First 30 days with a new CRM: a week-by-week checklist

A CRM doesn't usually fail at installation but in the first month, when the team decides whether to use it or carry on with their phones and spreadsheets. This order avoids the usual failures: dirty data, stages that don't reflect the real work and no tracking of adoption.

Reviewed on 2026-10-10. BAI Business offers a CRM and publishes this article. The thresholds in this guide are indicative and our own, not third-party measures.

Before you start: clean the data

Importing data without cleaning it is the most expensive mistake: with duplicate contacts and empty fields, the team sees a mess on the first day and loses trust. Before importing:

• Duplicates by email. Two contacts with the same email are the same contact.
• Duplicates by phone. The same number can be written three ways; standardise the format.
• Minimum fields. Name, email or phone, and where the contact comes from. Without the source you won't know which channel brings the best customers.
• Repeated companies. "Acme S.L.", "ACME" and "Acme SL" are the same company.

OpenRefine is a free, open-source tool for working with messy data, which lets you fix inconsistencies by merging similar values (source: OpenRefine, accessed on 2026-10-10). Remember that some CRMs create duplicates on import if the file has no unique identifier such as the email; HubSpot states this in its import guide (source: HubSpot, accessed on 2026-10-10).

Week 1: real stages and import

Define the stages from the last deal you closed. Open your email or messaging, and trace what happened: when the contact arrived, when the first real conversation took place, when you sent the proposal, when they said yes, when they signed, when you got paid. From that, define between five and seven stages. Each stage needs a clear exit criterion: "proposal sent" is one because the customer has the document; "in negotiation", on its own, is not.

Import and check three things. The total number of records against your sheet, a sample of twenty random records to see that each piece of data is in its field, and a search for your own company and a large customer to detect duplicates.

Connect the website form. Make it send contacts to the CRM and not only to an email, with a source field and a notification to whoever looks after them. Test it yourself: fill in the form and check that the contact appears.

Week 2: minimal training and manual operation

A short session with three workflows and nothing else: log a contact by hand, move a deal to the next stage with a note of the context, and create a task. If each person can do them unaided, the training is enough; the rest is learned when it is needed.

Then use the CRM by hand for a full week, with no automations. The aim is to observe which stages cause confusion, which mandatory fields nobody fills in and what is still done outside the CRM. Take notes and don't correct anything yet.

Week 3: automate only what already works

• Welcome email when a new contact arrives through the form.
• Inactivity alert if a deal has gone several days with no recorded activity.
• Follow-up reminder for a proposal that was sent and has gone a couple of days without a reply.

Leave email sequences, automatic assignment and contact scoring for later: they need a month of real data to be set up properly. Before that they only automate noise.

Week 4: measure whether the team is using it

• Recent activity. What share of open deals has some activity in the last seven days. Indicative, ours: nearly all.
• Direct contacts. What share of new contacts went straight into the CRM, rather than being logged later from a phone. Indicative, ours: the vast majority.
• Stalled deals. How many open deals have gone more than 30 days without anybody moving them. Indicative: none. If there are any, either nobody is managing them or they should be closed as lost.

Common problems in the first month

• The team carries on with WhatsApp and spreadsheets in parallel. There is usually a task they do many times a day that is slower in the CRM, almost always quickly logging a contact who has just called. Simplify that specific workflow.
• The records don't match reality. If someone spoke to twenty people and the CRM shows eight activities, it is a logging problem. Define what counts as an activity and follow up daily in the first week.
• The stages don't reflect the work. If almost all the deals sit in one stage, the criteria are not clear. Review them together and remove the stages nobody uses.

Where BAI fits

According to the sheet for its CRM solution, in BAI Business you upload your customers from your spreadsheet and the repeated ones are detected automatically; whatever arrives through forms, bookings or the receptionist goes into the customer's record and opens its deal; deals move through your stages with their next action in your calendar; and your ready-made automations do what you repeat every day. You can see it on the CRM for your business page.

Frequently asked questions

How long does it take for a CRM to be ready to use?

It depends on the size of the team and how clean the data is. What matters is following an order: clean data, real stages, minimal training and, only after that, automations.

What should I measure at the end of the first month?

Three things: what share of open deals has recent activity, what share of new contacts goes straight into the CRM, and how many deals have gone more than 30 days without moving. The thresholds we give are indicative, ours, not third-party measures.

What do I do if the team keeps using WhatsApp and spreadsheets in parallel?

Find out which task they do many times a day that is slower in the CRM than before, usually quickly logging a contact who has just called, and simplify that specific workflow.

You may also be interested in

• Migrating from Excel to a CRM without losing data: how to import your data.
• Sales KPIs for small businesses: seven indicators to track in your CRM: what to measure once the CRM is working.

Sources

• Prepare your import file in HubSpot, accessed on 10 October 2026.
• OpenRefine, a free and open-source tool for cleaning data, accessed on 10 October 2026.