What is a CRM? A guide from the ground up

Picture a freelancer with twenty clients. They know every name, remember the last call, and keep a few notes in a notebook. That works fine, because one head holds it all.

Now add two colleagues and 300 contacts. Anna emails a prospect on Monday. Ben calls the same person on Tuesday without knowing. The notes sit in three inboxes and a spreadsheet nobody updated. When Anna goes on holiday, those deals stall, because the next step lived only in Anna's head.

A CRM fixes this by giving the team one shared record for each person and company. That is the core idea. Everything else in this guide builds on it.

People use the term CRM in three ways. It helps to keep them apart.

This guide is mostly about the software. Software without a process is just a nicer spreadsheet, though, so the process comes up often.

The tools changed a lot. The job stayed the same: remember who you talked to, and what happens next.

1. 1950s

   The card file

   The Rolodex put each contact on a card on a spinning wheel. It held names and numbers, but no history.

2. 1987

   Contact managers

   ACT! put contacts and call notes on a desktop computer. Great for one person. Sharing meant copying files around.

3. 1990s

   Enterprise CRM

   Siebel Systems, founded in 1993, sold large systems to big companies, and the term CRM caught on. Setup often took months and a team of consultants.

4. 1999

   CRM in the browser

   Salesforce sold CRM as a web subscription. No servers to install, and you paid per user per month. Most CRMs still sell this way.

5. 2014

   Small teams

   HubSpot released a free CRM, and tools like Pipedrive focused on small sales teams. A CRM stopped being something only large companies bought.

6. 2023 on

   CRM with AI

   CRMs started using language models to log calls, sum up history, and draft follow-ups. The records underneath still matter most.

Every CRM is built from a few kinds of records. Product names differ, but the shapes are the same.

The value is in the links between them. A deal links to a company and to the people involved. An email links to the contact who sent it. Open any record and you see everything connected to it.

The words you will see

Contact

One person. Name, email, phone, role, and the company they work for.

Company or account

An organization. It groups its people, deals, and history in one place.

Lead

Someone who might buy but has not been checked yet. Some CRMs make this its own record. Others use a status on the contact.

Deal or opportunity

One possible sale, with an amount, a stage, an owner, and an expected close date.

Pipeline

All open deals, grouped by stage.

Stage

One step in your sales process, like Qualified or Proposal.

Activity

Something that happened: an email, a call, a meeting, a note.

Task

Something that needs to happen next, with a due date and an owner.

Owner

The person responsible for a record. Every contact, company, and deal should have one.

Every record has a timeline. Emails, calls, meetings, and notes land on it in date order. Most CRMs connect to Gmail or Outlook and your calendar, so this happens without anyone typing it in.

The timeline answers the question everyone asks before a call: what happened last time? A new teammate can read six months of history in five minutes instead of asking around.

It is also where CRMs fail. If people have to copy emails in by hand, they stop doing it and the timeline goes quiet. Automatic capture is the first feature to test.

A deal moves through stages, from first contact to won or lost. Most teams use five to seven. A common set:

1. 1Lead
2. 2Qualified
3. 3Proposal
4. 4Negotiation
5. 5Won

Fewer deals reach each stage, and that is normal. The shape tells you where deals drop out, so you know which step to work on.

Name stages after what the buyer has done, like “Budget confirmed”, rather than your own actions, like “Email sent”. An email can be sent to a deal that is already dead. A confirmed budget means the deal moved.

A deal with no next step tends to stall. So a CRM keeps a next step on every open deal.

Each task has a due date and an owner. When you finish a call, you set the next task before you leave the record. Each morning the task list tells you who to contact today.

Many CRMs also flag what slipped: a deal with no activity for two weeks, or a customer nobody has spoken to since last year.

Every deal has an amount, a stage, and a close date, so the CRM can add them up. That answers questions a spreadsheet struggles with: what is closing this month, where deals get stuck, and who has too much on their plate.

The most common forecast is the weighted pipeline. Each stage gets a win chance. Each deal counts for its amount times that chance.

The forecast is only as good as the records. If stages and close dates are out of date, the number is wrong, and nothing tells you so.

There are two common ways to sort CRMs: by the team that uses one, and by the job it does. Most products mix several types.

By team

By job

A CRM sits in the middle of the tools you already use. Email and calendar feed it history. Marketing tools feed it new leads. Support and invoicing tools add what happens after the sale.

Some vendors sell one suite that does all of it. Others focus on one job and connect to the rest. A suite means fewer logins. A focused tool is usually cheaper and simpler, but you depend on the connections working.

Before you choose, list the tools that must connect. If your email or calendar does not sync, the timeline stays empty.

A spreadsheet is a fine place to start. It is free, flexible, and everyone knows it. Trouble starts when more than one person depends on it.

Signs you have outgrown the spreadsheet

• More than one person talks to the same contacts.
• A deal was lost because nobody followed up.
• You cannot say what is closing this month in under a minute.
• Someone left, and their notes left with them.
• You juggle 50 or more contacts and deals at once.

If none of these fit, keep the spreadsheet for now. Come back when one does.

Skip the feature lists at first. Every CRM has contacts, deals, and tasks. The differences that matter show up in five questions.

1. 1

   Does it match how you sell?

   A team selling one-off projects needs different stages than a team selling yearly subscriptions.

2. 2

   Does it log email and meetings by itself?

   Test it with your real inbox during the trial, not the demo data.

3. 3

   What will it cost in year two?

   Check the per-user price, which plan has the features you need, and what add-ons cost.

4. 4

   Can you get your data out?

   A full export in a standard format means you are never locked in.

5. 5

   Will your team use it?

   Give two people a week with it. If they drift back to the spreadsheet, that is your answer.

1. 1

   Week 1

   Clean, then import

   Pull contacts from your spreadsheet, inbox, and old tools. Remove duplicates before the import, not after.

2. 2

   Week 1

   Set your stages

   Write down five to seven stages. Name each one after something the buyer has done.

3. 3

   Week 2

   Connect email and calendar

   Turn on sync for everyone, so the timeline fills itself.

4. 4

   Week 2

   Give every record an owner

   A contact with no owner is a contact nobody follows up with.

5. 5

   Week 3 on

   Hold a weekly pipeline review

   Fifteen minutes a week. Walk through open deals, update stages, and set next steps. This habit keeps the data honest.

AI in a CRM helps most with the dull parts. It can log a call from a recording, sum up a long email thread, and draft a follow-up for you to review.

It can also rank. By reading the history on every record, it can point at contacts that have gone quiet and deals that look stuck.

It cannot fix bad data. If emails are not synced and deals have no owner, AI has nothing to read. The basics in this guide come first.