $2M–$5M EBITDA runs roughly 4.0×, and $5M–$50M reached about 6.5× in Q3 2025. Industry and quality move you within a band; size determines the band.

What is the difference between SDE and EBITDA?

SDE adds the owner’s salary, benefits and discretionary spending back to profit, because a buyer stepping into an owner-operated business receives that full economic benefit. EBITDA reflects earnings after a market-rate manager is already paid, which applies once the business runs without the owner. Applying an EBITDA multiple to an SDE figure is the most common valuation error owners make.

Why do business owners overestimate their business value?

Four mechanisms, usually in combination: anchoring on revenue rather than earnings, borrowing a multiple from companies significantly larger or venture-funded, counting add-backs a buyer’s accountant will reject, and reasoning backwards from a retirement number. The size effect does the most damage because it is invisible — a headline industry multiple rarely states the size cohort it describes.

What lowers a business valuation?

Anything that makes the earnings less likely to continue without the owner. In rough order of impact: owner dependency, customer concentration, unreviewed or inconsistent financials, undocumented processes, thin management depth, at-will customer contracts, supplier concentration, and declining margins. Most are fixable within 12–24 months, and fixing them moves the multiple, which compounds across every dollar of earnings.

Can AI value my business?

It can help you build a defensible range; it cannot produce a valuation. Language models generate confident arithmetic errors and fluently invent multiples that sound plausible. Use AI in formula mode — have it produce the equation, the named inputs and the source for each, then compute in a spreadsheet yourself. Any valuation used for a transaction, dispute or tax filing requires a credentialed appraiser.

How much of the sale price do I actually keep?

Materially less than the headline. Enterprise value is reduced by debt repaid at close, advisory and legal fees, escrow or holdback, and any earnout not payable at closing — then by federal and state tax. The tax treatment depends on asset versus stock sale, purchase price allocation, entity type, holding period and basis, so it must be modelled with a CPA rather than estimated.

How long does it take to increase a business’s valuation multiple?

Twelve to twenty-four months for most factors, because several only count once they have demonstrated durability. A documented process is worth little until it has run for a year; a second-in-command is worth little until they have operated through a difficult period. Fixes started three months before a sale often read as staging rather than improvement, which is why the preparation window matters more than the effort.

About the author

E-E-A-T note for whoever publishes this: valuation is among the highest-scrutiny YMYL topics. The four strongest signals, in order — a named human with a linkable professional profile; a valuation or transaction credential (CVA, ABV, CEPA, CM&AA) rather than a general business one; an explicit statement of how you are compensated; and a quantified track record. Add a visible “Reviewed by [credentialed appraiser or CPA]” line if you can obtain one.

Sources and further reading

• International Business Brokers Association — Market Pulse. The quarterly survey behind most size-band multiple data cited here. Read the source report rather than secondary summaries where possible.
• BizBuySell Insight Report. Quarterly closed-transaction data for Main Street businesses, including average cash-flow multiples.
• CT Acquisitions — EBITDA Multiples by Industry (2026). Lower middle market band benchmarks.
• CT Acquisitions — How EBITDA, SDE and Revenue Multiples Actually Work. Useful on the SDE/EBITDA boundary.
• Iconic — Valuation Multiples: What Drives Your Business’s Price.
• Duran Advisors — What Businesses Actually Sold For.

On the figures in this guide: all multiples cited are drawn from the sources above and were checked on 13 August 2026. They describe closed transactions rather than asking prices, which is the distinction that makes them usable. Multiples move with credit conditions and deal volume — the $5M–$50M band moving from roughly 5.0× in 2022 to 6.5× by Q3 2025 is a reminder that a figure two years old may be materially wrong. Re-verify before relying on any of them. For a valuation with legal or tax standing, engage a credentialed appraiser; published band data is a sanity check, not an opinion of value.

Last updated

Version notes

• Multiple data — under-$500K SDE at 2.0×, $1M–$2M at 3.0–3.3×, Main Street average 2.7× (BizBuySell Q2 2026), $2M–$5M EBITDA ≈ 4.0×, $5M–$50M ≈ 5.3–6.5× (IBBA Q3 2025). Verified 13 August 2026. Re-verify quarterly. Never add a multiple to this page without a named source and period.
• Tax rates — deliberately absent. The proceeds bridge names the questions that determine the rate rather than stating one, because treatment depends on asset versus stock sale, allocation, entity type, basis and residency. Do not add a rate.
• Formula mode — the guide’s core discipline. If a future edit adds a prompt, it must carry the formula-mode clause. Degradation observed after roughly fifteen exchanges in a single session; the guidance to start fresh sessions should be retested after any major model release.
• Model assignments — written and tested against [MODEL NAME AND VERSION] on [DATE]. Multi-role prompts like the Three-Buyer Panel are the most fragile across versions.
• Worked example — illustrative composite, labelled as such. Not a real transaction. If replaced with a real anonymised case, confirm confidentiality obligations first.