What rekeying between CRM, ERP and accounts costs, and how to stop it

Where does duplicate data entry usually happen?

Wherever two systems need the same facts and nothing passes them across. In a typical manufacturer or distributor, that means:

• New customers. Set up in the CRM by sales, then again in the ERP by accounts when the first order arrives, often with slightly different names and addresses.
• Quotes and orders. A quote built in a spreadsheet or the CRM, then re-entered as an order in the ERP.
• Invoices. Raised in the accounts system from details copied out of a closed deal or an order confirmation.
• Contact changes. A new buyer, a moved delivery address or a changed credit term updated in one system and not the other.
• Order status. Sales asking operations where an order is, then typing the answer into the CRM so the next person can see it.
• Reports. Figures exported from both systems and stitched together in a spreadsheet every month.

None of these feels like a project. Each is a few minutes, done by someone conscientious, many times a week. That is why the total is rarely known.

How do you calculate its annual cost?

Count the time first, using four figures you can find in a week:

1. how many records are typed into a second system each week
2. how many minutes one takes, including looking it up
3. the hourly cost of the person doing it, including employer costs
4. how many weeks a year it happens

Then add the time spent correcting the records that go wrong, if you can estimate how many do. The calculator below does the arithmetic. It starts empty on purpose: the answer is only as good as your figures, and any figure we suggested would be a guess about your business.

Two cautions. Time the work rather than estimating it, because people consistently underestimate tasks they do in short bursts. And treat the result as the floor, not the total: it counts the minutes, not what the mistakes cost.

How do rekeying errors affect the business?

A typing error in a second system rarely stays there. Depending on which field it lands in, it can mean:

• Invoicing errors. A wrong price, quantity or reference, which leads to credit notes, queries and slower payment.
• Delivery errors. An old address or the wrong contact, which costs a redelivery and the customer's patience.
• Duplicate customers. The same company set up twice, so its history is split and nobody sees the whole account.
• Unreliable reports. Figures that disagree between systems, so someone reconciles them by hand before every management meeting.
• Lost follow-ups. A quote or order that never reached the CRM, so nobody chases it.

These costs fall in different departments, which is why they are seldom added up. If you can count credit notes, redeliveries or duplicate accounts over a quarter, they are worth setting beside the time figure.

Which records should be synchronised?

Only the records that more than one team genuinely needs, each owned by one system. A common starting point for a manufacturer or distributor:

The rule that matters most is one owner per field. Where two systems can both change the same field, they will eventually disagree, and nobody will know which is right.

Should the integration be one-way or two-way?

One-way wherever you can, two-way only where the business genuinely edits the same record in both places.

• One-way (for example orders and invoices from the ERP into the CRM) is simpler to build, easier to test and hard to get wrong. Most of what a sales team needs from the ERP can travel one way.
• Two-way suits a small number of fields that both teams maintain, such as a customer's address or main contact. It needs rules for which change wins when both systems change the same field, and monitoring for the cases the rules do not cover.

A design that starts two-way for everything usually ends up with records that change back and forth, and a team that stops trusting either system.

What should remain manual?

Anything that needs judgement, and anything where an automated mistake would be expensive or hard to reverse:

• approving credit limits and non-standard terms
• agreeing prices outside the price list
• deciding whether two similar records really are the same customer before they are merged
• releasing an order that has failed a credit or stock check
• anything that happens too rarely to be worth building

Automation should hand these to the right person with the information they need, rather than make the decision for them. The aim is to remove the typing, not the thinking.

What has Flowbird done to stop rekeying?

Two published examples:

• Sage 200 and HubSpot. A UK distribution business kept the same customers up to date by hand in two systems, so every new customer added administration. Flowbird built a secure tunnel between the on-premises Sage 200 and HubSpot, bringing customer and company records into HubSpot automatically. The result was less administration and better data, because each customer is written once; the project won the Creative, Digital & Innovation Award at the Ashford Business Awards 2025. Read how it works.
• Pipedrive and Xero. A UK property services business retyped invoices from closed deals, which took time and made invoicing inconsistent. Flowbird built a Make scenario so one click on a Pipedrive deal creates the draft invoice in Xero, with the reference built to the finance team's rules. The result was draft invoices created from the deal, with fewer errors and consistent references. Read the case study.

Note what both have in common: the person still decides, and the system does the copying. In the Xero example, a person still chooses when to raise the invoice; the click replaces the typing.

How do you start?

1. List the double entry. Walk through one customer's journey from enquiry to invoice and note every point where something is typed twice.
2. Measure it. Time a week of the largest two or three, and use the calculator above.
3. Agree ownership. For each record, decide which system owns it and who needs to see it.
4. Fix the data first. Duplicate customers and inconsistent codes will be copied faithfully by any connection. Clean them before anything syncs.
5. Connect, one-way first. Start with the records that move one way and save the most time, using whichever tool suits the job: a native connector, Make, n8n, BPA Platform or custom code.
6. Monitor it. Someone needs to know when a connection stops, before the sales team notices missing orders.

Where to start

The CRM & ERP Opportunity Audit is a fixed-price, two-week review that maps one revenue-critical process end to end, such as quote to order, including where it passes into your ERP and where information is typed twice. It ends in a written report and a plain go or no-go. It does not need access to your ERP and does not analyse ERP data; where the process passes into the ERP, the people who run it walk us through it.

If you already know what needs connecting, talk to an integration specialist. For what this looks like by sector, see manufacturing and engineering, wholesale and distribution and packaging and printing.

For a worked example, see how a UK manufacturer stopped rekeying between six tools by running every order, from request to Xero payment, in one Pipedrive pipeline.