Before you buy more leads, look at the records you already have.

Follow-up records can end up split across spreadsheets, CRM notes, inbox threads, old quotes, and an owner’s memory. When that happens, appropriate follow-ups may be missed. This guide helps you identify and organize records that deserve human review. It does not assume every record should be contacted or predict what any follow-up will produce.

It walks you through finding that leak and organizing it, with one rule underneath everything: nothing gets sent until you approve it.

What you need:a spreadsheet — Google Sheets, Excel, or Airtable all work. Set aside a focused block for the first pass; a larger or messier list may take longer.
What you don’t need: new software, a CRM migration, or any AI tool. You can do the whole thing by hand.

The five places your follow-ups leak

Before you spend another dollar on new leads, look in the five places you already have. Each one becomes rows in the single list you build in Step 1.

1. Old spreadsheets— customer lists, job logs, quote trackers, that one sheet from two years ago.
2. CRM notes— the notes nobody has opened in months, including contacts that were entered and never worked again.
3. Quiet inbox threads— conversations that just went silent. Search terms like quote, estimate, following up, and thanks for reaching out can help surface relevant threads.
4. Quotes, jobs, or inquiries that never received another touch— someone asked, you replied once, and it ended there.
5. Names or opportunities living only in your head— ones you remember but have not recorded. Write them down so you can assess whether any appropriate next action exists.

A follow-up you never wrote down is a follow-up you can’t act on. This gets all five into one place so you can finally see them.

Step 1 — Build one working list

Create a new spreadsheet and add these columns. One row per person or opportunity. Go through all five sources and add everyone you find — don’t clean, sort, or judge yet. Collection and judgment are separate jobs.

Paste-ready header row (tab it into cell A1):

name	company	contact	source	relationship	last_touch	welcome?	priority	draft	status	owner_decision	notes

Works identically in Excel, Google Sheets, or Airtable. If you use Airtable: one table, two views — Held for review and Approved, ready to send.

Blank vs. assumed — keep them honest.If you don’t know a value, write unknown— don’t invent one. If you’re estimating (a rough date, a best-guess spelling), mark it as an estimate (e.g. ~2024 or est.) so you can tell a real fact from a guess later.

Done when:every name you could find, from all five sources, is in one sheet — however messy.

Step 2 — Clean up duplicates (do this by hand)

A person may appear in more than one source under different spellings. Merge them so each person is exactly one row. This is a judgment task, not a find-and-delete:

1. Compare email and phone first.A matching email or phone is a strong indicator, but confirm it against name, company, and context — shared inboxes, household numbers, and recycled details can create false matches.
2. When identifiers differ or are missing, look at name, company, and context. “J. Smith at Acme” and “John Smith, Acme Ltd” with the same history may be one person; verify before merging. Two different Sarahs are not.
3. Merge the obvious duplicates into one row. Keep the most current contact details, the most recent last_touch, and the useful notes from both.
4. Flag anything ambiguous for a second look (e.g. notes: possible dup?) rather than force-merging.
5. Never merge records you’re unsure about.A wrong merge mixes two people’s history and produces a follow-up that references the wrong job.

Done when: each person appears once, ambiguous matches are flagged rather than force-merged, and no history was thrown away.

Step 3 — Add last touch and segment

Now add judgment to each row.

1. last_touch — the last meaningful interaction. Use the real date if you know it; label an estimate as an estimate; write unknownif you genuinely don’t know.
2. relationship — past_customer (you did work for them), quoted_no_close (you quoted, they never closed), or inquired_only (they asked, it went quiet).
3. welcome? — would a follow-up reasonably be welcome? yes / probably / no.
4. Permission / do-not-contact. If someone opted out, asked not to be contacted, or you only have their details from a purchased or scraped list, mark them and leave them out. Only reach out where you have both permission and a real relationship.

Rule: skip the no’s.If you’re unsure whether someone wants to hear from you, they probably don’t. Mark them and leave them out.

Done when: every row has last_touch, relationship, and welcome? filled, and anyone off-limits is clearly marked.

Step 4 — Rank what to work first

You can’t work the whole list at once, so put it in a sensible order. Set each row’s priorityusing your own judgment — no formula, no score:

• Age of last touch— how long they’ve been quiet.
• Warmth and relevance— how strong and recent the relationship is.
• A known open quote, job, or inquiry— something concrete already on the table.
• Permission and welcome— you’re allowed to contact them and a follow-up would land well.
• Owner judgment— cases where only you can decide whether it’s appropriate.

Turn that into a simple ordering: high (work first), medium, low, and do-not-contact(the no’s and off-limits rows, which you leave alone). This is a where to start ordering based on your judgment. It is nota prediction of who will buy — the guide makes no claim about conversion.

Done when: every workable row has a priority, and the do-not-contact rows are set aside.

Step 5 — Diagnose the leak

Before you write a single message, step back and look at the list you’ve built. Count and inspect:

• How many records made the list in total?
• How many are in each relationship group?
• How many have an unknownlast touch — people you’d genuinely lost track of?
• Where is the biggest cluster sitting? That is a useful place to inspect first.

That picture — your own numbers, from your own list — is the leak, made visible. It is a descriptive diagnosis: it shows where records are accumulating and how concentrated they are. It is not a revenue estimate, and this guide makes no claim about what any of it is worth in dollars.

Done when: the counts are written down and you can name where the leak concentrates.

Step 6 — Draft carefully (send nothing yet)

Write a short message for the rows you’ve prioritized. Use only the real context in the row. Put each draft in the draft column and set status to held.

Illustrative templates only— examples to rework in your own voice, filling [brackets] with details you actually know:

Past customer: “We did your [job] back in [time]— how’s it holding up? We’ve got room in the schedule this month if anything needs a look.”

Quoted, never closed: “Quoted you on [thing]a while back — priorities shift, no worries. If it’s still on the list, happy to refresh the numbers.”

Inquired, never quoted: “You asked about [thing]and it slipped through on our end — that’s on us. Still interested?”

The one hard rule for drafting: never invent.Don’t fabricate a job, a date, a price, a promise, or availability you can’t stand behind. If you don’t know a detail, leave it out rather than guessing.

If you use an AI assistant to draft:that’s fine — it’s good at first drafts, and it can work from the context already in the row. Drafting is the only job it gets. It never decides who to contact, and it never sends.

Done when:each prioritized, welcome contact has a personalized draft, and every draft’s status is held.

Step 7 — Hold every message before it sends

This is the rule the whole method rests on. Every drafted message stays at HELD until a person decides what happens to it.

1. Open the sheet and look only at rows where status = held.
2. Read each draft with the actual person in mind, then set owner_decision: approve (send it yourself, then set status to sent), edit (fix it, decide again), or reject/ skip (don’t send; note why).
3. One at a time. Eyes on every message.
4. When someone replies, set status to replied and write what happened in notes.

The queue rule: every message rests at HELDuntil a human chooses approve, edit, or reject. Nothing sends automatically. If you’re ever unsure whether a follow-up would be welcome, don’t send it. Anything sensitive or exceptional stays a human decision — always.

Done when: working the held queue is a short weekly habit and the notes column is filling up.

Step 8 — Take one useful first action

Don’t wait until the whole list is perfect. Pick a small handful of the clearest, warmest, most obviously welcome records — the ones at the top of your prioritycolumn. Draft those thoughtfully, review each by hand, and decide on each. That’s a complete first pass.

This is deliberately not a bulk-send playbook. A small batch is easier to review carefully, personalize honestly, and keep under owner control. Start small and inspect every message.

What this free guide does and doesn’t include

What it gives you:the complete manual method. With this guide and a spreadsheet, you can find your follow-up leak, organize it, rank it, draft carefully, and keep every message under your own approval — by hand, today, without buying anything.

What it does not include:

• Connecting your live tools — no integration with your real CRM, inbox, or booking system.
• Automated duplicate-matching across systems — the dedupe here is the manual version.
• Any credentialed access to your accounts or data.
• Monitoring, alerting, rollback, or production deployment.

Those are implementationconcerns — the work of building this into live systems and keeping it running reliably. They’re not information held back from you. The method on this page is the whole method. The line between doing it by hand and having it built into a business is accountable implementation, not a secret.

Before you start

You can begin this map with a spreadsheet. Start with a few records that clearly need review, add the required columns, and work through the steps in order. The judgment stays with you throughout. Go find the leak.