How to Size Your First Outbound Campaign: TAM, Signal and Campaign List

Size an outbound campaign before you write a word: count your TAM, signal and campaign list, then plan the data budget, mailboxes and email type.

This is a guide to sizing an outbound campaign before you write the first email.

We’ll go through one example step by step: Billowtide, an accounts-payable tool that sells to US manufacturers. Accounts payable (AP) is the team that pays the supplier invoices.

Say the market has 21,000 companies. 600 of them show a sign this month that they may need Billowtide. 250 of those have a named person with a working email or phone number. From these three numbers you plan the data budget, the number of mailboxes and the kind of email you write.

This guide is for founders about to send their first campaign, and for GTM (go-to-market) engineers who plan campaigns every month. If you already know your TAM, start at step 3.

1. Know the three numbers

Every campaign plan starts with three numbers:

  • TAM (total addressable market): every company that passes your test, whether you can reach it or not. Billowtide: 21,000 US manufacturers with 50 to 499 employees.
  • Signal: the companies inside your TAM that show a sign they may need you right now (a buying signal). Billowtide: 600 manufacturers that posted an accounts-payable job or named a new finance leader this month.
  • Campaign list: the signal companies where you also have a named person and a working email or phone number. Billowtide: 250.

Each number answers a different question. The TAM tells you whether the market is big enough for a year of campaigns. The signal tells you who to contact this month. The campaign list tells you who gets an email on Monday.

Mixing them up causes two mistakes. If you show only the signal companies, the market looks too small to be worth a campaign. If you plan with the TAM, you budget data and mailboxes for 21,000 companies when 250 are ready.

In the example, the campaign list is 1.2% of the TAM. That’s normal for a signal campaign. The other companies stay in the TAM until they show a signal of their own.

2. Take the TAM from your build, or estimate it

My TAM playbook shows how to build a TAM step by step: collect every company that could fit, then let AI check each one against your test. For sizing, you need two things from it: the count, and the companies that fit best.

If you built a TAM, use its count. It’s the most exact number you’ll have.

Without a TAM, estimate it from a filter count. Most company databases show how many companies match a filter before you export anything. Treat that number as a first estimate. A filter misses companies with the wrong industry label and lets in companies that only look right, so the real count comes from checking each company.

Compare your estimate with the Census. The Census Bureau counts US companies by industry and size for free. It counts 25,716 manufacturing companies with 50 to 499 employees (2022). Billowtide’s 21,000 sit below that, which makes sense: no database has every company, and some manufacturers fail the test. If your filter count is three times the Census number, the filter let in the wrong companies.

Then mark the companies that fit best. Fit is how closely a company matches your best customers, for example its size or its industry. Say most of Billowtide’s customers have 100 to 299 employees and come from four sub-industries. That’s 5,000 of the 21,000.

These 5,000 get the first campaigns, because they look most like the companies that already bought. The rest of the TAM comes later.

3. Pick a signal the prospect can check

A signal only helps the email if the prospect can check it. My test: can you put a link or a date in the row that the prospect could open themselves? If yes, the signal can go into the email. If no, use it only to decide which companies you contact first.

SignalWhere to find itCan the prospect check it?How long it stays useful (my default)
Job post for the role your product helpsCareers pages, job boards, job post searchYes, link and post dateWhile the post is open, plus a few weeks
New leader in the buying roleLinkedIn profile, press releaseYes, start dateThe first 90 days
Funding roundPress release, SEC Form D filingYes, filing dateA few months
New plant or locationPermits, local news, company siteYesA few months
Website visits, intent topicsVendor dataNoDays
Headcount growthCompany database estimatesHardly, estimates are often out of dateOnly to decide the order

Form D is the notice a US company files with the SEC within 15 days after the first sale in a funding round. Its date sits close to the event, which makes it easy to check.

Billowtide’s signal: a manufacturer posted an accounts-payable job in the last 60 days, or named a new controller or CFO in the last 90 days. Say that’s 600 companies this month.

Find the job posts with a search tool. BlitzAPI, LeadMagic and TheirStack all search job posts by job title and keyword. BlitzAPI includes job search in its flat-rate plan, one fixed price a month instead of a price per result. LeadMagic charges per job post it returns.

Search the full text of the post, not the title. A title like “Finance Associate” hides AP work, and a short summary drops the duties. Search the whole description, and list every form of each phrase: “accounts payable”, “AP”, “invoice processing”, “invoices”. A search for “invoice” may not find “invoicing”, so write every form down.

Use two time windows. A post from the last 60 days puts the company on the list, because the problem is still there. Name the post in an email only if it went up in the last 30 days and is still open. Companies whose post is 31 to 60 days old get an email without the post in the first line. A post that was taken down makes your first line wrong.

Count how many signal companies appear over time. Run the same search for posts from the last 30, 90, 180 and 365 days and write down all four counts. In my experience the 365-day count is far below twelve times the 30-day count, because the same companies post again and again. The four counts tell you how many new companies a month you’ll see.

Paste this into Claude Code to get the counts:

Read the API docs of <job search tool> at <link to the docs>.
Build tools/count_signal.py. Search US job posts whose full description
contains any of these phrases: <every form of each phrase>.
Only keep employers whose domain is in tam.csv.
Run it for posts from the last 30, 90, 180 and 365 days.
Print, per window: posts found, unique employers, employers in the TAM.
Save every raw response to raw/<window>/. Before the run, print the
expected number of requests and credits and wait for my yes.

4. Read the job post, not only the keyword

A keyword match tells you that a word appears somewhere in the post. It doesn’t tell you where it stands or why. Say you search job posts for “accounts payable”. Five posts, all matching:

  1. “AP Specialist. You’ll process 2,000 invoices a month and match them to purchase orders in Excel.” That’s the need. The job exists because the work is manual.
  2. “AP Automation Lead. Own the rollout of our new AP platform.” They already bought a tool. That’s a conversation for next year at the earliest.
  3. “Staff Accountant. Clear growth path toward AP Manager.” The keyword sits in the career path. The job is something else.
  4. “AP Clerk for our client, a manufacturer in Ohio.” A staffing agency posted it. The agency isn’t the company you’d sell to.
  5. “Questions? Contact Sam, Accounts Payable Manager.” A signature at the end of the post.

All five match the search. Only the first one is the signal you wanted.

Read the duties, because job titles are wrong in both directions. At a 120-person manufacturer, an “Office Manager” post can be mostly AP work. A “Finance Operations Associate” can be AP under another name. Decide by what the person will do all day.

The phrase and the right kind of role must be in the same post. A company may post a sales job that mentions “accounts payable” in passing. List the role families that count, like AP, accounting and finance operations, and only accept a post when both match.

Here’s the prompt I’d give Claude Code for this step:

You classify job posts for one question: does this employer need
<PRODUCT CATEGORY> right now?

For every place the keyword appears, label it: DUTY, REQUIREMENT,
CAREER_PATH_OR_BENEFIT, COMPANY_TEXT, SIGNATURE, OTHER.
A section heading before the keyword counts more than text after it.

Return NEED only if at least one hit is a DUTY or REQUIREMENT of the
advertised job, the job belongs to one of these role families:
<ROLE FAMILIES>, and the company posting it is the employer, not a
staffing agency.

Quote the sentence behind each label. If you can't quote it, return UNKNOWN.
Output one JSON line per post: {post_url, employer, post_date, label, quote, verdict}.

Test the prompt on posts you labelled by hand before the full run. Read 10 posts yourself and compare your answer with the model’s. When you disagree, fix the prompt and take 10 new posts. Stop when two rounds of 10 in a row match. Then run the prompt on every post, not on a sample.

5. Write the first line so it’s true in three weeks

The email goes out days or weeks after you found the signal. By then the job post might be closed and the job filled. So write the line in a way that stays true after the post is gone.

  • Weak: “You’re hiring an AP specialist right now.”
  • Better: “You posted an AP specialist role in October.”

The second line has a date and uses the past tense, so it’s correct in week three too. Neither line says how you found out. The sentence is about their situation, and the link to the post stays in your sheet as proof.

Three rules I’d keep:

  • Use the month you found the signal. Words like “now”, “currently” and “this week” are the first to become false.
  • Check the signal again before every follow-up. If the post is gone, the second email can’t talk about it as open.
  • Give each signal an end date. My defaults are 60 days for job posts and 90 days for a new leader. Companies past that date leave the campaign.

6. Turn the signal companies into the campaign list

The campaign list is what’s left of the signal companies after five checks. I call each check a gate: a company has to pass it to move on. Run the free gates first and the paid ones last. Write down a reason for every company you drop, so you can check the drops later.

For Billowtide, from 600 to 250:

  1. Country and size: 600 → 540. The job is in your sales territory, and the company is still in the TAM’s size range.
  2. Not a customer, open deal, competitor or partner, and nobody asked not to be contacted: 540 → 500. Match the website domains against your CRM, the database where you track customers and deals.
  3. Signal confirmed in the full post: 500 → 350. The prompt from step 4 removes agency posts and every keyword match outside the duties.
  4. Buyer named: 350 → 300. Every person with a buying title, like controller, AP manager or CFO. Take all of them, not one person per company.
  5. Reachable: 300 → 250. A verified email or a working phone number.

Gates 1 and 2 are lookups. Gate 3 costs a model run, which is cheap. Gates 4 and 5 cost money per person, so they only see what passed the first three. Gate 2 comes before them for that reason: you never pay for the contacts of a customer.

For gate 4, start with the flat-rate people databases. GetLeads and BlitzAPI are flat rate: one fixed price a month instead of a price per contact. Use Prospeo last, because you pay per contact there.

Gate 5 counts only emails a verifier checked. A verifier like MillionVerifier asks the company’s mail server whether the mailbox exists, without sending an email. It returns valid, invalid, unknown or catch-all. Keep only the valid ones and drop every “unknown”. Catch-all servers accept any address, so send to them only after a tool like LeadMagic or BounceBan confirms the address.

Before you pay for contacts, run gates 4 and 5 on 50 companies. Count how many end up with a named person and a working email or phone. That rate tells you what the full run will bring and cost. Measure it on the signal companies themselves, not on the whole TAM. Step 8 shows how far the two rates can differ.

7. Let each number set one part of the plan

Now each number answers one planning question. Write them down before you start writing the emails.

NumberExampleWhat it decides
TAM21,000Whether the market is big enough for a year or more of campaigns
Best fit5,000Where the first campaigns go
Signal per month600How often you run the gates: every week, on that week’s new signal companies, so the list keeps filling
Campaign list250This month’s data budget, sending volume and copy

The campaign list tells you how many emails you’ll send. Say gate 4 finds two people per company on average, and each of them gets three emails over three weeks. That’s 250 × 2 × 3 = 1,500 emails.

A mailbox is one sending email account. Say each mailbox sends at most 30 emails a day, on 15 working days. 1,500 emails in 15 days is 100 a day, so four mailboxes cover it. And you pay for contact data on 500 people, not on 21,000 companies.

The copy follows from the number too. With 250 signal companies, every first line can come from the company’s own signal. The 5,000 best-fit companies without a signal get a different email with a broader offer, and you should expect fewer replies from it.

8. Measure every rate on the group you’ll send to

A rate you measured on the whole market says little about a smaller group you picked out of it.

Say you checked 1,000 random companies from the 21,000. At 70% of them you found a named finance leader with a verified email.

Then you plan a more targeted campaign on one part of the signal: manufacturers hiring their first AP person, 120 companies. You apply the 70% and promise 84 reachable companies. The contact search comes back with 30, which is 25%.

The signal picked the group. A manufacturer hiring its first AP person often has no finance team yet. It’s smaller, younger and harder to find in every database. In my experience the most specific signals often point to the companies with the least data.

  • Never use a rate from one group for another group. Measure it again on the group you’ll send to, even if that takes an hour. The hour costs less than a budget approved on the wrong number.
  • Write down what a percentage is based on. “70% of 1,000 checked” and “70% of 11 companies we could count” mean very different things.
  • Reply rates work the same way. A rate from a broad campaign won’t be the same in a signal campaign, and only a measurement tells you which way it moves.

9. Fill in the sizing worksheet

Copy this into a sheet, one per campaign:

SIZING WORKSHEET

1  TAM
   Test question: ________
   TAM count: ________     Built or estimated: ________
   Census count for the same industry and size: ________
   Best-fit definition: ________     Best-fit count: ________

2  SIGNAL
   Signal: ________     Link or date in the row: ________
   Phrases, every form: ________     Role families: ________
   Find window / name-in-email window in days: ____ / ____
   Count at 30 / 90 / 180 / 365 days: ____ / ____ / ____ / ____
   Prompt tested by hand: two rounds of 10 matched (y/n)
   Signal companies this month: ________

3  CAMPAIGN LIST          in      out     main reason
   Gate 1 country, size   ____    ____    ________
   Gate 2 do not contact  ____    ____    ________
   Gate 3 full post       ____    ____    ________
   Gate 4 buyers named    ____    ____    ________
   Gate 5 reachable       ____    ____    ________
   Campaign list: ________

RATES (one line per rate, measured on the group you'll send to)
   Rate | measured on which group | companies counted | value | date

DECISIONS
   Contact data budget this month: ________
   Emails and mailboxes: ________
   First line, with the month: ________
   Date to check the signals again: ________

Where to start today

You can size your next campaign today in three steps:

  1. Get your TAM count, from your build or from a filter count you compared with the Census (step 2).
  2. Count your signal with the script from step 3, for the last 30 and the last 365 days.
  3. Run the five gates on this month’s signal companies. The companies that pass are your campaign list, and the worksheet from step 9 turns it into the budget and the mailboxes.

Blog

Keep Reading

Start With a System Check

Start with a free target market map or a short system check. Either way, you leave with something useful.