CRM setup checklist: build it before you send a single email

This CRM setup checklist has two parts. First you build the structure, in order: objects, lifecycle stages, required fields, stage timestamps and a scoring gate. Then you run five cleanup passes over the data you already hold, before you send a single email.

"It is relatively clean." We hear it on sales calls, and it is rarely anyone's fault. Half the records sit in HubSpot, half in a spreadsheet someone still updates, and the two disagree. Nothing looks broken until you need a report you can trust.

This is the order we run the fix in. It is the same on any CRM, whether that is HubSpot, Attio or something older. It is one of the first jobs in GTM engineering. Every system built afterwards reads from it and writes back to it.

01The CRM setup checklist at a glance

The two parts of the CRM setup checklist are different in kind. The five build steps are a sequence: you cannot choose required fields before you have decided what a record is. The five cleanup passes are parallel passes over the same data. They can run in any order, and all five have to happen.

02Part one: how to set up a CRM from scratch, in this order

If you are setting up a CRM from scratch, these five come before the first import. Each one depends on the one above it, so the order is the checklist.

1. 01Objects before anything elseDecide what a record can be before you import one: companies, contacts and deals. Why it matters: every stage, field and report after this refers to these objects, and records imported before they exist end up filed wherever there was room.
2. 02Lifecycle stages you can observeName each stage after something that actually happened, such as a reply, a booked meeting or a sent proposal, and give one owner the decision to move a record. Why it matters: a stage named after a feeling gets applied differently by every person, so the report compares different things under one label.
3. 03Required fields, kept shortEvery required field is a tax on everyone who touches a record. If nobody reads it in a report, it should not be required. Why it matters: a long list of required fields gets filled with whatever gets the record saved, and that junk then looks like data.
4. 04The pass-through rates you will want laterPick the stage-to-stage rates you will report on, then stamp the date a record enters each stage. Why it matters: without those dates you cannot measure how long anything takes, and they cannot be filled in honestly after the fact.
5. 05Qualify and score before you importResearch the company, check it against the ICP and score it A, B or C. Work A and B, and exclude C. Why it matters: cleanup usually gets treated as a pass over what is already there, and it is far cheaper as a gate on what goes in.

Scoring before import saves the most cleanup time. A team that scores first spends its cleanup time on the records it actually wants to work.

03Part two: the cleanup before you send a single email

These five run on the data you already have. None of them depends on another, so split them across people if that is faster. What matters is that all five are finished before the first send.

1. 01Duplicates and hierarchyDedupe on domain. Then fix the parenting: find the right parent for each subsidiary, recheck the ones already linked, and confirm the parent exists at all. Why it matters: a duplicate splits one account's history across two records, so neither of them tells you the truth.
2. 02Dead contactsCheck the person still works there before you keep the record. Re-enrich the email and phone, and confirm the title is current. Why it matters: people change jobs, and an email to someone who has left bounces, which costs you sender reputation.
3. 03Unsubscribes and suppressionsEvery unsubscribe, hard bounce and do-not-contact goes in one list, checked last before a send. Why it matters: this is the list that protects the people who asked to be left alone, and your domain along with them.
4. 04OwnershipOne owner per record. Why it matters: unowned records are how one person gets emailed twice in a week by two people from the same company.
5. 05The fields nobody fills inList every custom field and when it was last written to, then backfill it or delete it. Why it matters: an empty field is worse than none, because someone will filter on it and believe the answer.

04Where every reply, call and meeting gets logged

"Where does that activity get logged?" is the question we hear on sales calls as soon as the CRM comes up. The answer we build to is that every reply, call and meeting lands in the CRM, against the right contact and the right owner.

The list you are working lives in the sending tool, for example Smartlead or EmailBison for email and HeyReach for LinkedIn. When someone raises a hand, three things happen.

1. 01The record is written. The contact and company are created or matched in the CRM, deduped on domain, with the reply attached.
2. 02The owner is told. A Slack alert goes to the record's owner, so sales can act the same day.
3. 03The stage is stamped. The record enters its first lifecycle stage with a date, so the pass-through rates start counting from the right moment.

Calls and meetings follow the same rule: they log against the same contact, and a booked meeting moves the deal to its next stage.

One case prospects ask about is the field rep. Let's say a rep reaches a customer through outreach, then takes an inbound call from the same person a week later. Both land on one record only if the duplicates and ownership passes were done: one record per domain, and one owner who sees both.

The write-back itself is an API job, and every current CRM has an API. The time goes into the conversation before it: which fields get written, what hierarchy you want and how granular it should be. The scoring from step 5 runs upstream, and its grade is written into a field you agreed on in step 3.

05Why the CRM goes before outbound

For cold outbound, the list lives in the sending tool and the CRM receives the people who raise a hand. The CRM matters most when you send to data you already own: your customers, old opportunities, an export from two years ago. Then your CRM is the list.

That is also why the CRM setup checklist goes first. Outbound into a dirty CRM still produces activity, and none of it can be reported on. You cannot tell a stage that converts badly from one that was recorded badly, and every report from those weeks gets rebuilt.

Owned lists also carry the most risk. An old customer who gets a cold first-touch email has a worse experience than a stranger. A suppression that was never merged in shows up as a complaint. The dead-contact, suppression and ownership passes are there for exactly these sends.

06Who runs this checklist

In a small team the CRM setup checklist usually lands with whoever owns the tooling. Often that is a GTM engineer, since the same person builds the outbound system that writes back into the CRM. In a larger team it sits with RevOps.

By RevOps we mean three jobs: one owned record per account, stages and dates that match what actually happened, and reports that tie every meeting back to the campaign that sourced it. We see growing volume as mostly a tooling question. A well-built system can carry more leads, more campaigns and more records without adding anyone.

The person becomes necessary when decisions arrive faster than the team can make them. Before that hire, four things should already exist: a working motion, a defined ICP, data the team trusts and reporting that survives being questioned. This checklist is how you get the last two.

When a client later brings the system in-house, the same checklist is what makes the handover work. A clear, standardised CRM is something a new hire can take over and run from the first week.

07Running the CRM setup checklist on an existing CRM

Our builds usually start from a CRM that already exists. The checklist still applies in the same order, with one change. Steps 1 to 4 become an audit of what is there. Step 5 becomes the gate for everything imported from now on.

• Audit the objects. Look for leads stored as contacts, deals with no company and companies used as folders.
• Rename the stages. Map each old stage to an observable event, and merge the ones nobody can tell apart.
• Cut the required fields. Anything no report reads goes back to optional.
• Start the timestamps today. History before this point will be incomplete, so report on it separately.
• Gate the next import. Score it A, B or C before it touches the CRM, then run the five cleanup passes on what was already there.

We run this CRM setup checklist for every client before anything gets sent. It is also the core of our RevOps service, where we fix the stages, the data, the routing and the reporting in dependency order.