CRM implementation that survives contact with your sales team

What this actually solves

Two very different businesses arrive at this page. The first has no CRM. The owner cannot see what is coming in this month, asks the sales team, gets told everything is fine, and then misses the number. Someone resigns and takes the customer relationships with them, because the relationships lived in that person's phone and never in the company. Two people call the same customer in the same week. Nobody can say for certain whether a quotation was ever sent. None of that is a software problem yet — it is the cost of a business running inside individual heads.

The second already bought a CRM, two years ago, from someone who configured it and left. Today it holds thousands of contacts, many of them duplicated because "Ali Khan" and "ali khan" were imported as different people. The pipeline is full of deals that should have been closed last year. One salesperson uses it, another uses it occasionally, and a third says openly that their own spreadsheet is better. The owner's conclusion is that the software does not work. The software works fine. What was never built was the agreement underneath it — what a deal stage actually means, who is responsible for moving it, and what the salesperson gets in return for typing. They bought a licence and assumed a system would arrive with it.

Both problems have the same fix, and it is mostly not technical. It is a decision about how the work runs, written down, then configured, then taught.

How the work runs

Software is not opened until the fourth step. That ordering is deliberate — what the system gets configured to do is decided in the discovery sessions and in the state of your data, never in the tool, so opening the tool first only builds the wrong thing faster.

1. Discovery, with the people who do the job. Two to four sessions. The owner comes first and explains how the company should run; the rest of the time is spent with the salespeople and the admin who know how it does run. The system is built on the second answer. One question does most of the work — show me one real deal, from the day the enquiry arrived to the day the money landed. Two hours on a small business, up to two days on a large one.
2. Process map and platform decision. The current process is written out step by step with every point where something stalls or gets lost marked, then the proposed version beside it. You see your own business on one page before anything is bought. The proposal states what is included, what is explicitly not, and what we need from you - files, decisions, and an hour of each user's time. Two to three days.
3. Data review, before anything is built. Every spreadsheet, every old export, the phone contacts. How many rows are real, how duplicates will be identified, what format the phone numbers are in, which records are simply dead. This is also where we agree what will not be brought across — usually the last two years of active data comes over and the rest is archived to a file you keep. One to three days.
4. Build, shown to you weekly. Objects and fields first, then the pipeline, then roles and permissions, then templates, then three to five automations, then the dashboard. Each week there is a short call showing only that week's piece and asking one question - does this match your real work. Small mistakes stay small that way. One to three weeks.
5. Migration, in stages. Export to a copy, never the original. Clean the formats, dedupe on email then phone, produce a field mapping sheet and get it approved in writing. Then twenty rows imported and checked field by field on screen, then two hundred, then the whole file, then a count reconciled against the source and handed to you. Every imported record carries a dated import stamp. Three days to two weeks, depending entirely on the state of the data.
6. Testing with real work. Each person who will use the system is given a genuine task of their own — today's enquiries, entered the old way and the new way — and asked what broke. Fixes are sorted into blocking, post-launch, and next phase. Two to three days.
7. Training, cutover and hypercare. Each role is taught only the handful of things they personally do, on their own records rather than demo data, in short recorded sessions, plus a one-page printed sheet. Go-live is never on a Friday. The old system goes read-only rather than being deleted. Then daily check-ins for a week and responsive support after. One week, plus two to four weeks of hypercare.

What moves the price

The range is $1,500–$25,000 and three things decide where you land in it.

How many people have to change their habits. Five salespeople on HubSpot is a straightforward configuration and a handful of training sessions. Fifty seats across several teams means role design, permissions, and training that runs for weeks — and training hours, not licence count, are what actually scale the cost.

The state of the data. A clean export from a previous CRM is a day. Four thousand rows assembled from spreadsheets, phone contacts and an old mailing list export, with inconsistent phone formats and no reliable unique field, is weeks — and it is honest work, not padding.

How many other systems have to be joined to it. A CRM standing alone is one job. A CRM that has to receive website forms, push signed documents into deal records, and reconcile with an accounting package is several, and each connection has to be tested for what happens when it fails rather than only when it works.

What usually goes wrong

The spreadsheet gets a glance and an import. This is the most expensive mistake in the category, because it does not fail loudly. Four thousand contacts go in, everything looks right, you are pleased. Three weeks later somebody notices that phone numbers landed on the wrong records, or that hundreds of companies exist twice, or that every "last contacted" date is the day of the import. By then there are three weeks of real work sitting on top of the bad data — new calls, new notes, new deals — so the bad records cannot simply be deleted. Staged imports and a dated import stamp on every record are what make that recoverable later.

Deal stages get named after feelings. "Negotiation" and "Warm" mean something different to every person on the team, so cards stop moving, the forecast becomes fiction, and within a month the owner is back to asking people verbally. Every stage needs one condition that is verifiably true or false — the quote has been sent and the PDF is attached to the record - along with who is allowed to move it.

Thirty custom fields on day one. Spread across contacts, companies and deals it feels thorough, and it is how adoption dies. Three months later twenty-five of them are empty, the contact record looks abandoned, and the salesperson who was reluctant to type in the first place learns to skip the ones that do matter — including the field a stage condition depends on, which is how the pipeline stops being true. Every field is designed for the least willing user on the team, because that is who decides whether the CRM survives.