From TAM to outbound campaign

A TAM goes out as a series of small campaigns, one per segment. My test for a segment: can you write one opening sentence that is true of every account in it? If the sentence needs a “depending on your situation”, the segment is still too wide and the sequence will read as if it were written for nobody.

Throughout this chapter, say you sell payroll software to French SMEs with several sites, and your TAM holds 6,200 scored accounts. All the figures below are illustrative.

Cut the TAM until one sentence fits

Start from the tiers built in the scoring chapter, then split the top of the list by what you can honestly say to each group.

Segment	Filter on the TAM	Sentence true for all of them	Accounts
A. Retail, growing	Retail NAF codes, 5+ sites, payroll or HR role open in the last 90 days	You are adding sites and payroll now needs its own hire	180
B. Hospitality	Hotel and restaurant NAF codes, 3+ sites	Your headcount moves with the season and payslips follow	420
C. Multi-site, no signal	Top-tier sectors, 5+ sites	Each site adds its own rules to your payroll run	1,150
Parked	Everything else	No sentence yet	4,450

The parked line matters. Those accounts stay in the database until a signal or a new angle gives you something true to say, and the signals chapter is where that comes from.

Keep one persona per campaign. A CFO and a payroll manager at the same company have different problems, so they get different sequences, coordinated at account level so both don’t hear from you in the same week. The contacts chapter covers who to pick.

Remove everyone who should not receive this

The size of a segment and the size of a campaign are two different numbers. Run the exclusions before you write a line of copy.

SELECT c.contact_id, c.first_name, c.email, c.linkedin_url,
       a.company_name, a.site_count, a.last_signal_role
FROM contacts c
JOIN accounts a ON a.account_id = c.account_id
LEFT JOIN suppression s ON s.email = c.email OR s.domain = a.domain
WHERE a.segment = 'A'
  AND c.persona = 'payroll_manager'
  AND a.status NOT IN ('customer', 'open_deal', 'partner', 'competitor')
  AND c.email_status = 'valid'
  AND s.email IS NULL AND s.domain IS NULL
  AND (c.last_contacted_at IS NULL
       OR c.last_contacted_at < CURRENT_DATE - INTERVAL '120 days');

The same logic works in a spreadsheet with filters and a suppression tab. Write the loss at each step, because it tells you where the TAM is weak.

Step, segment A	Remaining
Accounts in the segment	180
After customers, open deals and partners	164
With a payroll or HR contact found	131
With a valid email	118
Not contacted in 120 days, not opted out	112

Your real campaign is 112 people. Before launch, read the CNIL page on electronic direct marketing, which separates messages to professionals from messages to individuals, and check the equivalent authority for each country you target. I am not a lawyer and this is not legal advice: have counsel validate your setup, and check the terms of use of each data source and platform. The chapter on upkeep covers opt-outs.

Size the test before you write the sequence

Decide what you can learn from the volume you have. Split 112 contacts into two variants and each one gets 56. At that size, a gap of two replies is noise, so test things that could change the outcome by a wide margin: the opening angle, the persona, the channel order. Subject line tests can wait for a segment of several thousand.

Then check capacity. Set a daily limit per sender after reading each platform’s rules, and count replies as work too. If one rep starts 25 new contacts a day, 112 contacts take about five working days, and follow-ups will still be landing three weeks later. Launch the next segment when the previous one has cleared, so that nobody is handling three sets of replies with three different arguments.

Write the sequence from the fields you have

Each step has one job and uses a field that exists in the TAM.

Step	Day	Channel	Job of the message	TAM field used
1	0	Email	State the observed fact, ask one question	site_count, last_signal_role
2	3	LinkedIn connection, no pitch	Become recognizable	linkedin_url
3	7	Email	One concrete example from the same sector	sector_label
4	12	Phone, if you have a number	Talk about the problem	phone, site_count
5	18	Email	Close the loop and offer to stop	first_name

For segment A, the first email could read:

Subject: payroll across {{site_count}} sites

Hello {{first_name}},

I saw that {{company_name}} is recruiting a {{last_signal_role}} while running {{site_count}} sites. At that stage, collective agreements and site-level variables tend to land on one person’s desk.

Is that where you are, or is it already handled?

A blank variable sends a broken message, so block the launch on this check, which has to return 0 for each required column:

=COUNTIFS(Segment!A:A,"A",Segment!F:F,"")

Move the segment into the sequencer

I co-founded Derrick, so take this section with that in mind. Here is how the last mile works there, from the app or the Google Sheets sidebar.

Company Hiring Signal takes LinkedIn company URLs and returns whether the company is hiring, how many roles are open and their titles, for 1 credit per company checked. That fills the signal column. Email Finder takes a LinkedIn URL, or a name and a company, and charges 5 credits per email found. For addresses that came from elsewhere, Email Verification costs 1 credit per address submitted. Push to La Growth Machine, a tool we partner with, costs no credits: you connect your API key once, pick the audience, map the LinkedIn URL column and your custom fields, and a column confirms each row went through. I would create one audience per segment.

Derrick stops there. It does not send anything, build the sequence or write results back into your CRM, so reply handling and CRM updates stay on your side. Without it, any finder and verifier will do, followed by one CSV per segment imported into your sequencer with these columns.

Column	Example	Rule
contact_id	c_01842	Stable TAM ID, kept as a custom field
segment	A	One value per file
variant	A1	Assigned before import
first_name	Camille	Cleaned, correct capitals
email	valid status only	No role addresses on a personal sequence
linkedin_url	full profile URL	Required for step 2
company_name	Maison Martin	Legal suffixes removed
site_count	12	Number, no blanks
last_signal_role	payroll manager	Written the way a person would say it

Write the results back into the TAM

Export the outcomes per contact_id and update the database, or the next campaign will repeat this one’s mistakes.

Segment	Contacts	Replies	Positive	Meetings	Decision
A	112	14	7	4	Widen to the next batch with a signal
B	240	12	3	1	Rewrite the sentence, retest on 100
C	300	9	1	0	Stop and go back to the segment definition

Store last_contacted_at, campaign_id and one outcome from a closed list: no_reply, positive, not_now, wrong_person, bounced, opt_out. A not_now gets a recontact date. Opt-outs go straight to the suppression table. A high rate of wrong_person or bounced means the data is at fault, and no rewrite of the copy will repair it.

Your next step: take the segment with the most recent signals, write its one sentence, and run the exclusion table. If at least 100 contacts survive, build the five steps above. If fewer do, widen one filter and run the table again.