To compare data across several companies in the same group, you do not have to bring everything into a data warehouse first. When each company has its own database, you can put the same question to all of them, see the answers side by side and only then decide whether consolidation is worth it. What makes the comparison reliable is a shared set of indicator definitions, more than the technology you choose.
Many groups have grown through acquisitions or by setting up new companies, and each business kept its own system, its own codes and its own way of counting. In practice, the monthly comparison between companies is assembled in a spreadsheet by the finance team, arrives days after the close and only answers the questions someone thought of in advance. This article sets out the options, what needs to be agreed before comparing, and how to choose a tool.
One note before we start. This is about comparison for management purposes, the kind that tells the board which company sells more, where margin is slipping or which team collects cash faster. Statutory consolidation of accounts is a different process, with its own rules, and remains the job of the accounting team.
Why is it hard to compare the companies in a group?
The difficulty is rarely getting hold of the data. It is making sure that one company's figures measure the same thing as another's.
- Different systems. One company runs an ERP, another runs a different version of the same ERP, a third uses an in-house application. Tables and fields do not match.
- Different codes. The same customer or product has a different code in each company, and cost centres were designed by different people.
- Different definitions. "Sales" may mean invoicing with or without credit notes, counted by invoice date or by order date.
- Calendars and currencies. Companies in other countries may have different financial years or currencies.
- Access. Group management sees everything, but the management of each company does not need to see the others' data.
What are the options for comparing the data?
There are three common approaches. They are not mutually exclusive, and many groups use more than one.
| Approach | How it works | When it makes sense | Main limitation |
|---|---|---|---|
| Export and combine in a spreadsheet | Each company exports its data and someone combines it by hand | A few indicators, once a month | Slow, prone to copying errors and to several versions of the same file |
| Data warehouse or consolidated BI model | Data is copied into a common repository, with harmonisation rules | Many stable indicators and a data team to maintain them | A long project, and each new question may require changes to the model |
| The same question to each database | The question runs against each source and the answers appear side by side | New questions, exploration, and before deciding what to consolidate | Needs similar structures or aligned definitions, and does not fix different codes |
The third approach is the newest and the least known. It does not replace a data warehouse when a group needs one, but it lets you start comparing without waiting for it, and it helps you see which indicators deserve to be consolidated.
How does asking several databases the same question work?
A conversational analytics tool turns a question written in plain language into a database query, runs it and presents the result. With several sources connected, the same question is translated and run against each of them.
- The question is written once. For example, "what was gross margin by month over the last six months?".
- Each source answers on its own. The query is prepared for the structure of each database, and each returns its own result.
- Answers come together when they are comparable. If the results have the same columns, they can be combined in a single table and a single chart, with one line per source.
- The reader confirms what was measured. Before drawing conclusions, it is worth opening each source's query and checking that it measured the same thing.
Imagine a group with three distribution companies, each with its own database. The question about monthly gross margin runs against all three and returns a table with one row per company and per month, and a chart with one line per company. If one line sits well above the others, the first step is to check that company's query. It may be a real difference, or that company's margin may be calculated on standard cost while the others use actual cost.
This is the same care you should take with any AI answer about data. We wrote about it in how to check an AI's answers about your company data.
Which definitions need to be agreed before comparing?
A single page with the definitions of the main indicators prevents most arguments about numbers. These are the ones that most often differ between companies in the same group.
- Sales. Including or excluding VAT, with or without credit notes, by invoice date or by order date.
- Margin. On standard cost or on actual cost, with or without freight.
- Active customer. Anyone who bought in the last 3, 6 or 12 months.
- Days sales outstanding. Calculated on the month-end balance or on the average for the period.
- Period. Calendar month or fiscal month, and the date from which a month counts as closed.
- Currency. The reporting currency and the exchange rate used, when some companies are outside the euro area.
The companies do not all have to change the way they work. It is enough that, for each indicator being compared, it is written down how it is calculated in each one.
How do you choose a tool to compare the companies in a group?
These criteria help separate a polished demonstration from a tool that serves a group day to day.
- It connects to the systems you already have. Check which database engines and file formats are supported, and whether anything has to be migrated.
- It asks several sources the same question. And it shows each source's result separately, not just the total.
- It shows the query run against each source. Without that, there is no way to confirm that two companies were measured in the same way.
- It controls access by person and by table. The management of one company should be able to ask about their own business without seeing the restricted data of the others.
- It logs who asked what. So you know where every figure presented in a meeting came from.
- It works in your teams' languages. Important when the group has companies in other countries.
- It exports the result. As a table, a chart and a report, ready for the board meeting.
How does EngiAnalytics compare several sources?
In EngiAnalytics, the Engibots conversational data analytics platform, with several databases connected, the same question can be put to all of them at once, up to five. Each one answers on its own and, when the answers have the same columns, they come together in a single table and a single chart, with one line per source.
EngiAnalytics connects to MySQL and PostgreSQL databases, including ERP and CRM databases running on those engines, and to CSV and Excel files, without migration or system replacement. The SQL query behind each answer can be opened and copied, next to the table with the data returned. Access profiles define, table by table, what each person can query, and every question is logged, with who asked, the source, the query run and the outcome. Questions and answers can be in Portuguese, Spanish or English, and results can be exported as CSV, an image or a PDF.
A good place to start is to pick three indicators, write down the definition of each and put the same question to the companies in the group. If the answers are comparable, the board gains a view that used to depend on the month-end spreadsheet. If they are not, you learn exactly where the definitions diverge, which is already half the work of any consolidation project.
Frequently asked questions
Do you need a data warehouse to compare the companies in a group?
Not necessarily. For new questions and one-off comparisons, asking each database the same question is enough. A data warehouse pays off when there are many stable indicators and a team to maintain it.
What is the difference between comparing companies and consolidating accounts?
Consolidation of accounts is an accounting process with its own rules. Management comparison puts each company's indicators side by side to support decisions, and does not replace consolidation.
What happens if the companies use different systems?
The question can be the same, but the query is different for each source. That is why it matters to see the query run against each one and confirm it measures the same thing.
How do you stop each company seeing the others' data?
With access profiles defined by person and by table. A question outside that scope should be refused without exposing restricted information.
How many databases can EngiAnalytics compare in a single question?
Up to five. Each one answers on its own and, when the answers have the same columns, they come together in a single table and a single chart, with one line per source.