Size the Market Before You Source It
Outcome: three counted numbers — total, addressable and workable — plus a credit estimate for the full sourcing run, all produced before you import a single row.
- Surface
- App and MCP server
- Level
- Intermediate
- Uses
- find_companies (count only) · check_credits
- Credits
- ~2
- Prerequisite
- Lesson 01's filterable ICP
Counting is nearly free; importing is not
A search returns a count before it returns rows. That count is the cheapest information in this course and most people skip straight past it.
Read the count first, adjust the filters, read it again. Only import once the number is one you meant.
The three numbers
Total market — the outer boundary
Every company matching your broadest defensible filters: industry and geography, nothing else. Big, mostly not addressable, and useful only as a denominator.
Addressable market — the real one
Total, cut by the filters that decide whether you can actually sell: size band, stack, business model, compliance constraints. This is the number to plan against.
Workable market — what your team can touch
Addressable, cut by capacity. If two reps can work 80 accounts a week, your annual workable market is roughly 4,000 — and if addressable is 40,000, your problem is prioritization, not sourcing.
The gap between addressable and workable is what the scoring engine in lesson 07 exists to manage.
Counting properly
Add filters one at a time and record the count after each. The sequence tells you which filter is doing the work.
| Filters applied | Count | Removed |
|---|---|---|
| Industry: software | 184,000 | — |
| + Country in target list | 41,200 | 78% |
| + Employees 50–500 | 6,400 | 84% |
| + Founded after 2015 | 4,100 | 36% |
| + Funding in last 24 months | 1,850 | 55% |
Two things fall out immediately: geography and size do most of the cutting, and the funding filter halves the market — which is a big commitment to make on a criterion you may not have validated.
Any filter removing more than about half your market deserves a validation check: does it hold for your existing best customers? A criterion that excludes 55% of the market and 40% of your customer base is not a qualifier, it is a mistake.
Sanity checks
A count can be wrong in both directions, and both are diagnosable.
| Symptom | Likely cause |
|---|---|
| Suspiciously small | Over-filtered; a filter is a signal not an attribute; a value is spelled differently than the source expects |
| Suspiciously large | A filter is not applying — check that region strings and industry tags resolved |
| Round number at exactly a cap | You hit the result limit, not the market size |
| Zero | A filter combination that cannot co-occur, or a typo in a value list |
Cross-check against something you know: how many companies in your market do you believe exist? If your count is ten times that, one filter is not applying.
Estimating the run
Once the count is right, the cost estimate is arithmetic. Use expected survival rates between stages.
| Stage | Rows | Cost each | Total |
|---|---|---|---|
| Source | 1,850 | 0.1 | 185 |
| Enrich company | 1,850 | 0.5 | 925 |
| Tech stack | 1,850 | 0.5 | 925 |
| Filter → 40% survive | 740 | 0 | 0 |
| Find people, 2 each | 1,480 | 0.3 | 444 |
| Email + verify | 1,480 | 1.1 | 1,628 |
| ~4,100 |
Now you can decide whether to run it, tighten the definition, or split it into phases — before spending anything.
Do this now
Count the total market
Industry and geography only.
Add filters one at a time
Record the count and the percentage removed after each.
Flag any filter removing over half
Validate it against your best customers before keeping it.
Compute the workable number
Weekly capacity × 50 weeks.
Run the sanity checks
Especially the “did I hit a cap” one.
Estimate the run cost
Per stage, with survival rates.
Check your balance
check_credits, against the estimate plus 20%.
Check your work
- You have three numbers: total, addressable, workable
- You know which filter removes the most
- No count sits suspiciously at a round cap
- You have a per-stage cost estimate
- Your balance covers it with headroom
Where this breaks
Importing before counting is how a sourcing run becomes a four-figure surprise. A definition that looks tight can return 40,000 companies because one filter silently failed to apply — and by the time you notice, the enrichment columns have already run against all of them. The count is free. Read it every time, including on re-runs after a filter change.
Further automation
Recount monthly without importing. The addressable number moving is itself information: growing means your market is expanding or your definition has loosened, shrinking means the opposite. Lesson 08 turns the recount into a scheduled refresh.
Next lesson
03 — Source companies at scale, turning the counted definition into rows.
Reference for this lesson: find_companies, Credits, check_credits, Company.