Viability test · 29 July

Your free data reaches a third of your market. I'd been selling you the smallest slice of it.

You asked me to test whether the library database is actually worth it. I couldn't get into it — that needs your ID at a counter — so I tested the question underneath it instead: how much of your territory can you already reach for free? The answer changes what I recommended yesterday.

17,080
businesses with 20–49 staff in your eight counties
33%
you can reach with a phone today, counted generously
11,362
you have no way to phone at all
6%
share that is construction — the file I called the winner

The reversal

I've spent two days telling you the contractor licence file is the best thing on the table. On its own merits it still is — phone on every row, renewal dates nobody else sells. But I never checked how big construction actually is in your size band, and it's 1,054 companies out of 17,080. Six percent.

Restaurants and hotels are 4,061 — four times bigger — and you can only see the Santa Clara County ones. Retail is 1,974 and you have nothing. Professional services 1,886, nothing. Auto shops and salons 821, nothing.

So I undersold the library card, not oversold it. I pitched it as a nice free extra and spent the detail on a $104 mobile-number tool. It's the other way round: general coverage is the gap, and that card is the only free thing that fills it.

Every sector in your size band, and whether you can phone them

Sector20–49 staffSanta ClaraRest of BayCan you phone them?
Restaurants & hotels4,0611,1372,924partial county food file — Santa Clara only
Retail1,9745401,434none
Health care & social1,9604631,497yes care facilities + provider registry
Professional & technical1,8864611,425none
Construction1,054289765yes licence file, 100% have a phone
Admin & building services843212631weak injury file, no phones
Auto repair, salons, other821193628none
Manufacturing788275513weak injury file, no phones
Wholesale739196543weak injury file, no phones
Education583168415none
Finance & insurance566111455none
Information555165390none
Arts & entertainment397103294none
Transport & warehousing35466288partial trucking census only
Real estate29977222none
Company management17026144none

Census County Business Patterns, 2023 — the government's own establishment count by county and employee band. Their band is 20–49; yours is 24–49, so these totals run slightly generous. Shaded rows are where you currently have no phone route.

How I got to 33%, and why the real number is lower

I counted all construction and all healthcare as covered, plus Santa Clara restaurants, plus the 1,567 companies with a retirement filing. That's 5,718 of 17,080.

It's generous on purpose, and it's still only a third. The honest figure is lower because the licence file only covers licensed contractors, not every construction firm; the care-facility and provider files don't reach every healthcare employer; and the 1,567 retirement-filing companies overlap with the buckets I already counted. Call it a ceiling, not an estimate.

What I could not test, and why

I tried three ways into the database itself and all three closed:

So the gate is a photo ID at a counter, and that's yours to do — one visit, and the main library downtown is walking distance from your office. Until then I can't tell you whether their employee counts are any good, and that remains the open question.

When you have it, the first test takes ten minutes: look up twenty companies whose real headcount you already hold from their government injury filings, and compare. That settles the estimate question with your own data.

One thing this doesn't rescue

A general directory gives you a company and a main line for the missing 11,362 — which lands you back with the receptionist. It fixes coverage, not access. The mobile-number problem is separate and still costs money.

Worth being clear about the order: coverage first, because you can't dial a company you can't find. Then the cell number, and only for the rows worth it.

What I'd do differently now