Poor customer data rarely looks like a data problem. It shows up as extra work in Sales, Service and Marketing. Here is a simple way to put a price on it and see what is worth fixing first.
Sales spends extra time preparing for follow-up.
Service ends up with duplicate cases.
Marketing struggles to build the segments it needs.
Finance spends another afternoon reconciling account data.
IT gets separate requests to “fix the system”.
Nobody calls it the same problem. But they may all be paying for the same one.
And, of course, AI is already in the conversation. What could it automate? How much faster could teams work?
But AI will use the customer context you give it. If people already check three systems before they trust that context, AI does not make the problem disappear.
It often starts with something much smaller
Say Sales wants better account prioritisation.
A customer group appears as one account in CRM, while ERP records purchases separately for each legal entity. Before planning follow-up, a salesperson exports the figures and matches customer IDs in a spreadsheet to see the group’s total purchases.
The request was for better prioritisation. Now the question is: Which customer record can we actually trust when we make a decision? And who decides which record is right?
This is where a “data issue” stops looking like an IT clean-up exercise. The systems often reflect years of business decisions and workarounds.
We often find the real problem in manual checks people do outside the system. A salesperson might keep a spreadsheet to work out which ERP records belong to the same customer group in CRM before trusting the sales figures. That workaround shows us where to look: how those records are linked and who is responsible for keeping those links accurate.
Dean Kostov
Salesforce Senior Solution Consultant, Agilcon
Who approved €89,000 a year for poor customer data?
Nobody did.
The cost sits inside work that feels normal: checking another system, finding an existing case, cleaning a list before a campaign.
Take three simple examples, using an illustrative €30 fully loaded hourly cost for a senior specialist.
Sales: Twenty salespeople have four meaningful customer conversations a day. Five extra minutes checking CRM, ERP, email or previous interactions adds up to 1,467 hours, or about €44,000 a year.
Service: A team of 20 handles 25 cases each per day. If only 5% need ten extra minutes because history is missing or work started elsewhere, that is 917 hours, or €27,500 a year.
Marketing: Eight campaigns a month, with six team-hours spent checking identifiers, exclusions, duplicates and segments, becomes 576 hours, or €17,280 a year.
Together: Call it €89,000 a year.
That €89,000 is really a measure of capacity tied up in workarounds. At this scale, that is more than the annual cost of one and a half senior commercial team members, freeing up significant time for customer conversations, follow-up and work that moves revenue.
The assumptions are deliberately simple. Replace them with your own people, volumes and hourly cost. And remember: this only prices visible manual work. It excludes delayed opportunities, customer impact, Finance reconciliation and other work created downstream.
The same problem may also be paid for several times. Marketing fixes the record. Sales checks it later. Service searches for the history. Finance reconciles it at month-end.
A better question than “How many duplicates do we have?” is: Where are we paying people to compensate for data they do not trust?
Find the hidden cost
Pick one customer process your team works around. We’ll help trace what sits behind it and what is worth fixing first.
And then AI arrives
Imagine two companies with similar sales teams. In one, AI prepares useful account context before a meeting because the underlying customer history and account data are connected.
In the other, the salesperson still checks ERP before trusting the AI summary.
The first company starts removing preparation work. The second keeps the five-minute workaround and adds AI on top.
Service faces the same issue. An AI-generated case summary is only as useful as the history behind it.
This is where the difference starts to show. Some companies automate the work. Others are still paying people to piece the context together by hand.
Start with the workaround, not every field
This is also where we would start at Agilcon.
Take one important customer process and sit down with the people who work it.
What do you check before deciding?
Where does the information come from?
What happens when two systems disagree?
For the account prioritisation example, that could mean agreeing which legal entities belong to each customer group, who maintains that structure and how ERP customer IDs link to CRM accounts.
The practical test is whether Sales can prioritise those accounts without rebuilding the picture in a spreadsheet.
Then put a number on the workaround:
Volume × extra time × employee cost
You will not capture every consequence. You do not need to.
One process may already show whether poor customer data is an IT annoyance or a business cost worth fixing.
If one process already feels more manual than it should, map it with us. We’ll help you find where the friction sits and what is worth fixing first.
Talk to our team
Book a free consultation and see how we can help you reach your goals.
