“Order number, delivery date, customer name, location, everything has to be replicated for every single item. On this project we had 150 machines. She had to copy it 150 times.”
That’s what the manager of an industrial equipment company said, in a recorded meeting. He’s not talking about software. He’s describing a person copying the same piece of information one hundred fifty times in one afternoon. And there’s no process error in this scene: the record was simply designed to hold an order, not a project with 150 machines. When the record doesn’t match the shape of the work, someone pays the difference with their own hands.
Of the 342 pains DualWin extracted from 169 meetings with 38 companies, 49 are about fields and records. It’s the second most cited theme, behind only spreadsheets, and the quietest one: nobody opens a meeting saying “our records are bad.” It shows up disguised as a report that doesn’t add up, a customer who got billed twice, a salesperson who couldn’t find the contact.
The record is the source of everything that comes after
The sales-by-region report only exists if someone filled in the region. The renewal reminder only fires if the contract date sits in a field, not inside an attached PDF. When the record is loose, the effect shows up three months later, in a dashboard nobody trusts.
A manager at an electronics manufacturer described exactly that: “This one doesn’t bring me anything. It shows how many deals were opened in the pipeline, what she marked as lost and why, and that’s it.” The dashboard wasn’t broken. It faithfully showed everything that had been recorded.
So the right question isn’t “which fields does the system offer.” It’s: which decisions do we make every week, and what information needs to be filled in to make each one?
The four symptoms of a broken record
The same information gets typed more than once. Customer name on the proposal, on the contract, on the work order, in the finance spreadsheet. Four entries, four possible versions of the same name.
There are fields nobody fills in. Every empty field teaches the team that filling it in is optional. Then the field that actually matters ends up empty too.
The “notes” field became the system. Agreed deadline, authorized discount, reason for the loss, all in a text box. It works for whoever wrote it and for no one else: it doesn’t filter, doesn’t count, doesn’t alert.
The same customer shows up twice. With “Inc.” and without, with an accent and without. Three records, three histories. It’s not carelessness: it’s the predictable result of there being no rule for searching before creating.
Two or more of these symptoms, and the problem isn’t your team. It’s the design.
A good field is a field someone uses to decide
The criterion fits in one sentence: a field only deserves to exist if someone uses what’s in it to make a decision, filter a list, or trigger an alert.
“Customer segment” passes, if you tailor your approach by segment. “Tax ID” passes, because the invoice depends on it. “Number of employees” guessed by the salesperson doesn’t pass: it decides nothing and still gives the false impression that the company knows its customer base.
Two practical consequences: a required field is a management decision, not a system one (too many required fields produce a database that’s full and wrong, which is worse than an empty one); and a field with no owner rots (every field needs someone who fills it in, when, and what happens if it’s left blank).
The minimum record
Resist the forty-field list. Start with the minimum and grow only with what practice demands. In most sales and service companies, there are three blocks:
Who they are. Legal name, tax ID, billing address, segment. The invoice and the contract require it, so it has to be right.
Who we talk to. Name, title, phone and email for each contact, kept separate per person. This is the block that suffers most in a spreadsheet, because a company has several contacts and a spreadsheet has one row.
What’s happening. Stage, value, expected date, owner, reason for loss. This is what feeds any sales forecast.
Conversation history, visit notes and proposal attachments don’t belong here: they’re records, not fields. Mixing the two is what makes a record bloat.
7 steps to rebuild your records without stopping the operation
- List the decisions, not the fields. Ten questions the company needs to answer every week. That’s the target.
- Mark where each answer comes from. A question with no field is a missing field. A field that doesn’t answer a question is a candidate to go.
- Write the definition of every field that stays. Is “delivery date” the promised one or the actual one? That ambiguity alone can ruin a report.
- Resolve duplicates before importing. Sort the database by the company’s tax ID, never by name.
- Define the creation rule. Before creating a new customer, search by tax ID. Without this, the database duplicates within weeks.
- Run in parallel for two weeks. If the new record answers the ten questions and the spreadsheet doesn’t, the spreadsheet leaves the stage.
- Schedule a review for 90 days out. An empty field after three months of real use isn’t a forgotten field, it’s an unnecessary one.
If you’re doing this alongside leaving the spreadsheet behind, the plan is in how to migrate from a spreadsheet to a CRM without stopping the operation.
How to know it worked
The same information gets typed only once. The notes field has emptied out, because the information now has somewhere to live. And someone answers one of the ten questions on their own, in under a minute.
None of this requires switching systems. It requires deciding what your company needs to know and no longer storing the rest just in case.
If it becomes clear that the problem isn’t the design, but information scattered across spreadsheets, email, WhatsApp and the heads of three people, the work is bigger than a field review. That’s what DualWin has done for more than five years, in more than 100 companies. Talk to a specialist.
Frequently asked questions
Which fields should exist in a customer record?
Only the ones someone uses to make a decision, filter a list, or trigger an alert. The minimum has three blocks: who the company is, who you talk to, and what's happening in the deal.
How do you avoid duplicate customer records?
With a rule agreed on with the whole team: before creating a new customer, search by the company's tax ID, never by name. Cleaning up existing duplicates is also done by sorting by tax ID.
How long does it take to reorganize customer records?
The seven-step plan runs in two to three weeks without interrupting the operation, because it includes two weeks of running in parallel with the current system.
