8 min read
A small team can set up a working CRM in one day, and a team of fifty can do it in a week. Most of that time goes to decisions, not clicking. If you make the decisions in the right order, the software configures itself around them in an afternoon.
This guide is for founders, sales leads, and ops owners setting up a CRM for the first time, or resetting one that never stuck. When you finish, you have a CRM with your real customers in it, a pipeline your team agrees on, and a daily habit that keeps the data current. If you are still picking a tool, start with how to choose a CRM. If you want the basics first, read what is a CRM.
The whole setup, in order:
Write down what the CRM must answer
Decide who owns it
Model your data
Define pipeline stages
Import your data
Connect email and calendar
Automate the first three chores
Run a one-week pilot
1. Write down what the CRM must answer
Open a blank document and list five questions you want the CRM to answer without anyone asking a person. Good ones sound like: Which deals are closing this month? Who last spoke to this customer? Which leads have gone quiet for two weeks? Which accounts renew next quarter?
These questions drive every later choice. Each field you add, each stage you name, and each automation you build should trace back to one of them. If it does not, leave it out. The most common reason a CRM becomes a chore is that someone added forty fields on day one and nobody could say what they were for.
The decision here is scope. Default to five questions. You can add more once the first five are answered reliably, and adding a question later costs far less than removing a field everyone has learned to ignore.
2. Decide who owns it
Name one person who owns the CRM. Their job is to approve new fields, fix broken data, and say no to requests that do not serve the five questions. In a team of five, that is usually the founder or the head of sales. In a larger team, it is a revenue operations hire or a sales manager with a few hours a week.
Ownership matters because a CRM with shared ownership drifts. Every rep adds a field that fits their own workflow, and within a quarter two people mean different things by "qualified." One owner keeps the definitions stable.
The decision is how much time the owner gets. Default to two hours a week for a team under twenty, and review the setup together once a month. Put the review on the calendar now, before the first record exists.
3. Model your data
Decide what kinds of things you track, and how they connect. Most B2B teams need three: companies, people, and deals. A person belongs to a company, and a deal belongs to a company with one or more people attached. Partnerships, investors, vendors, and recruiting pipelines can each become their own object when they follow a different pattern.
Keep each object to the fields you need to answer your five questions. A company needs a name, a domain, an owner, and perhaps a segment such as size or industry. A deal needs a value, a stage, an expected close date, and an owner. Use selects for anything you will filter on, like segment or lead source, and use free text only where a human has to read the answer.
The decision is flexibility versus structure. Default to structure for anything you report on, and flexibility for notes. A flexible data model, where you can add objects and attributes without a developer, lets you start small and extend later. If your CRM forces your business into fixed shapes, you end up storing real information in a notes field where no report can reach it.
4. Define pipeline stages
A pipeline is the path a deal takes from first conversation to signed contract. Write each stage as something that has happened, not a feeling. "Discovery call completed" is a stage. "Interested" is a mood.
Five to seven stages is enough for most teams. A typical set looks like this: Lead, Discovery call held, Demo held, Proposal sent, Negotiation, Won, Lost. For each stage, write one sentence that says what must be true for a deal to enter it, and one sentence that says what the rep does next. Pin those sentences where the team can see them, or put them in the stage's description inside the CRM.
The decision is whether to build one pipeline or several. Default to one. Split only when the sales motions differ in length, owner, or buyer, such as self-serve upgrades versus enterprise contracts. A second pipeline doubles your reporting work, so it has to earn its place.
5. Import your data
Gather every place customer information lives today: spreadsheets, a previous CRM, a shared inbox, a notes app, a billing tool. Export each to a CSV, then clean the files before you import them. Remove duplicates, fix obvious typos in company names, and make sure every row has a company domain, because domains are the most reliable way to match people to companies and to catch duplicates later.
Import companies first, then people, then deals, so each later file can link to records that already exist. Most CRMs let you map CSV columns to fields during import. Spend the time on this mapping screen. A wrong mapping imported across five thousand rows takes far longer to fix than it took to prevent.
The decision is how much history to bring. Default to everything from the last 18 months, plus every open deal and every active customer regardless of age. Old, cold records clutter every search and every view. You can always import them later if someone misses them.
6. Connect email and calendar
This step turns the CRM from a database someone has to feed into a record of what is happening. When each rep connects their inbox and calendar, emails and meetings attach to the right company and person on their own. Nobody has to log a call to prove it happened.
Connect accounts before anyone starts working in the CRM, so the first day of use already has history behind it. Tell the team what the sync shows to colleagues and what stays private, and set that before you connect, because trust in the tool depends on it. Most people accept shared email history when they know exactly what is shared.
The decision is the sharing level for email. Default to sharing that a conversation happened, and who was in it, with the whole team. Limit message content to the people on the thread unless your team has a reason to be more open. Call recording works the same way: choose where recordings and summaries live, and who can see them, before the first meeting is recorded.
7. Automate the first three chores
Look at what your team does by hand every day that a rule could do. Three jobs cover most of the early value: assigning new leads to an owner, creating a follow-up task when a deal changes stage, and posting a message in Slack when a deal is won. Build exactly those three and stop.
Assignment can be round-robin, or by territory, or by segment. Follow-up tasks should land on the deal owner with a due date, two days out by default. The won-deal message goes to a shared channel, which is the cheapest way to show the whole company that the CRM is where results appear. Workflows in Attio can run all three from templates, and Ask Attio can build a workflow from a plain-language description.
The decision is how much to automate before you have seen real usage. Default to three, and review the list after a month. Automations built on guesses tend to fire on conditions nobody expected.
8. Run a one-week pilot
Before you announce the CRM to everyone, pick two or three people who will use it for real work for a week. Choose a mix: one enthusiastic rep, one skeptic, and the CRM owner. Each pilot user works every deal and conversation through the CRM, and sends you a message every time something feels slower than their old way.
At the end of the week, sit down with the five questions from step one and check whether the CRM answers each of them without a manual lookup. Fix the gaps: a missing field, a confusing stage name, an import that attached people to the wrong company. Then run the full-team session.
The decision is the cutover date. Default to the Monday after the pilot ends. On that day, the old spreadsheet becomes read-only, and every new deal, contact, and update goes into the CRM. Keeping both systems alive for "just a few weeks" is the surest way to end up with neither one trustworthy.
Common mistakes
Adding fields before writing the questions they answer.
Letting every rep define "qualified" in their own way.
Importing every record ever created, including contacts from 2016.
Building automations before anyone has used the pipeline for a week.
Skipping the cutover date, so spreadsheets and the CRM live side by side.
What to set up next
Once the pilot is done and your team is working in the CRM daily, turn to reporting. Pick the three numbers you review every Monday, such as pipeline created, win rate by stage, and average deal cycle, and build one view for each. Review those views with the whole team each week, and the CRM becomes the place where pipeline conversations start.