Most people's first experience of a CRM is being given a login and a two-hour walkthrough of a product with sixty screens, after which they go back to their phone and their notebook.
That is a training failure, not a software failure. The actual daily job inside a CRM is four actions long. This guide covers those four, plus the two weeks around them: what to do before anyone logs in, what a manager does each week, and the five mistakes that reliably kill a rollout in the first month.
If you are not yet sure what the software is for, read what is a CRM first — this article assumes you know.
Before anyone logs in
Write your stages down on paper
A stage is simply a step that a possible sale passes through. Most businesses need five or six:
New → Contacted → Qualified → Offer sent → Negotiating → Won / Lost
Get these right first, because everything else is built on top of them. Your reports count them, your screens group by them, and any automatic action the system takes is triggered by moving between them. Changing the stages later, with real customers already sitting in them, is the one job that is genuinely painful.
Two rules. Use the words your team already says, not sales-textbook language — if your people say "sent him the quotation", the stage is "Quotation sent". And give every stage a rule anyone can check. "Going well" is not a rule; within a month half your customers will be sitting in it and it will tell you nothing. "Meeting booked" is a rule, because anyone can look and see whether there is a date.
The commonest beginner mistake is too many stages. Twelve of them produce a screen nobody can read and weekly arguments about whether something is at stage six or seven.
Decide the handful of facts you always need
Alongside the stages, agree the small set of things you want recorded about every customer — the four to six facts that determine what you do next. Which ones those are depends entirely on your trade, and there is a longer treatment of choosing them well in our guide to what to ask and what to record, written for brokerages but true anywhere.
The only universal advice: pick fewer than you want to. Every question the system forces someone to answer adds work to every single call. Ask for too many and your team will either fill them in carelessly or stop logging calls at all.
Getting your data in
Import the customers you are actually working
Export everything you have — spreadsheets, phone contacts, the old system — into one file. Then, before importing, do the boring part:
- Remove duplicates. The same customer usually exists three times with three spellings.
- Fix the phone numbers. One consistent format, country code included. This is what lets the CRM recognise an incoming WhatsApp message as an existing customer rather than creating a new record.
- Delete what you cannot interpret. A row with a first name and a dead number is not data.
Import live customers first. Historical archives can come later or not at all — moving ten years of dead records on day one makes the new system feel exactly like the old mess.
If you are moving off an existing CRM rather than starting fresh, the migration has its own traps, and they are covered properly in the migration guide.
Put the open conversations in, with dates
Every live enquiry goes in with a stage, an owner, and a date for its next step.
This is the step people skip, and skipping it is why rollouts fail. A CRM that only contains leads from the day it launched shows you a sliver of the business, so the team keeps using the notebook for everything else — and now they are running two systems, which is worse than one.
Connect where enquiries arrive
Connect your website contact form, your advertising enquiry forms and your business WhatsApp number to the CRM, so that a new enquiry creates a customer card by itself.
Until this is connected, every enquiry costs a salesperson a retyping job, and retyping is the single most common reason a team decides the CRM is "extra work". You are asking them to do the same job twice for a benefit they cannot see yet.
The four actions: what the job actually is
Once it is running, here is a salesperson's entire daily interaction with the system.
1. Open your list for today
Not the whole database. The list of people you are supposed to speak to today, which the system built from the next steps everyone set yesterday and last week.
This list replaces memory. Until you have one, the customers who get called back are simply the ones who happen to come to mind — and it is the quiet, patient customer who never comes to mind. Of everything in this article, this is the change your team will feel first, on the first morning.
2. Log what happened
After each conversation, two or three lines on the customer's card: what they said, what they need, anything that changes the next move.
Write it for a colleague who has never met this person. That is the test. Not "called, no answer" — that is a status, not information. "Didn't pick up, third attempt this week, always answers evenings" tells the next person what to do.
3. Move the stage
If the conversation moved things on, move the card. If nothing changed, leave it where it is — a deal sitting still is information too. Nudging everything forward because it feels better to see progress produces a screen that looks healthy and a report you cannot trust.
4. Set the next step
One action, one date, one name. Then close the card.
The reason this matters is not organisation, it is silence. A customer with no next step makes no noise. Nobody notices them. They simply stop existing, and three weeks later they buy from someone who called.
That is the whole job. Four actions, a couple of minutes per customer.
What the manager does
The team's job is those four actions. Management's job is different, and if nobody does it the system decays into a filing cabinet.
Every morning, briefly. Look at what arrived overnight and whether it has been picked up. New enquiries sitting untouched are the most expensive thing on the screen.
Once a week. Read the pipeline stage by stage and ask two questions: what has not moved, and what has no next step. Those two lists are where deals die. Also watch for the opposite problem: a deal that moved through three stages in one week. That usually means somebody updated a week of old records in one sitting, rather than recording each step as it happened.
Once a month. Look backwards. Where did enquiries come from, how many became customers, and where in the stages do they most often stop. Month one of this is not very interesting. Month four is the reason you bought the software.
Always. Ask "is it in the system?" instead of accepting a spoken update. This sounds bureaucratic, and it is the only thing that works. The first time a manager takes the answer in the corridor and does not ask for it to be written down, the whole team learns that using the CRM is optional.
The five mistakes that kill it in the first month
1. Running two systems at once. The old spreadsheet stays open "just for now". Everyone does everything twice, resents it, and eventually picks one — and it will not be the new one. Choose a date after which the old file can be looked at but not added to or changed, and hold to it.
2. Customising before you understand it. Businesses spend the first month building custom fields and elaborate automations based on how they imagine they work. Use the plain version for a few weeks first. You will customise about a third of what you originally planned, and you will pick a better third.
3. Making everything required. Fifteen mandatory fields on a lead form produces exactly one outcome: fifteen fields of garbage, entered as fast as possible. Required fields should be the two or three you would genuinely refuse to work without.
4. Treating it as a database instead of a diary. The information the system holds is worth something; the next step is worth more. A CRM that is only a well-organised list of names is a contact book with a subscription fee.
5. Nobody is responsible for it. Someone has to own the thing: fix bad data, answer questions, decide what a stage means. Without a named owner, small problems accumulate for six months and then everyone agrees the software is bad.
What good looks like after a month
Three signs, none of which involve the software itself.
Your morning starts with a list instead of a decision. Nobody sits down and works out who to call today; the list is already there and the argument is only about the order.
Someone asks what happened to an enquiry from three weeks ago and gets an answer in ten seconds — from the record, not from a group chat and three people's memory.
The monthly number stops being built by hand. It is something you look up rather than something someone assembles on the last afternoon of the month, and because it was not assembled by hand, you start to believe it.
Those are early signs, not the full return. Putting an actual figure on that is the next article in the series: what a CRM changes, and when it does not pay off. If your open question is instead the practical one — logins, phones, who can see what — that is how to access a CRM.