Dental Practice Valuation Explained: Prepare Your Figures With AI

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Dental Practice Valuation Explained: Prepare Your Figures With AI.
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Dental Practice Valuation Explained: Prepare Your Figures With AI.

A dental practice is usually valued as normalised annual earnings multiplied by a market multiple, plus or minus the value of equipment, premises and working capital. AI cannot set that multiple, but it can tidy your accounts, list add-backs and assemble the patient and income figures a broker and accountant will ask for.

The word that does the damage is "normalised". Your profit and loss account shows what the practice earned with you in the chair on your current pay and your current habits. A buyer wants to know what it would earn under a new owner who pays a market rate for clinical work. The gap between those two numbers is where most of the haggling happens, and it is also where AI saves the most preparation time.

Follow me on Instagram@sagnikteaches

How a buyer turns your accounts into a price

Valuers work from the same short chain of steps, even when the final price differs between buyers. Understanding the chain tells you which of your documents deserve attention first.

Connect on LinkedInSagnik Bhattacharya
  1. Start with reported profit. This is net profit from the last three years of accounts, before tax, interest and depreciation, which is where the abbreviation EBITDA comes from (earnings before interest, tax, depreciation and amortisation).
  2. Replace the owner's pay. If you work four days at the chair and draw a profit share, the buyer replaces your clinical work with what an associate dentist would cost. If you pay yourself less than that, normalised earnings drop. If you pay yourself more, they rise.
  3. Add back one-offs and personal costs. A refit paid in one year, a legal fee for a dispute, the family member on the payroll who does no work, the car lease. Each is an add-back, and each needs paper to support it.
  4. Apply a multiple. The multiple reflects how safe and transferable the earnings look: a larger, steadier practice with several associates usually earns a higher multiple than a single-surgery practice that depends on one dentist's list. I won't quote figures here because they move with interest rates and buyer appetite. Ask a specialist broker for current evidence from comparable sales.
  5. Adjust for assets and debt. Equipment, the lease or freehold, stock and cash, less loans and deferred liabilities. Whether these sit inside the headline price or on top of it is a negotiation point and should be explicit in the heads of terms.

Some sales are quoted instead as a percentage of annual fee income. Treat that as a shortcut, because two practices with the same turnover can have very different profits depending on lab bills, associate splits and rent.

Subscribe on YouTube@sagnikteaches

A worked example with illustrative numbers

Take a three-surgery private practice. These figures are invented to show the arithmetic and are not market evidence.

LineAs reportedNormalised
Net profit before tax, interest and depreciation$310,000$310,000
Owner's clinical pay already deducted$60,000Replace with associate rate: $150,000 for the same hours
Add-back: one-off surgery refit expensed in the year$0+$28,000
Add-back: owner's car lease and family payroll line$0+$19,000
Adjustment: rent below market (owner's relative is the landlord)$0-$14,000
Normalised earnings$310,000$253,000

The calculation behind the last row: $310,000 plus $28,000 plus $19,000 gives $357,000. Replacing the owner's $60,000 with a $150,000 associate cost takes off a further $90,000, giving $267,000, and the $14,000 rent adjustment leaves $253,000. Re-add any table like this yourself, because an AI model can get a column total wrong without saying so.

Using $253,000, a broker's evidence suggesting a multiple of, say, 4 would give a headline of about $1.01 million, and a multiple of 3.5 about $886,000. Half a point of multiple moves the price by roughly $126,500, more than four times the refit add-back, which is why the evidence behind the multiple deserves a conversation with your broker and not a chatbot.

Notice what the example shows. The owner's biggest lever was not the multiple. It was the owner's own pay, a $90,000 swing that is easy to overlook when you have never paid yourself a market wage.

The Complete Excel Course with AI Integration thumbnail

The Complete Excel Course with AI Integration

From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.

By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner

Buy now

What buyers actually look at beyond the profit line

Earnings are the start. Buyers and their advisers test whether those earnings will survive the handover.

  • Patient base. Active patients (state the window you use, such as patients seen in the last 18 or 24 months, because practices and buyers define it differently), new patients per month, how many left in the last year, and how many appointments were cancelled or missed.
  • Income mix. The split between private fees, plan or membership income and publicly funded contract work. Contract income may not transfer automatically, so buyers read the terms closely.
  • Dependency. What share of income comes from the owner's own list, and whether associates have notice periods, restrictive covenants and contracts in order.
  • Premises. Lease length, rent review dates, assignment rules. A short lease can lower value more than a mediocre year of profit.
  • Equipment and compliance. Age of chairs, compressors, X-ray equipment, service records and registration paperwork.
  • Data protection. Whether patient records are handled lawfully, because the buyer inherits them. This is where careless AI use can cost you; the tutorial on which AI tools dental staff may use under GDPR sets out the ground rules.

Five jobs AI does well while you prepare

1. Turning an accounts export into a clean three-year table

Practice accountants and software produce profit and loss reports in different layouts each year, so comparing them by eye is slow. An AI model can read three exports and return one table with consistent line names, provided you remove names first. Upload only the profit and loss lines, not the payroll or patient ledgers.

I'm preparing a dental practice for sale. Below are three years of profit and loss lines (names removed). Build one table with years as columns and consistent row names. Flag any line that changes by more than 20% between years, and list anything that looks like a one-off. Do not guess; if a line is unclear, ask me.

A realistic reply (illustrative) returns the table plus a note such as: "Lab fees rose from $41,000 to $58,000 in year three; materials fell over the same period, so this may be a reclassification." What you would fix: the AI cannot know that the lab increase was a new digital scanner workflow. You write that explanation in, then give it to your accountant.

2. Building an add-back list with the evidence column filled

Ask the model to turn your bookkeeping notes into a table with four columns: item, amount, why it is not an ongoing cost, and the document that proves it. Items without a document go into a separate "needs evidence" list. Buyers' accountants reject add-backs they cannot trace, so this list protects the price you were hoping for.

Illustration: the refit above becomes "Surgery 2 refit, $28,000, one-off capital work expensed in year two, supported by the contractor's invoice and the March bank statement." The car lease line would be marked "needs evidence" until the lease agreement is attached.

3. Summarising patient numbers without sending patient data

You do not need a language model to see patient names. Export counts from your practice software, such as new patients by month, active patients, recalls overdue and cancellations, and paste only the totals. Ask for trends and questions a buyer might raise.

Here are monthly counts for 36 months: new patients, patients seen, cancellations, no-shows. No names or identifiers are included. Describe the trends in plain English, show the three weakest months and suggest what a buyer might ask about each.

If the answer says new patients dropped in months 14 to 17, you can check whether the cause was a dentist leaving, a closed surgery or a change in reporting, and write a one-paragraph explanation before anyone asks. Lapsed patients who could be recalled are a separate opportunity, covered in winning back lapsed dental patients.

4. Drafting the information memorandum and data-room index

Brokers usually write the sales document, but you can speed them up by supplying a clear fact sheet. An AI tool can turn your notes into a one-page description of the practice: surgeries, opening hours, team, equipment, services, lease, and growth plans a buyer could pursue. Then ask it to produce a document index (accounts, tax returns, lease, associate contracts, equipment list, registrations, policies) with a status column, so you can see what is missing.

Check every sentence of the output against your records. A model will happily describe a "modern, well-equipped practice", which tells a buyer nothing and may be untrue. Replace adjectives with facts: "three surgeries, intraoral scanner installed 2024".

5. Rehearsing the buyer's questions

Give the model your normalised table and ask it to act as a sceptical buyer's accountant. It will list the questions: why did lab costs rise, how much income depends on you, when does the lease end, what happens if the two associates leave? You do not need its answers. You need yours, written down and supported before a real buyer asks.

A filled-in readiness checklist

Here is a checklist for a practice twelve months from a possible sale. The statuses are an illustration of how an owner might fill it in.

ItemStatusNext action
Three years of accounts in one comparable tableDone via AI draft, accountant to confirmSend to accountant with notes
Add-back list with evidenceTwo of five items missing paperLocate lease and contractor invoice
Associate contracts and notice periodsOne contract expiredRenew before marketing
Lease term and rent reviewFour years remainingAsk landlord about extension
Active patient count, definition written downCounted two ways, figures differChoose one definition and state it
Data-protection recordsNo list of AI tools in useWrite the approved-tools list

Where AI helps the least, and where it misleads

Three errors show up repeatedly when owners try an AI-only valuation.

  • Invented multiples. A chatbot asked "what is my practice worth?" will produce a confident range. It has no access to current transactions, and the number will sound more authoritative than it is. If you see a figure, ask where it came from. It will not have a source.
  • Mixing definitions. Turnover, profit, EBITDA and "seller's earnings" are different numbers. A model may swap them mid-answer. Ask it to restate the definition each time it gives a figure, or ask for the working in a table you can check.
  • Silent arithmetic slips. Re-add the columns yourself, or paste the output into a spreadsheet. The tutorial on cleaning messy data shows a fast way to check totals.

There is also a tax and legal layer: whether the sale is of shares or assets, how goodwill is treated, what warranties you give. That belongs to your accountant and solicitor, and AI should not be used to pre-empt their advice.

Keeping the data safe while you prepare

Financial data and staff pay are confidential, and some figures will come with patient-linked detail if you are careless. A few working rules keep preparation clean.

  • Use a business-grade plan where your content is not used for training by default, not a free consumer account. OpenAI's business data page says ChatGPT Business content is not used for training by default, and Anthropic says the same of its commercial plans such as Claude Team. On personal Free and Plus ChatGPT accounts, by contrast, data sharing for training is on until you switch off "Improve the model for everyone". Check the setting on the vendor's own page on the day you upload anything.
  • Remove names from payroll lines (use "Associate A", "Hygienist B") and leave out patient identifiers entirely.
  • If a prospective buyer has signed a non-disclosure agreement, read what it permits before putting their questions or documents into an AI tool.
  • Keep a short log of what was shared with which tool and when. If a buyer's adviser asks, you can answer.

The same habits apply in other professional businesses; anonymising client data before you paste it into AI walks through the method.

Who does what in a sale

A sensible division looks like this. You and your practice manager gather facts and explain unusual items. AI prepares drafts: tables, lists, summaries, question rehearsals. Your accountant confirms normalised earnings and advises on tax. A specialist broker supplies market evidence, finds buyers and manages the process. A solicitor handles the contract, lease assignment and warranties. Nobody on that list is replaced by a chatbot, and the preparation you do beforehand makes each of them faster and cheaper to use.

If you would rather map which of your reports are safe to automate this way, that is the kind of job my AI implementation consultation covers.

Three practice types, three different valuation problems

The same method produces different worries depending on how the practice is built. Each of these sketches is illustrative.

The single-surgery practice. One dentist, one hygienist, a loyal list. Earnings are mostly the owner's own work, so normalising pay removes a large share of the profit. The useful AI job here is a table of how much income depends on the owner versus the hygienist and the recall system. If hygiene appointments are 20% of fee income and recalls are tracked, the buyer sees something that continues after you leave. If the practice has no recall data, that is the first thing to fix.

The associate-heavy practice. Four surgeries, five associates on percentage splits. The risk is attrition: if two associates leave after completion, income falls with them. Ask AI to draft a one-page retention summary from the contracts (notice periods, covenants, split percentages, start dates) after you have removed names. Then have a solicitor confirm it. The associate dentist contract checklist lists the clauses a buyer's solicitor reads first. Buyers pay for contracts that look secure, so sorting an expired contract is worth more than polishing the brochure.

The membership-plan practice. A third of income arrives as monthly plan payments. Buyers like the predictability but will ask about churn, so the table to prepare shows members at the start and end of each month, joiners, leavers and average monthly fee. A quick AI pass over twelve months of counts will flag a month where leavers spiked, which prompts you to find out why before a buyer does.

A mistake that happens: the chatbot number that stuck

Picture an owner who asks a chatbot what a practice with $1.4 million of fee income is worth, gets a range that sounds sensible, and repeats it to a family member and a prospective associate. The broker's figure comes back lower because the practice's rent is below market and the owner's pay is low. The owner now has to walk back a number they gave people and has lost goodwill in the conversation. The fix is simple: use AI for preparation, keep every price talk until the accountant has normalised earnings, and describe any early figure as "a range I'm testing".

Rehearsing a buyer's questions with a sample exchange

Here is how the rehearsal in step five can look in practice, shortened and illustrative.

Act as a sceptical accountant acting for a buyer. Using this normalised table (names removed), list the 8 questions you would ask first, ordered by how much they could reduce the price. For each, say which document would answer it.

A typical reply puts the lease expiry first ("Four years remaining: what is the assignment clause, and will the landlord extend?"), the owner's reduced clinical hours second, and the lab cost jump third. What you would fix: the model tends to stress generic questions such as staff turnover even when you have none. Delete those, add any question you know your own practice invites, and write a two-line answer plus the supporting document for each. This sheet is the one you hand to your broker.

A twelve-month preparation timeline

  • Months 12 to 9: get the three-year table, define active patients, list add-backs, and note which contracts and the lease need work.
  • Months 9 to 6: fix what you found: renew associate contracts, document processes, remove personal costs from the practice books where your accountant agrees.
  • Months 6 to 3: run the practice normally and watch the numbers monthly, so a buyer sees steady figures. A simple AI-built monthly summary of new patients, cancellations and income makes this cheap.
  • Months 3 to 0: assemble the data room, rehearse questions, then hand over to your broker and accountant.

Questions owners ask before a valuation

Can ChatGPT value my dental practice?

It can explain the method and build a worksheet, but it cannot know what buyers are paying for practices like yours this year. Treat any number it gives as a rough illustration. The figure that matters comes from a broker's market evidence and an accountant's normalised earnings.

Is it safe to upload my practice accounts to an AI tool?

Accounts name the practice, staff pay and sometimes patient-linked data. Use a business plan with training switched off, strip names and patient details first, and avoid free consumer plans. If a buyer has signed a confidentiality agreement, check it allows you to share material with a processor.

Should I value the practice before I decide to sell?

A rough valuation a year or two ahead is useful, because it shows which numbers a buyer would mark down while you still have time to fix them. An informal accountant review costs little compared with discovering a problem during due diligence.

What do brokers charge for a valuation?

Practices differ, and fee structures vary between a flat fee and a percentage of the sale price. I haven't verified current rates, so ask two or three brokers for written terms and compare what each includes.

Further reads

Sources: general valuation practice for owner-run healthcare businesses; OpenAI business data page and ChatGPT data controls FAQ; Anthropic privacy centre on commercial products (checked November 2026). No market multiples are quoted because they change with conditions; ask a specialist broker for current ranges.

Want your practice figures sale-ready?

On a 1:1 call we can map which of your reports an AI tool can tidy safely, which still need your accountant, and how to set that up so it keeps running.

Book a 1:1 call with me