How to Find a Competitor's Average Deal Size: 10 Sources

Search this question and every result explains how to calculate your own average deal size. That is a genuine gap rather than an oversight: your own figure is a division you can do in a minute, and a competitor’s is not published anywhere as a number. What is published, and what almost nobody uses, is better than an average.

The approach that works is to bound the answer from several directions rather than to hunt for a single figure. A pricing floor sets the bottom. A disclosed customer distribution shapes the middle. Public contract records give you specific real deals. Your own contested deals tell you what they quote in your segment. Report the range and the sources, and expect the spread between them to be the most informative thing you produce.

1. 1Decide which number the decision actually needs. The deal you will face against them in your segment, the typical customer they serve, and the revenue concentration of their business are three different questions with three different answers. Pick one before collecting anything.
2. 2Fix the definition before you compare anything. First-year value or total contract value, one year or three, list or net of discount, with services or without. Two figures built on different definitions are not comparable, and nobody discovers this until the two are already in the same slide.
3. 3Take the disclosed distribution where one exists. Listed companies routinely report how many customers sit above stated revenue thresholds and what share of revenue those accounts represent. That is a distribution, which is strictly better than any average, and it costs nothing to read.
4. 4Compute the crude average, then label it crude. Revenue divided by customer count gives a figure in seconds. Keep it, because it bounds the problem, and write beside it that it is a mean over a skewed population, which in software usually means it describes almost nobody.
5. 5Pull real contract values from public buyers. If the competitor sells to government, education or healthcare, the awarded value and the term of specific contracts are published by law in several jurisdictions. These are complete deals with a named buyer, which is as close to ground truth as outside research gets.
6. 6Bound it from below with their commercial floor. Entry price times any seat or volume minimum is the smallest deal they will take without an exception. Combined with the top of the distribution, you now have a range rather than a single misleading number.
7. 7Anchor it with the deals you have contested. Record what the buyer told you the competitor quoted, in every deal you win or lose against them, with the seat count and the term. Twenty of those beat any published figure, because they are drawn from exactly the population you sell into.

Take a real, checkable example rather than an assertion. In its annual report for the year ended 31 January 2026, the automation company UiPath disclosed revenue of $1,610.6 million, annual recurring revenue of $1,852.6 million, and approximately 10,747 customers. It also disclosed that 2,565 customers carried annual recurring revenue of $100 thousand or more and 357 carried $1 million or more, and that those two groups accounted for approximately 89% and 52% of revenue respectively.

Divide revenue by customers and you get an average of roughly $150,000. Now use the rest of the disclosure. The 357 largest accounts, 3% of the base, take about half the revenue. Everyone outside the $100 thousand band, which is three quarters of all customers, shares the remaining 11%.

The customer counts, the revenue shares and the totals are as disclosed. The final column is computed from them, so anyone can check it. The point is the gap between the first row and the last: the headline average of about $150,000 sits close to seven times above the roughly $21,700 that three quarters of their customers represent. Walk into a deal quoting the average as what this vendor charges and you will be wrong in the direction that loses the argument.

There is a vocabulary consequence too, and it is measurable. The exact phrase annual contract value turns up in more than 400 annual reports filed with the US securities regulator; average deal size turns up in roughly 150. Companies disclose aggregate and per-customer measures and almost never an average deal. So a search for the words you started with returns nothing and you conclude the data does not exist, while the better version of it sits three pages earlier in the same document under a name you were not looking for.

For any competitor selling to government, education, healthcare or other public bodies, complete deals are a matter of public record. Not a unit price, which is what a marketplace listing gives you, but the awarded value of a whole contract with the buyer named and the term attached. That is precisely the shape of a deal, and it is the closest thing to ground truth available from outside the company.

Two practical notes. First, the figure reported in an award record is usually the amount committed rather than the ceiling of the contract including every option, so read which one you have before quoting it. Second, public buyers negotiate hard and buy through frameworks, so these values sit towards the lower end of what the vendor charges rather than in the middle. Use them as real points inside the distribution, not as the average, and read the per-unit half of the same documents under competitor pricing.

1. Their pricing page, seat minimums and plan floors

The bottom of the distribution, and the only part of it that is unambiguous. Take the entry price of the lowest plan a business would realistically buy, multiply by any stated seat or volume minimum, and note whether an annual commitment is required. That product is the smallest deal they will take without an exception. Where an enterprise tier says contact sales, the floor of that tier is often revealed by the feature that forces the upgrade, such as single sign-on or an audit log.

2. Investor filings and customer cohort disclosures

The single best source when the competitor is listed. Look for three things in the annual report and the quarterly filings: total customer count, the number of customers above one or more stated revenue thresholds, and the share of revenue those cohorts represent. Then read the definition of a customer, which is always stated and rarely the same twice. Track the cohort counts across two or three years and the direction of the business becomes obvious well before anybody announces a strategy.

3. Public sector contract award records

Search by supplier name rather than by product, and search the parent entity as well, since public bodies contract with legal entities and often with a reseller rather than the vendor. Record the awarded value, the start and end dates, and any option years, because a five-year award is not five times an annual deal in any meaningful sense. Ten awards give you a real distribution for one segment of their business, which is more than any vendor-supplied benchmark will ever give you.

4. Public records requests to public sector buyers

Where the award notice is thin, the buying authority holds the contract and freedom of information law usually obliges it to release a version. This is legitimate, routine and used constantly by journalists and analysts. Two things to expect. Commercial terms may be redacted as confidential commercial information, and in some regimes the supplier is notified and given an opportunity to object before anything is released, so assume the competitor may learn a request was made.

5. Earnings calls and investor presentations

Management commentary is where the direction lives even when the number does not. Search the transcript for deal size, land, expand, upmarket and enterprise, and read what is said about seven-figure deals in particular, since companies count and celebrate those. A statement that average deal sizes increased is a segment move being described in public. Pair it with the cohort counts from the filing, which either support the claim or quietly contradict it.

6. Review sites and buyer community threads

Individually unreliable and collectively useful. Reviewers occasionally state what their organisation pays, and review platforms record the reviewer’s company size, which turns a bare figure into a data point with a segment attached. Community threads asking whether a product is worth the money often produce several such figures in one place. Treat every one as unverified, record the company size beside it, and never build a claim on fewer than several consistent statements.

7. Their job adverts and disclosed pay ranges

An indirect route that pay-transparency rules have made much more usable. Several US states require salary ranges in job adverts, and the European directive on pay transparency required member states to legislate by 7 June 2026, obliging employers to give applicants pay information before interview, although only four member states met that deadline so coverage is uneven for now. A seller’s on-target earnings imply a quota, and a quota divided by a plausible number of deals per year gives an order of magnitude for deal size. It is a chain of assumptions, so use it to sanity-check a figure rather than to produce one.

8. Industry and analyst reports

Useful for orienting yourself in a category you do not know, and almost never usable as a statement about a specific competitor. Category-level average selling prices are built from surveys with undisclosed samples, mixed definitions and long lags. Several of the large market-data services are available free to library members, which is worth knowing before anybody buys a subscription. Whatever you take from them, record it as a category figure rather than as a competitor figure.

9. Your own closed-lost and competitive deal records

The most accurate source available to you and the one most often left uncollected. Add three fields to any deal where a named competitor was present: what the buyer said the competitor quoted, the seat or volume count, and the term. After twenty such deals you have a distribution drawn from exactly the population you sell into, which no external source can produce, and it doubles as the input to your win rate analysis by competitor.

10. Win/loss interviews with buyers who evaluated both

The only source that explains the number rather than reporting it. A buyer who chose the competitor will usually describe the shape of the offer: the term they agreed, whether pricing ramped, what was bundled to close the gap, and which line item the discount actually came off. That is worth more than the headline value, because it tells you what they will do in the next deal. Our free win/loss interview questions include the commercial section this feeds.

Every method here has a different error profile, and stating that profile beside the number is what separates research from guessing. Work down the ladder until you have two or three that agree, then report the range rather than picking a favourite.

One rule holds all of this together: never mix methods inside a single figure. A range built from a disclosed floor and a disclosed cohort is defensible and can be shown to an executive. An average of a filing-derived mean and three review comments is a number with no meaning that will nonetheless be quoted for a year. The size of their business as a whole is a different question and belongs with competitor revenue.

1. 1Name the definition on the figure itself. First-year value, annual contract value or total contract value, new business or all business. A number without that label cannot be compared with anything and will eventually be compared with everything.
2. 2Check the population, not just the maths. A mean across an installed base including free-to-paid conversions is a different animal from a mean across enterprise deals. Read how the company counts customers before dividing by them.
3. 3Convert currency and term at a stated date. Contract values in public records are in local currency and often span several years. Normalise to one currency and one year, and record the rate you used.
4. 4Test it against the floor. If your estimate is below the smallest contract their pricing permits, the estimate is wrong. This one-line check catches more errors than any other on this list.
5. 5Ask two sellers whether it matches reality. People who have lost deals to this competitor in the last quarter know the shape of their commercial offer. If your number surprises them, find out why before it reaches a slide.

Everything on this page is either published by law or volunteered by a buyer, and that combination keeps the work straightforward. The constraints that matter here are about somebody else’s obligations rather than your own, which is an unusual shape and the reason it catches people out.

• The confidentiality clause usually binds the buyer. Enterprise quotes and contracts routinely restrict what the customer may disclose. Listening when a buyer volunteers a figure is one thing; pressing them to forward the written proposal asks them to break their own agreement, and the relationship you damage is your own.
• A records request is not anonymous. Freedom of information requests are made by a named requester, and several regimes require the authority to notify the supplier before releasing information it has marked as confidential commercial information. Assume the competitor may learn that you asked, and make the request one you would be comfortable explaining.
• Never discuss commercial terms with the competitor. Researching what a rival charges is ordinary competitive work. Talking to them about what either of you charges, intends to charge, or will not go below is a different activity that competition authorities pursue vigorously, and it does not require an agreement to be a problem.
• Do not seek a quote under a false identity. Requesting pricing while presenting as a genuine prospect is misrepresentation, and it produces a figure you cannot use in any document that names its source. The published contract values on this page are better evidence and carry no such problem.

• A private company’s actual average. It exists in one spreadsheet inside their revenue operations team and nowhere else. Proxy: a range bounded by their pricing floor and their largest publicised deployments.
• The median, which is what you really want. Nobody discloses it, because the mean flatters and the median does not. Proxy: the residual band computed from disclosed cohorts, as in the worked example above.
• Their discount floor. How far they will go to win a specific deal is decided in the room and recorded nowhere public. Proxy: the lowest verified public sector award, which usually sits near the bottom of what they will accept.
• The split between new business and expansion. Very few companies break this out, and it changes the interpretation of every figure above it. Proxy: retention disclosures, which are covered under competitor churn.
• What they charged a specific named account. Unless that account is a public body, this is confidential on both sides. Proxy: what the buyer tells you directly in a win/loss conversation, which is often more than you expect.

Run the derived figures quarterly for listed competitors, because that is the cycle their disclosures arrive on and re-reading one filing takes fifteen minutes. Re-check the floor whenever their pricing page changes, since that is the fastest-moving input and the one that moves the whole range. Add to the deal-level record continuously rather than in batches, because it only becomes useful through accumulation and a backfill six months later is guesswork.

Three changes justify rebuilding the estimate immediately rather than waiting. A change to packaging or to the minimum commitment, which resets the floor. The appearance of a new top tier, which usually means they are chasing larger deals than the ones you have been modelling. And a first public sector contract, which both adds a real data point and tells you a new segment has opened. Note the date of every rebuild, because a stale deal-size figure is repeated in negotiations long after it stops being true.

The inputs here move on different clocks and none of them announce themselves. A packaging change quietly raises the floor of every future deal. A new enterprise tier appears one Tuesday with no post about it. A cohort disclosure shifts by two hundred accounts inside a document nobody on the commercial team reads. Each is small on its own, and together they are the difference between a seller expecting a forty thousand deal and meeting a hundred and twenty thousand one. That gap is discovered in the negotiation, which is the most expensive possible place to discover it.

Watching pricing pages, packaging, tiers and filings across a competitive set is exactly the repetitive checking that competitive intelligence platforms exist to carry. Flares watches the commercial surfaces where those edits land and reports what moved, so a reset floor reaches your sellers before a customer does. What no platform can produce is the number itself. No software knows what a private competitor charged in a room you were not in, and the closest anyone gets is asking the buyers who were, which remains a human conversation.