How we calculate value

How are trade businesses valued? Buyers take a measure of yearly earnings and multiply it by a number of turns that reflects how risky and transferable those earnings are. Our calculator does the same thing in the open: it works out your earnings two ways, picks a multiple band for your trade and size, moves it up or down for what you told us, and shows a low, likely and high value. This page shows every step, the source behind every number we could source, and the parts that are our own estimate.

Everything here describes engine version 2.0.0 (CALC_VERSION 2.0.0), running engine config 2.0.0, released on November 5, 2026. Every multiple below is illustrative: no trade's numbers have been checked against verified closed-deal data yet.

The formula in one look

SDE              = net profit + owner pay + interest and depreciation
                   + accepted add-backs + rent normalization - one-off income
Adjusted EBITDA  = SDE - market-rate replacement manager salary
Value            = earnings x multiple

Two things decide your number: which earnings figure a buyer would use, and which multiple they would put on it. The rest of this page is how we choose each one.

What we ask you

The quiz asks every owner the same base questions, then a short set for your trade and a few follow-ups that depend on your answers. A typical flow stays under 25 questions and takes about 5 minutes. The money questions are:

• Revenue for the last three full years.
• Net profit before tax for the last full year (it can be negative).
• Total pay for one working owner: salary, draws or distributions, bonuses and the payroll tax on them.
• Loan interest plus depreciation.
• Add-backs: personal or one-time costs run through the business, either itemized or as one rough total.
• Whether you pay yourself rent on a building you own, and what market rent would be.

The rest are about how the business runs: who holds the license, how much the company depends on you, recurring revenue, customer mix, crews and technicians, books, and your state and metro. We never ask for your SSN, EIN, tax returns or bank logins. Every number is self-reported, and we treat it that way.

Two bases: SDE and adjusted EBITDA

SDE (seller's discretionary earnings) is what one full-time owner-operator takes home before paying themselves and before debt payments. It adds back one owner's full pay. If two owners work in the business, only one owner's pay is added back. This matches how BizBuySell reports the SDE multiples our SDE bands are built on.

Adjusted EBITDA is what the business earns for an owner who does not work in it. We start from SDE and subtract what it would cost to hire a general manager to do your job. Private equity groups and larger buyers price on this number.

The threshold between them

We value on SDE when weighted adjusted EBITDA is under $500,000, on adjusted EBITDA at $1,000,000 or more, and on a straight-line blend in between:

EBITDA weight = (adjusted EBITDA - 500,000) / 500,000, kept between 0 and 1
Value = EBITDA weight x EBITDA-path value + (1 - EBITDA weight) x SDE-path value

• $1M is the full EBITDA point because private equity add-on buyers generally need about $1M or more of EBITDA. Below that, most buyers are individuals with SBA loans who price on SDE.
• $500K is where the blend starts because that is where advisers start quoting EBITDA bands, for example 4.0x to 6.5x for $500K to $2M of EBITDA for HVAC. That starting point is our estimate.
• Why a blend and not a switch. A hard switch creates a cliff. In a made-up example, a roofer with $900K SDE is worth about 2.8 x $900K = $2.52M on SDE and about 4.0 x ($900K minus $140K) = $3.04M on EBITDA. A $1 change in profit should not move the answer by half a million dollars, so the blend keeps value smooth.
• If the business cannot run without you, the EBITDA weight is capped at 0.5, however large the profit. No institutional buyer underwrites a company that leaves with its owner without tying the price to an earnout.

Add-backs: what we accept

Buyers do not take every add-back at face value. A quality-of-earnings review tests each one. So each case uses a different acceptance rate. These rates are our estimates.

Itemized and rough add-backs together are capped at 15% of last year's revenue (our estimate). Above that, double counting is more likely than real savings, and the excess is left out. Roofing storm one-off costs are only added back if the storm revenue is also normalized, because buyers will not add back the costs and let you keep the windfall.

The replacement manager salary

On the EBITDA path we subtract the cost of a general manager, fully loaded:

Manager cost = salary for your revenue band x state wage index x 1.20

The anchor is the Bureau of Labor Statistics national median wage of $101,280 for general and operations managers (May 2023). The same table reports a mean of $129,330, which fits larger firms paying more. The two upper bands are our estimate. The 1.20 burden covers payroll taxes and benefits (our estimate, range 1.15 to 1.25). The wage index is sourced for California (1.11) and New Jersey (1.09) from the BEA's 2024 regional price parities; for other states it is our estimate until the full BEA table is loaded. The SDE path never subtracts this salary.

Weighting three years

We only ask for last year's profit, so we estimate earlier years by applying last year's margin to each earlier year's revenue. Then we weight the three years by the shape of your revenue. The weights are our estimates.

The reason: buyers pay for an average year when revenue swings, for today's run rate when it is falling, and give most weight to the latest year when growth is steady. If you skip earlier years, we use last year only and your confidence score drops 10 points.

Storm normalization (roofing only)

If 30% or more of your roofing revenue comes from storm or insurance work and you tell us your quietest year in the last five, we use the average of that quiet year and your weighted revenue. Buyers re-base storm earnings: one adviser gives the example of a $4M hail-year roofer that is "really a $1.8M business". Using the midpoint of the two is our estimate of how buyers split the difference. If you skip the quiet-year question, we keep your weighted revenue, keep the full storm adjustment and take 10 points off confidence.

Base multiple bands per trade

Each trade has six bands: three on SDE and three on adjusted EBITDA. Low, mid and high stand for a lower-quartile, typical and upper-quartile buyer for a business with average answers on every driver. Each band applies exactly at its anchor point, and between anchors we interpolate in a straight line, so earning $1 more never drops your value.

HVAC

Illustrative estimate

The HVAC S2 band is rounded from BizBuySell's 2021 to 2025 sold HVAC deals: lower quartile 1.99x, median 2.58x, upper quartile 3.33x SDE. The FY2025 average was 2.80x on 123 sales. S1 sits below that, in line with the IBBA's 2.3x SDE median for deals under $500K across all industries. The EBITDA bands draw on Next Era Legal's adviser ranges, First Page Sage, Axial letters of intent at 2.1x to 10.9x, GF Data's 7.2x for $10M to $500M deals via CIBC and PKF's "north of 10x" for top service platforms. We do not use Capstone's 14.0x disclosed-deal average, because those are platform-sized deals.

Plumbing

Illustrative estimate

The plumbing S2 band is rounded from BizBuySell's 2021 to 2025 sold plumbing deals: lower quartile 1.66x, median 2.24x, upper quartile 3.15x SDE. S3 keeps the same gap below HVAC that the sold medians show. The EBITDA bands lean on two disclosed deals by a public buyer, Kingsway, for Southside Plumbing and M.L.C. Plumbing, both with seller notes or earnouts, plus IBBA survey medians, GF Data via CIBC and Ad Astra's illustrative ranges.

Roofing

Illustrative estimate

There is no published sold-deal multiple for small roofing companies. The roofing S2 band is built from four things: BizBuySell roofing listings, with an asking median of 3.16x SDE and quartiles of 2.24x to 4.01x; BizBuySell's HVAC data, where sale prices run about 0.92 to 0.94 of asking, which would put a sold roofing median near 2.9x to 3.0x; BizBuySell's sold construction average of 2.60x; and Sofer Advisors' 2.5x to 3.5x for small owner-operators. We set the mid at 2.5x, below the implied 2.9x, because asking prices lean toward the businesses owners think are worth most, and roofing carries storm and crew risk. S1 draws on Chisel's 1.9x to 2.7x for small or storm-dependent shops. The EBITDA bands draw on Axial roofing letters of intent at 2.8x to 7.0x, Sofer's 4x to 6x for managed firms, CriticalPoint's 4x to 8x and Portus, set below Portus because its range covers larger firms.

The clamp on the final multiple

After adjustments, the multiple is kept between a floor and a ceiling: the floor is the higher of 1.0x SDE (or 2.0x EBITDA) and 75% of the band's low; the ceiling is 115% of the band's high. These limits are our estimates. They stop an unusual mix of answers from producing a multiple no buyer would pay.

Adjustments and caps

Each answer you give can move the mid multiple up or down by a set number of turns (a turn is 1.0x of earnings). Every adjustment size is our estimate, because no public source isolates the effect of one driver on a private trade-business price. On the SDE path most adjustments are halved, because SDE multiples are smaller. Two work differently on SDE: owner dependence (heavy owner involvement is normal for SDE-path businesses) and the quality of your books (SBA lenders check numbers against tax filings, so books matter more there).

Adjustments are grouped, and positives and negatives are capped separately, first per group and then in total. If a cap binds, every adjustment of that sign is scaled down evenly, so the dollar impacts in your report still add up.

Why the upside is smaller than the downside: with every positive answer, an HVAC company in the S2 band lands at about 2.6x plus 0.75, or 3.35x, which is close to BizBuySell's upper quartile of 3.33x. On the downside, weak companies often do not sell at all rather than sell cheap. Separate caps for positives and negatives mean a strong company's strengths cannot hide a weakness that a buyer would still price.

We also stop the same risk being counted twice. For example: size is priced only through the band and never as an adjustment; a license penalty is scaled by how strict your state's license rules are; a declining trend is penalized less because the earnings already use last year only; and storm penalties are halved when storm normalization has already been applied.

Geography

Your state and metro move the multiple a little, never a lot:

geography adjustment = state buyer density + metro adjustment,
                       capped at plus or minus 0.5 turns (0.25 on SDE)

State adjustments reflect how many active buyers we can document, for example Florida and Texas leading HVAC consolidation in 2024. No published source breaks trade multiples out by state, so every state and metro adjustment is our estimate. Your state also sets how strict the license rules are, which scales the license penalty, and the wage index for the manager salary.

Low, likely and high

For each path we find your band at your earnings, add the capped adjustments to get the likely multiple, then step down to low and up to high by the band's own spread, widened or narrowed by your confidence score. Low, likely and high earnings differ only in how many add-backs are accepted. Then we blend the two paths and round: values under $2M to the nearest $10,000, and $2M and over to the nearest $50,000, so the result never looks more precise than it is. You always get a range, never one number.

Confidence

Confidence starts at 100 and drops for things that make the estimate less certain:

A score of 80 to 100 is high and gives the tightest range, 60 to 79 is medium, and under 60 is low with the widest range. The point sizes are our estimates.

Readiness score

Readiness is a separate score from 0 to 100 that shows how ready a buyer would find your company. It never changes the value.

Each part scores how close your answers are to the best answer. Some answers cap the total: a roofing company with a current insurance claims problem tops out at 35, a business with zero or negative earnings at 40, an owner who runs everything with no second in command at 50, no books at 55, and a sole license holder with no plan in a strict-license state at 60. The bands read: 80 and up market ready, 60 to 79 getting ready (about 12 to 24 months of work), 40 to 59 needs work (about 2 to 3 years), under 40 not ready yet. The weights and time frames are our estimates.

Value gap

The value gap shows what a buyer might pay if you fixed your five biggest fixable drags. We re-run the engine with one answer moved one realistic step up at a time (for example books kept yourself to a bookkeeper), holding your earnings and basis fixed. We keep the five with the largest gain, then re-run with all five together, so the total is not just the parts added up. Geography, years in business, trend, margin, seasonality and customer type are not counted as fixes. We show it as "a buyer might value the business about $X to $Y higher, all else equal" and never promise the gain.

What we could not find

We say plainly where the evidence runs out:

• No sold-deal multiples for small roofing companies. BizBuySell publishes sold data for HVAC and plumbing but only asking prices for roofing. Every roofing band rests on asking prices, letters of intent and adviser pages.
• Thin EBITDA evidence for all three trades. Only plumbing has hard closed-deal points (Kingsway, two to three deals), and they include earnouts.
• No public measure of what one driver is worth. Every adjustment size is our estimate. We plan to calibrate them once we have enough closed-deal outcomes, shared with consent.
• No state-by-state multiples. State adjustments are inferred from deal counts, not prices.
• Owner age by trade. Only construction-sector figures exist.
• A Census split of plumbing versus HVAC firms. Census counts them together.
• How many trade businesses actually sold in the last 10 years, or how many listed ones never sold.
• Full Capstone residential services multiples. The report text could not be extracted.
• Published referral or lead fees paid by buyers, wealth managers or lenders for owner introductions.

The illustrative label rule

Until a trade's multiples have been checked against verified closed-deal data, every page, result, report, PDF and email that shows a multiple carries an "Illustrative estimate" label, and the disclaimer adds that the multiples are illustrative assumptions, not verified market data. Today that applies to all three trades. When a trade is verified, we will record the date in the change log below.

What we do not do

• We do not appraise businesses. This is an educational estimate, not a formal valuation.
• We do not let AI produce or change a number. The value comes only from the engine described here.
• We do not change a stored result. Each estimate keeps the engine and config version it was made with, so an old report stays the same after we update the formula.
• We do not present our multiples as market facts. See where our numbers come from for every source and how much we trust it.
• We do not earn more if your number is higher or if you sell. See how we make money.

To see the method on your own numbers, start a free estimate for roofing, plumbing or HVAC.

Change log

• November 5, 2026: engine 2.0.0, config 2.0.0. Replaced the placeholder multiples with per-trade bands built from published data (section above). Added the SDE and adjusted EBITDA bases with the $500K to $1M blend, add-back acceptance rates, the replacement manager salary, three-year weighting, roofing storm normalization, grouped adjustments with separate caps, geography, confidence, the readiness score and the value gap. Plumbing and HVAC estimates launched.
• Engine 1 (until November 5, 2026). Roofing only. Used the same placeholder multiples of 1.5x, 2.5x and 3.5x SDE for every trade, set on purpose as placeholders and not market data, with a simpler set of factor adjustments. Results made with engine 1 keep their stored numbers.

Corrections log

No corrections yet. If you think a number, a source or a sentence on this page or any other page is wrong, tell us through contact. Include the page and what you think is wrong, with a source if you have one. We will check it, fix it if we got it wrong, and list the correction here with its date.