How an offer becomes a campaign map: the market counted, every account classified, the public record mined for pain, and copy written for each situation.

Understand the service and the industry before touching data

Everything downstream depends on two questions answered precisely: what does the offer actually do, mechanically, and who in the market has a reason to care this quarter rather than some day. Spend the first block of time here. The maps look like data projects; they are offer projects that use data.

What to do

1. Write the mechanism in one paragraph that a skeptic would accept. Not the pitch. What changes, for whom, under what rules, and what has to be true for it to work. If the mechanism depends on a tax position, a regulation or a filing threshold, name it and cite it.
2. Write the compliance lock. The sentences that can never appear in copy, the words that are banned, the claims that need a professional in the room first. Do this before copy exists so the copy is born inside the fence rather than edited into it.
3. Name who is in the room when it closes. The owner, the CFO, the controller, the HR lead, the broker, the CPA, the actuary. Each one is a potential recipient and a potential blocker, and the copy for a blocker is not the copy for a buyer.
4. Find the natural clock. Most offers have a date on which the buyer is already forced to think about the subject: a renewal, a plan year end, a filing deadline, a funding notice, a contract expiry. The campaign calendar is built around that clock, not around your sales targets.
5. List the public records the market leaves behind. Regulatory filings, licensing databases, court records, permits, procurement notices, job postings. The one that names the buyer, the date and a size figure is the spine of the map.
6. Read the industry's own material. Carrier renewal letters, trade association surveys, the regulator's guidance notes, the two or three studies the industry quotes to itself. The copy will borrow their vocabulary; it has to be correct.

What it produces

A short internal brief: mechanism, compliance lock, buyer and blockers, the clock, the public records, the industry vocabulary. On the Section 125 map that brief was a separate document written before any data was pulled and it is cited on the page as the source for the ICP and the compliance rules.

Identify how many are in the market

The market is counted, not estimated. Start from the full public file, apply each filter one at a time, and write down the count after every filter so the funnel is auditable. A reader should be able to see exactly where a company fell out and why.

What to do

1. Pull the raw file fresh. Do not reuse a file pulled for a different offer, even from the same source. The pension map and the Section 125 map both start from Form 5500, and they share almost no rows: one keeps pension codes, the other keeps welfare code 4A. A stale or borrowed file is the most common way a map goes wrong silently.
2. Learn the file's own rules first. Who is required to file, who is exempt, what each code means. Two rules decided the shape of the Section 125 market before a single row was counted: cafeteria plans themselves never file (IRS Notice 2002-24), and small insured or unfunded welfare plans under 100 participants are exempt (29 CFR 2520.104-20). So the visible market starts at 100 participants, and the 133,388 US firms with 50 to 99 employees (Census SUSB 2022) are invisible on this record and need a different source.
3. Filter one condition at a time and log the count: the right plan type, the right entity type, US only, not a final filing, one row per employer (dedupe on EIN, keep the largest plan), the size band, the exclusions the offer demands (union plans, government, the wrong funding type).
4. Check a code against the instructions, not against intuition. On the first Section 125 run the entity code filter kept code 1, which the instructions define as multiemployer, and the universe came back as 1,523 union sponsors. Code 2 is single employer. The fix was one character; the lesson is to read the layout document before trusting a count.
5. Record the denominators. Every percentage on the map names what it is a percentage of. "48% renew in Q4" is wrong if 64% of insured plans and 66% of all plans do; both were put on the page with their denominators.
6. Put the funnel in a table on the page and keep the script that produced it beside the page.

What it produces

A universe file, one row per account, with a stable key (EIN, plan number), the filter log, and a funnel table on the page. Everything after this step reads from that one file.

Classify every account: the grid

The grid is the heart of the map. Two axes, every account in exactly one cell, every cell assigned to a campaign. The axes are not chosen for neatness; they are the two things that most change what you would say to the account and when you would say it.

What to do

1. Choose the timing axis. It comes from the clock found in Step 0: the quarter the renewal lands, the plan year end, the filing window. Accounts whose clock strikes soonest go first, which is why the first campaign on both maps is the nearest quarter.
2. Choose the fit axis. The one or two attributes that change the message itself: size band, state tax regime, industry type, funding type, whether a consolidator broker sits between you and the buyer. On the Section 125 map, income tax state changes which paragraph the employee benefit gets, industry changes whether the retention line is used, and broker type changes who is written to at all.
3. Write the assignment rule as code, not prose. Each account must land in one campaign by a rule a script can run. If an account could land in two, the rule is not finished. The Section 125 rule, in order: no medical contract on Schedule A, campaign C; renewal in Q1, Q2, Q3, campaigns A1, A2, A3; renewal in Q4, split by band, tax state and industry into B1 to B4.
4. Count every cell and put the counts on the page with the campaign tag in each cell. The grid should answer "how many are in B2" without a query.
5. Name the campaigns by what they are, not with marketing names: "Q1 renewals, insured" is better than a code name, and the code (A1) is only there for the sequencer.
6. Write the exclusions into the grid as well. The bands and types left out, and why, in a table beside the grid. A reader should not have to guess why 500 plus is absent.

What it produces

A group column on the universe file, a grid on the page with a count and tag in every cell, and a campaign list with one line of definition each.

See what public information can enrich the ICP, before paying for anything

Enrichment has two tiers. The first is free: every field on the public record that can be merged into copy. The second costs credits: names, verified emails, LinkedIn profiles. The map lists the free tier completely, with coverage, before a single credit is considered, because the free tier is what makes the copy specific and the paid tier only makes it deliverable.

What to do

1. Inventory every usable field on the record and any schedule that joins to it. For each: the source field name, what it means in plain words, the coverage (the share of the universe where it is populated), and how it appears in copy.
2. Join the schedules. On Form 5500 the health plan's carrier, policy dates, covered lives, premium, commission and broker all live on Schedule A and its Part 1, joined on the filing's ACK_ID. The join is where most of the specificity comes from.
3. Pull the prior year and diff it. Year over year change is the richest free signal there is: who changed carrier, who changed broker, whose headcount moved, who filed for the first time. None of it is on a single filing; all of it is in two.
4. Publish the coverage honestly. A field at 13% coverage (premium on the Section 125 record) is a sentence that is dropped where missing, not a field the copy depends on. A field at 0% (first name) is stated as 0% and is the reason paid enrichment exists.
5. Check a field's meaning, not just its presence. The "carrier" on the largest Schedule A contract is sometimes a voluntary benefits carrier, not the medical carrier. The page says so, and the copy drops the carrier line until the contract type is verified.
6. Price the paid tier per account and per group with the credit schedule written out, so the decision to enrich is a number: how many rows, at what cost per row, for which group first. On the Section 125 map that is an editable calculator on the page. Spend nothing until the list is final and the group is chosen; the ranking in Step 4 tells you which rows deserve the first credits.
7. Write every merge field down with its fallback. If {{firstName}} is missing, what goes in its place. If {{brokerCommission}} is missing, the sentence is dropped. The fields table on the map is the contract between the data and the copy.

What it produces

The custom fields table (field, source, coverage, use in copy, fallback), a universe file enriched with the free tier, and a cost calculator with the paid tier priced per row.

Identify the pain points and the problems, then measure them

There are two kinds of pain on a map: pain you can see in a field, and pain you believe is there. Both belong on the page, but they are labeled differently and used differently. Measured pain becomes a trigger line and a score. Believed pain becomes a paragraph in the body and a hypothesis to test by reply rate.

What to do

1. List the pains the offer brief claims, in the brief's own ranking, and set them beside the record. For each one ask: which field, or which year over year change, would show this pain if it were real?
2. Define each flag as a rule with a threshold: headcount down 10% or more, headcount up 15% or more, carrier changed, broker changed, first year filing, waiver rate 25% or more. Thresholds are written down so they can be argued about and changed.
3. Measure the share of the market carrying each flag. That share is also the first honest test of the brief: if the brief says everyone is in pain and the data says 15% changed carrier, the campaign is built on the 15%, and the page says so.
4. Build a score that adds the flags with weights, adds fit (band, state, industry) and subtracts obstacles (a consolidator broker in the way). Keep it additive and documented in a table so anyone can recompute a row by hand. The score orders the group; it does not replace the group.
5. Rank each campaign by score and count the bands (70 plus, 50 to 69, 30 to 49, under 30). The top band is who gets enriched first and written to first.
6. Separate the audience from the pain. A broker of record is not in pain; a broker is in the way, or is a channel. The map routes the 988 A1 accounts behind consolidator brokers to a broker track with its own copy rather than pretending the owner will act around their broker.
7. Keep the assumptions visible. Pains that could not be measured (margin pressure, turnover cost, owner fatigue with benefits) are listed under "assumed" with the reply rate that would confirm or kill them.

What it produces

A flag table with shares, a score table with weights, a score column on the universe file, score bands per campaign, and the audience split into buyer, blocker and channel.

Design the copy for every pain, problem and situation

Copy is the last step, not the first, because by now the writer knows exactly who is being written to, what is known about them, what date they are already thinking about, and what cannot be said. The copy is a matrix: a variant for each fit segment, a trigger line for each measured flag, and a track for each audience that is not the buyer.

What to do

1. Fix the sequence before the words. Both maps use three emails over fifteen days (day 1, day 5, day 8) with LinkedIn alongside (connect day 2, follow up day 6, breakup day 10), never the same email twice to the same person, and a cap on touches per channel per quarter. The sequence is the same for every campaign so that only the words change.
2. Write the base variant for the largest segment in full: three subjects per email, body, call to action, sign off, footer. Everything else is a diff against it.
3. Write each other variant as a diff: "replace the employee paragraph with", "drop the sentence beginning". Diffs are reviewable; four full rewrites are not. On the Section 125 map V2 (salaried industries) and V3 (no income tax states) are three and two changes against V1.
4. Write one trigger first line per measured flag, ordered by priority, and prepend the highest priority one that is true of the row. The first line is the only place the account sees that this email was written for it, so it carries the measured fact and nothing else.
5. Write the track for the blocker or channel. The broker track on the Section 125 map writes to the producer, not the employer, offers them the illustration, and lets them say no. If they go silent for 60 days the account re enters the buyer track with an opener that says so.
6. Write the re entry for 60 days after breakup with a different opening angle, capped at three cycles per account, so no account is written to forever.
7. Run the compliance lock over every line before the page is published: banned words, dollar figures, guarantees, the footer on every email. On the Section 125 map that is a zero count check on the copy file.
8. Put the full copy on the map, campaign by campaign, with the merge fields visible, so the sequencer load is a copy and paste from the page and the reviewer sees exactly what will be sent.

What it produces

A copy file per campaign group with the segment to variant map, trigger lines, the base variant, the diffs, the blocker track, the re entry and the footer; the same copy rendered on the map under each campaign.

Put the plan in numbers: volume, cost, calendar

With the groups counted and the copy written, the plan is arithmetic. The map carries it as editable calculators so the person approving spend can change a rate and see the result rather than ask for a new deck.

1. Volume to outcome. Daily sends, delivered rate, reply rate, positive reply share, meeting rate, close rate, each editable, projecting replies, meetings and signed accounts per group. Every default rate is labeled as an assumption until the campaign has its own data.
2. Cost per group. Rows to enrich, credits per row, the credit schedule, inbox count and cost, sequencer seats. Cost per meeting falls out of the two tables together.
3. The twelve months. A Gantt row per campaign, placed by the clock from Step 0: the nearest renewal quarter first, the Q4 block split into its four campaigns, the self funded group last because its copy is different and its pain is less measured.
4. Behind the page. A claims register (every number, where it came from, the script that made it) and a sources list. This is what makes the page auditable a month later by someone who was not there.