How to Map Business Data Flows
Ask three people in your company where a customer's phone number lives and you will get three answers. Sales says the CRM. Accounting says the invoicing tool. The technician in the field says it is in the text thread on her phone. All three are right, and that is the problem. When the customer changes numbers, one of those places gets updated, and the other two keep calling the old one.
A data flow map is the cure for that confusion. It is a plain record of what information your business holds, which systems touch it, which way it moves, how often, and who answers for it. You do not need special software or a consultant to make a useful first version. You need a few hours, a spreadsheet, and the willingness to ask people how they really work, not how the process document says they work.
This guide teaches a method an operations lead can run in a week. It ends with a sample map, a worked example, and notes on privacy.
What a data flow map is, and what it is not
A data flow map answers one question. For each important kind of information, where is it created, where does it go from there, and what happens to it along the way?
It is not a network diagram, which shows servers and cables. It is not a process flowchart, which shows who does which task. It is not a data dictionary, which defines every field. Those are all useful, and the map can point to them, but mixing them makes a drawing nobody can read.
Think of it as a street map for information. You are not recording every house, only the roads between neighborhoods and who maintains each road.
The five attributes to capture
The method rests on five attributes, and every row of your map should have all of them. If you cannot fill one in, that gap is itself a finding.
- Entity. The kind of thing the data describes, such as a customer, an order, an invoice, an employee, a product, a vendor, a support ticket.
- System. Where a copy of that entity lives, including the unglamorous places like spreadsheets, email inboxes, shared drives and paper files.
- Direction. Whether data moves from this system to another, from another into this one, or in both directions.
- Frequency. How often it moves. Real time, hourly, nightly, weekly, or only when a person remembers.
- Owner. The person who is accountable for the data being correct in that system, and who is called when it is wrong.
Two terms are worth defining once. A source of truth, sometimes called a system of record, is the one place that is treated as correct when copies disagree. A sync is any mechanism that copies or updates data between systems, whether an integration tool, a script, a scheduled export, or a person retyping.
Step 1. Choose the entities, and keep the list short
Start with the information the business cannot run without. For most small firms that list is five to eight entities.
- Customers or clients, and their contacts.
- Quotes and orders, or projects and engagements.
- Invoices and payments.
- Products, services or inventory items.
- Vendors and purchase orders.
- Employees or contractors.
- Support requests or service tickets.
Skip the long tail for now. You can add marketing audiences, equipment logs or meeting notes in a second pass. A first map that covers six entities well is more useful than one that covers thirty badly.
Step 2. Inventory the systems, including the informal ones
List every place those entities live. Walk through the company by role and ask a simple question. "When you need to look up a customer, where do you go?"
Expect to find the following.
- Business applications, such as accounting, CRM, scheduling, e-commerce, payroll, help desk.
- Spreadsheets, especially ones that were meant to be temporary.
- Email inboxes and shared mailboxes.
- Messaging apps, including personal phones.
- Shared drives and cloud folders.
- Paper files and whiteboards.
- Third-party platforms where your data lives but you do not host it, such as a payment processor or a website form tool.
Do not judge what you find. People build shadow systems because the official ones did not fit the work, and the point of the exercise is to learn that, not to assign blame.
Step 3. Draw each flow as one row
Now create a table with one row for each flow, meaning one entity moving between one pair of systems. Columns are entity, source system, destination system, direction, mechanism, frequency, owner, and a notes column for anything strange.
Mechanism is worth a column of its own. The usual mechanisms are the following.
- Manual entry, where a person retypes or copies and pastes.
- File export and import, such as a CSV that someone uploads.
- A native integration, which is a connector offered by one of the vendors.
- An integration platform or automation tool.
- A custom script or API call.
- A webhook, which is a message one system sends automatically to another when something happens, such as a new payment.
The mechanism tells you how fragile the flow is. A webhook with a signature check and retry logic is sturdy. A person who copies totals into a spreadsheet on Friday afternoons is not, and the map should say so without judgment.
A sample map
Here is a short, hypothetical map for a ten-person professional services firm. The systems and the details are invented for illustration.
| Entity | Source | Destination | Direction | Mechanism | Frequency | Owner |
|---|---|---|---|---|---|---|
| Client contact | Website form | CRM | One way | Native integration | Real time | Marketing lead |
| Client company | CRM | Accounting | One way | Manual entry | When a deal is won | Office manager |
| Invoice | Accounting | Client email | One way | Built-in sending | On approval | Bookkeeper |
| Payment | Payment processor | Accounting | One way | Bank feed and import | Daily | Bookkeeper |
| Project | CRM | Project tool | One way | Manual entry | When a deal is won | Operations lead |
| Time entries | Project tool | Accounting | One way | CSV export | Weekly | Operations lead |
| Employee | HR spreadsheet | Payroll | One way | Manual entry | At hire and change | Office manager |
| Client contact | CRM | Personal phone contacts | One way | Manual copy | Ad hoc | Each consultant |
Even in a small table like this, patterns jump out. Three flows rely on a person remembering to type something. One entity, the client contact, has a copy on personal phones that nobody owns. Client company data is entered twice, once in the CRM and once in accounting, which is exactly the setup where spelling and tax IDs drift apart.
Step 4. Mark sources of truth and conflicts
For each entity, circle the one system that should be the source of truth. This is a decision and not a discovery, and sometimes it is a debate. A workable rule is that the system where the data is first created, or where its accuracy matters most, wins. For client billing details, accounting wins. For a client's preferred contact person, the CRM wins.
Then look for places where two systems both accept edits to the same field. That is a two-way flow, and it is where conflicts live. Ask what happens if both are edited within the same hour. Some tools let the last write win, which can quietly erase a correct change. If you must run a two-way sync, define which system wins per field and record it on the map.
Step 5. Score the risks
Walk through the table and mark each row against a handful of questions.
- Is the flow manual?
- Does it run less often than the business needs? A nightly sync is fine for a monthly report and wrong for stock levels.
- Does only one person know how it works?
- Is there a way to tell when it fails?
- Does it carry personal, financial or health information?
Rows with several yeses are your priorities. They are the places where an error is most likely, hardest to notice, or most costly.
A worked example with illustrative numbers
Here is a hypothetical scenario. A small distributor takes orders by email and phone. An office clerk types each order into the accounting system, and a warehouse lead types it again into a spreadsheet used to pick and pack. During the mapping exercise, the team times the work and finds that each order is typed twice, at roughly four minutes per entry. At 25 orders a day that is about 200 minutes of typing, or a little over three hours, of which half is the duplicate. Over a 22-day month, the duplicate typing is roughly 37 hours.
More important than the hours is the error exposure. The map shows two copies of each order with no sync and no owner for reconciliation. In this illustration, a spot check of one week of orders finds three where quantity or ship-to address differ between the two copies. Those are the flows worth fixing first, and the map makes the case without argument. The numbers here are invented to show the method and are not from a real company.
Privacy and security notes for your map
A data flow map is also the first step in any serious privacy effort, because you cannot protect or delete what you do not know you have. A few practical points.
- Flag the personal data. Mark rows that include names, contact details, government identifiers, financial account data, health information, or data about children. Treat these as a higher tier.
- Note where data leaves your control. Every processor or vendor that receives personal data is a recipient. Record who they are and where your contract with them is stored.
- Record the purpose and the retention period if you know them. Under many privacy laws, you should be able to say why you collect something and how long you keep it. Requirements differ by country and by state and sector, and nothing here is legal advice, so check with counsel about what applies to you.
- Track cross-border flows. If a vendor stores data in another country, note it.
- Watch the informal copies. Spreadsheets emailed to personal addresses and exports left in download folders are common leaks. The map should make them visible so someone can decide what to do about them.
- Restrict the map itself. A complete data flow map is a useful guide for an attacker. Keep it in a place with controlled access, and do not paste credentials, API keys or passwords into it. Reference where secrets are stored without writing them down.
If your business handles health records, financial accounts, or other regulated data, the map is a stepping stone to a more formal assessment. A short list of the regulated flows is a good thing to bring to a lawyer or a compliance advisor.
Keeping the map alive
A map that is out of date is worse than none, because people trust it. Build in light maintenance.
- Give the map one owner, usually the operations lead.
- Update it whenever you add, replace or cancel a system. Make that a line item in your software purchase and offboarding checklists.
- Review it every quarter in a 30-minute meeting. Ask which flows changed, which broke, and which are still manual.
- Store it in a shared spreadsheet or a simple document, with a version date at the top. Simple tools survive staff changes better than elaborate diagrams.
If you operate in Puerto Rico, add one more column or note for resilience. Which systems are cloud hosted, which depend on the local office network, and what happens when power and internet are down for a few days after a storm? A flow that depends on a single desktop computer in the office is a risk the map should show plainly.
What you can do on Monday morning
- Pick six entities and write them down.
- Spend one hour interviewing two or three people, asking where they look up each entity and where they put new information.
- Build the table with the columns entity, source, destination, direction, mechanism, frequency and owner.
- Mark each entity's source of truth, and circle every row that is manual, owned by nobody, or carries personal data.
- Share the draft with the team and let them correct it. They will fix it in minutes, and they will start thinking in flows.
After a week, you will have a document that explains how your information really moves, and a short, ranked list of what to fix. Ozcorp InfraOps helps small teams turn that list into working integrations when they want a hand, with InfraOps Base starting at $500 per month and vendor and license charges billed separately. Either way, I hope the map saves you from the next argument about which phone number is correct.
Discuss your infrastructure requirements
Review the current service offer or schedule a conversation about the scope you need.
Stay in touch
Follow Ozcorp
Leave your details for occasional notes about new articles and Ozcorp updates.