Most private companies cannot say how many companies could buy from them. Nobody owns the question, sales defines the market as its pipeline, and the honest number threatens someone's story. So the count never gets made.
That gap is the reason marketing ends up measured on closed revenue. Awareness has a denominator. Without the count there is no awareness metric, and the company fills the hole with attribution.
I expect the number to be missing. The plan below is how I produce it in the first quarter and what I need agreed before I start.
Define the buying unit. Not the industry, the thing that needs the product. A machine, an instrument, a process step. Units times attach rate times spend per unit is the equation, and every later step puts a number on one of those three terms.
Interview the owner, sales, product, and finance separately. Each gets the same question. How many companies could buy this, and how do you know? The answers set the baseline I am correcting, and the spread between them is the first finding.
Pull the customer base from the CRM and the ledger. Segment by size and application. Take median annual spend per segment, not mean.
Build the long list. Firmographic filters first, then the signals only a real buyer throws off. Engineering job postings, patents citing the technology, trade show exhibitor lists, competitor case study customers, distributor catalogs.
Name the proxy markets. Every niche has an upstream or downstream neighbor somebody counts. Tool shipments, source unit volumes, installed base reports.
Draw a random sample of 150 to 200 accounts from the long list and qualify each one by hand. Do they buy the unit, and about how often. The qualified fraction times the list is the count, and the sample size is the error band. This is the step everyone skips, and I will run it with AI agents doing the first pass and a person confirming the sample, so it takes days instead of a quarter.
Call ten customers across the segments and ask what share of their spend in the category we get. That share-of-wallet answer converts our revenue in an account into the account's total demand.
Sum the competitors. Segment disclosures for public ones, headcount times revenue per head for private ones, rep and distributor intelligence for the rest. Competitors plus us is the floor the total cannot sit under.
Three estimates. Bottom-up from the qualified list, the competitor floor, and the proxy market times attach rate. If they land within a factor of two of each other, the range holds. If they do not, the gap identifies which assumption is wrong, and that becomes the next piece of work.
Publish a range, never a point. A point invites argument about the number. A range invites argument about the assumption, which is the useful argument.
Present the regime finding and the metric to leadership. From here marketing reports penetration of the universe every quarter. Every closed-lost reason of no need removes a name. Every new logo that was not on the list widens the filters. The number tightens each year because the CRM feeds it.
The list will show some of the believed market is not worth reaching, and the total may come in under the number in the owner's head. I will say so. Agreeing to that in advance is what makes the work worth doing.
Few companies can count the companies that could buy from them. I will deliver that count in my first quarter, with the error band and the assumptions behind it. From then on marketing reports how much of that market it has reached, and we stop arguing about which closed deal marketing gets credit for.
Back to dougjones.me
Doug Jones