Four systems, none of them agree: connecting ERP, CRM and spreadsheets without replacing the ERP

The director asks what the pipeline is worth. Someone opens three files to answer. Sales has one customer list, accounting has another, and the ERP has a third. The monthly report is assembled by hand, and by the time it's ready, half of it is out of date.

This is the normal state of most companies, large and small.

How disconnected companies are

• The average organisation in the MuleSoft survey runs 897 applications (MuleSoft, 2025).
• 26% of company data is considered unreliable. Only 49% of leaders can get insights from data in time (Salesforce, 2026).
• Three in five companies don't know how much they spent on digital projects or what they got back (EY-Parthenon, 2022).

Smaller manufacturers aren't exempt. Before joining the UK's Made Smarter programme, 97% of participating companies already had some digital technology. They had adopted it piece by piece, and the systems needed integrating (Ipsos for DBT, 2025). Data and systems integration became the programme's most supported category, ahead of new machines.

Why it costs more than time

MIT's Center for Information Systems Research studied 721 companies. Those with fragmented systems — what the researchers call "silos and spaghetti" — grew revenue 12.8 percentage points below their industry average, with margins 10.6 points lower. Companies with their own integrated internal platform grew 12.4 points above average (MIT CISR, 2025).

That is correlation, not proof of cause. But it is a large gap, and it points the same way as everything else.

Why buying BI rarely fixes it

A BI tool draws charts from the data it receives. When the CRM says one thing and the ERP another, the dashboard shows one of them, and the meeting turns into an argument about which number is right. The problem is rarely the chart. It is the data underneath.

Research on data-driven decision-making found productivity 5 to 6% higher in companies that decide on data (Brynjolfsson, Hitt and Kim). The effect appeared only where decisions actually changed. A dashboard nobody looks at does nothing.

One data model, in your words

What works is less glamorous.

1. Agree the vocabulary. What is a customer? An order? A won deal? Each department has its own answer. Write one down together.
2. Decide which system owns each fact. The ERP owns invoices and stock. The CRM, or the new system, owns the pipeline. Nobody keeps a second copy.
3. Connect, don't replace. Existing systems stay and feed the model.
4. Build the screens on top. The director's dashboard, the catalogue with its real structure, usage statistics that show what people actually do.

What the director's screen looks like

For the panel maker in Quoted by hand. Then not., the director had no view of the pipeline before. Now there is one screen: revenue by month, the funnel with amounts at each stage, average deal size, conversion, time to quote, top managers by volume, and a list of orders that have been in one stage for more than 14 days, each one clickable.

The director sees the business without asking anyone what's in it. The ERP stayed.

Where to start

Write down the five numbers you need every Monday. For each one, note where it comes from today, who assembles it and how long it takes. That list is the specification for the first screen.

Related: Your business runs on a spreadsheet · Build or buy? A checklist for mid-sized companies