PHASE 04

ONE SOURCE OF TRUTH,
OR FOUR VERSIONS.

Almost every small business runs on several systems that each hold a partial version of the truth. The cost of that is paid in time, in errors, and in decisions made from the wrong number.

AI-First Operations › CRM & Data

The cost of four versions

Count the places a new customer's information gets entered in your business. For most operations the honest answer is three to six: the CRM, the accounting system, a spreadsheet somebody maintains, the project or job tracker, sometimes a shared folder with a naming convention, sometimes a text thread.

Two costs follow, and the second is much larger than the first.

The time cost is visible. Minutes per record, several records a day, across everyone who touches intake. It is boring, it is measurable, and it is the one people cite.

The error cost is not visible, and it is worse. Manual re-entry introduces mistakes at a rate nobody tracks. A transposed number, a misspelled name, an address that is right in one system and wrong in another. Then someone updates a phone number in one place and not the other three, and now you have four records that disagree and no way to tell which is current. Every downstream decision made from the wrong one is a compounding consequence.

The failure mode is not dramatic. It is a follow-up sent to a dead email, a job scheduled at a prior address, an invoice to a name that has changed. Individually small, and each one costs credibility with the customer.

Pick the system of record

Before any integration work, one decision has to be made and communicated: for each type of information, which system holds the truth.

Not which system has the most complete data today. Which one is authoritative going forward. When two systems disagree, that is the one that wins, and everything else gets corrected to match it.

A reasonable shape for a small business looks like this. Contact and pipeline information lives in the CRM. Financial records live in the accounting system. Job or project status lives in whatever runs delivery. Documents live in one structured place with a naming convention people actually follow.

Whatever you choose, write it down and say it out loud to the team. Half of all data problems in a small business are not technical. They are that two people each believe a different system is the real one, and both are maintaining theirs.

What actually needs to sync

The instinct once you start integrating is to connect everything to everything. Resist it. Every connection is something that can break silently, and a web of them is unmaintainable in a business with no technical staff.

Sync only what is needed in more than one place to do the work.

Usually worth syncingUsually not
Contact details and the identifier that links recordsInternal notes and commentary
Stage or status changes that trigger someone's workEvery field on every record
The address and the property or job referenceHistorical activity nobody queries
The agreed price where it drives invoicingDraft documents and working files

Direction matters as much as scope. Most connections should be one-way, out from the system of record. Two-way sync doubles the failure modes and creates conflicts that need rules nobody wants to design.

And decide what happens when a sync fails, because it will. Silent failure is the dangerous kind: for weeks everything looks fine and the data has been drifting apart the entire time. Somebody gets notified, and somebody checks.

The discipline that keeps it clean

Integration handles the mechanics. It does not handle the habits, and the habits are where most systems degrade back to where they were.

  • Enter it once, at the point of capture. If someone is taking notes on paper and entering them later, you have an unreliable step no integration can repair.
  • Required fields are the ones you will actually use. Every unnecessary required field makes bad data more likely, because people type anything to get past it.
  • One convention for names, addresses and stages, written down. Free text where a list belongs is how duplicates are born.
  • Someone owns data quality. A named person who periodically looks for duplicates and stale records. Without an owner it decays.
  • Fix the record, not the report. The instinct to correct a number in a spreadsheet before a meeting is how a second version of the truth starts.

Do not start by replacing everything

A common and expensive reaction to this page is to conclude the whole stack should be replaced with one platform that does it all.

Sometimes that is right. Usually it is a large project undertaken to avoid a small one. Migration is disruptive, all-in-one platforms are typically mediocre at several of the things they include, and you will have rebuilt everything before finding out whether the new arrangement actually fits how you work.

The cheaper sequence is to name the system of record, connect the two or three points where information genuinely has to move, fix the entry habits, and run it for a quarter. If a system is still clearly the wrong tool after that, you will know exactly why, and you will be replacing it with a clear specification rather than a hope.

Frequently asked

Questions people actually ask

Which CRM should I use?

The one your team will actually keep current. Fit to your process and ease of daily use matter far more than feature lists. A simple system used consistently beats a powerful one that half the team works around.

Do I need everything integrated?

No, and trying to is how small businesses end up with connections nobody can maintain. Sync what is needed in more than one place to do the work, one-way where possible, and leave the rest alone.

What is a system of record?

The system that holds the authoritative version of a given type of information. When two systems disagree, it wins and the other gets corrected. Naming it is a decision you announce, not a technical setting.

How do I clean up data that is already a mess?

Name the system of record first, then clean forward rather than trying to reconcile everything historically. Fix active records as they are touched, and set a date beyond which stale records are archived rather than maintained.

Is a spreadsheet ever the right answer?

For a genuinely small operation, yes, provided everyone agrees it is the system of record and it is not being copied. Spreadsheets fail at scale because they get duplicated, and the duplicate is what people work from.

Who should own data quality?

One named person with the authority to set conventions and correct records. It does not have to be a large part of their job. It has to be somebody's.

Make your next move

A year from now, what will you be glad you started today?

You don't need another promise that everything will be easy. You need something useful to learn — and a next step you're willing to take.