Adaptive CRM. One place where the data is always right.
The customer file on the office computer, the stock list in the warehouse, the contacts on the salesperson's phone. Three different versions of the truth about the same business, drifting a little further apart every week. This piece leaves just one, and keeps it up to date on its own.
You enter data in one place only. Customers, enquiries, orders and availability all live in one database, and every screen reads from it. The second copy disappears, and that's where the discrepancies start.
The fields follow your process. Statuses, steps and names are taken from how you work now, including the ones that only exist in your line of work. That's why it's called adaptive.
Data is checked as it comes in. A code that doesn't exist or an impossible date is rejected straight away, instead of being found by a customer three weeks later.
This page covers a single piece of the system. The other pieces, and how we choose which one to start with, are on the services page.
What changes in practice
Simple questions get answered in ten seconds. How much has that customer bought in the last year, where does that enquiry stand, how much stock is really left: in many businesses today that takes three open files and interrupting two people, and the answer still isn't certain.
The second change is that nobody retypes anything. Data goes in once, where it originates, and everyone else sees it from there. When someone corrects it, the record shows who did it and when, so arguments about who was right are settled at a glance.
The problem with spreadsheets isn't that they're spreadsheets. It's that there are three of them, and nobody knows which one is current.
Why spreadsheets drift apart, and why nobody notices
In the spreadsheets businesses actually use, errors are the rule, not the exception. Across 85 operational spreadsheets examined in six independent studies, 94% contained at least one error, with error rates per cell of between 1.2% and 2.5%.
The figures were compiled by Raymond Panko of the University of Hawaii in What We Don’t Know About Spreadsheet Errors Today, table 2. His conclusion: “spreadsheet programs are not error-prone. People are error prone”.
An error rate of 1% per cell sounds small until you look at the size of real spreadsheets. In the same paper, a survey of two organisations found 65,806 spreadsheets on their internal servers, with an average of more than 4,000 formulas each. At that scale, the chance that the total at the bottom is wrong stops being hypothetical.
Rereading doesn't close the gap. In the research on human error Panko summarises, people catch about 81% of simple errors and only 66% of complex ones, so a third of the tricky ones stay in even after a careful check.
Your process decides the fields, not the other way round
An off-the-shelf product comes with its own way of working and asks the business to adapt: predefined fields, predefined statuses, predefined steps. It works when the business fits the model, and in Italy it hardly ever does.
Here we start from the opposite end. We look at how the work is done today, write down the real statuses an enquiry goes through, including the ones no product anticipates, and the database takes that shape. A producer who ships samples before an order gets a status for samples, because in their line of work that status exists.
The same goes for words. If you call something a pratica where others would say opportunità, the database says pratica. A system that makes people translate their own trade every time they type into it gets abandoned within three months.
What the system does, step by step
All data comes in through a single point, and every change leaves a log entry with who made it and when, so you can always see what it was before.
| Step | What happens | What you get |
|---|---|---|
| Importfrom today's spreadsheets | Existing files are imported with their history, duplicates included, and duplicates are flagged rather than merged at random. |
You start from what you already have, without spending a week retyping customer records. |
| Structurefields and statuses | The fields, statuses and rules are taken from your process during the analysis, and can still be changed later. |
The people who use it recognise their own terms on day one, so they actually use it. |
| Checkon the way in | Data is validated as it comes in: codes that must exist, dates that make sense, mandatory fields where they're needed. |
An error costs ten seconds now instead of a phone call to a customer three weeks from now. |
| Syncwith your programs | Where business software can be connected to from outside, the data flows across on its own; where it can't, a file is prepared for import. |
Nobody has to change their business software, and anyone still using the old program won't notice a thing. |
| Historywho and when | Every change is logged with who made it and when, and the previous version can always be looked up. |
Arguments about who changed what are settled in a glance instead of half a day. |
Checking at the point of entry, which hardly anyone builds
Nearly every project of this kind leaves the clean-up until the end: import everything, then spend a week fixing it. That's how the manual work you meant to get rid of comes back under another name, and why the 94% Panko measured refers to spreadsheets that someone had already checked.
Here the check happens at the door. An order without a price doesn't get in, a customer code that doesn't exist doesn't get in, a delivery date before the order date doesn't get in. The person entering it finds out immediately, while the document is still in front of them and they remember what they were doing.
Duplicates are handled the same way. Two records that look like the same company aren't merged automatically: they're shown side by side with the differences highlighted, and a person decides. Merging the wrong two customers is the kind of mistake you only discover once the invoice has gone out.
It's also why the database is the foundation the other pieces stand on. Payment reminders, customer reactivation and follow-ups only work if someone knows for certain who bought what, and when.
Italian SMEs are on the wrong side of the gap
The advantage isn't theoretical; it's been measured. In the 2025 European survey on technology use in enterprises, 65% of large companies used customer management software, against 25% of small ones: a gap of forty points on the same tool.
The gap is even wider for tools that analyse the data: 69% against 11% for business intelligence, according to Eurostat figures published on 20 May 2026. Businesses with their data in order know which customers are worth a call; businesses with three spreadsheets call in alphabetical order.
It's worth noting that the survey only counts enterprises with ten or more employees, so the smallest part of Italy's business landscape isn't even in the figure. The full argument about the gap is on the page about how we look at sectors.
Same mechanism, different name in every trade
The database is the same; the spreadsheet standing in for it today isn't. It's worth looking at your own case, because that's where you see what's being lost.
| Sector | The spreadsheet that holds everything today | Where you see it |
|---|---|---|
| Food and agricultureand export | The list of foreign distributors, with their last order, samples sent and agreed terms, updated when someone remembers to do it. |
|
| Hospitalityaccommodation and events | The calendar of enquiries, with dates on hold, quotes sent and deposits received, split between a spreadsheet and an inbox. |
|
| Restaurantsand bars | Suppliers, agreed prices and the regulars who book the big tables, all of which currently live in the head of the person who has worked the floor longest. |
What this piece doesn't do
It doesn't replace your business software, and it doesn't force a new product on people who already have their own tools. It works alongside them, takes the data and hands it back, and where a program stays closed it stops one step short and provides the file.
It doesn't decide for you who's a good customer. It keeps the data clean and shows it to you; qualification and reactivation are a separate piece, handled by customer qualification and reactivation. The reasoning behind keeping a person in the loop is on our page about the principles we build by.
It doesn't promise a return multiple. Published estimates for customer management projects vary enormously depending on how much the system is actually used, and anyone quoting a single number is picking whichever end of the range suits them. The number that counts is what keeping a second copy costs you today, and that can be measured in a week.
Questions and answers
Do we have to throw away our Excel files?
No, and it's not worth trying on day one anyway. Existing spreadsheets are imported with their history, and where someone has used them for years the sheet can stay on as a read-only view that updates itself from the database.
What goes away is writing the same thing in several places: you enter data in one place and read it wherever you like.
How is it different from a subscription CRM?
An off-the-shelf product comes with its own way of working and asks the business to adapt: the fields, statuses and steps are whatever its makers had in mind. Here the fields are taken from the process you already use, including statuses that only exist in your line of work.
It's also why we don't promise a return multiple: published estimates for customer management projects vary enormously depending on how far the system is actually adopted.
Who makes sure the data in it is right?
Data is checked as it comes in, which is the only point where checking is cheap. A customer code that doesn't exist, an impossible date or an order without a price is rejected straight away, instead of being discovered three weeks later.
Rereading afterwards is no substitute: in research on human error, people catch about 81% of simple errors and 66% of complex ones, so some always slip through.
Do our business software and other programs stay as they are?
Yes. The database works alongside what you have rather than replacing it: where a program can be connected to from outside, data flows across on its own; where it can't, the system prepares a file in the layout that program imports.
Nobody has to change their business software to make the rest work, and that rule applies to every piece of the system.
How do you measure whether it's working?
By counting the leads and follow-ups that no longer get lost, before and after, over the same period. The second number is how often the same data gets typed by hand in two different places, which we start by sampling over a week.
The third is how long it takes to answer a simple question, such as how much a customer has bought in the last twelve months. In many businesses today, that means opening three files and asking two people.
Notes on sources
- The 94% of operational spreadsheets with at least one error, the per-cell rates of 1.2% to 2.5%, the survey of 65,806 spreadsheets and the detection rates of 81% and 66% are in Raymond R. Panko, What We Don’t Know About Spreadsheet Errors Today, 2016, table 2 and section 2. The table covers 85 spreadsheets examined in six studies between 1995 and 2001, mostly audited financial models: the sample is small and old, and we say so. The direction is confirmed by every later study cited in the same paper.
- The figures of 65% against 25% for customer management software and 69% against 11% for business intelligence come from Eurostat's 2025 survey on ICT usage in enterprises, published on 20 May 2026. The survey covers European enterprises with ten or more employees, so it leaves out micro-businesses, and it compares large and small enterprises across the EU, not country by country.
- On financial return, we don't publish a multiple. The most widely quoted estimate, Nucleus Research's $8.71 for every dollar spent, comes from an analysis of selected cases, and later surveys by the same firm give much lower figures. It's a range that depends on adoption, not a promise you can carry over.
- This page doesn't report results achieved for a client, because this piece hasn't yet been delivered to a client. The tests mentioned are functional checks run in a test environment.
The other pieces in this group
Enquiries and contacts lost across channels and languagesFifteen minutes, with your case in front of us.
How many different files do you have to open today to find out how much a customer bought in the last year? If the answer is more than one, that already tells you the size of the problem. In fifteen minutes on the phone we'll look at it together and tell you where it makes sense to start, even if we never end up working together.
You'll speak to Mattia Esposito, who then builds the system: there's no salesperson in between. If you'd rather measure things yourself before talking, the Diagnostico (in Italian) is twenty questions and five minutes.