Do You Vet Your Business Partners? Financial data that can do it in just a few clicks
www.posiv.cz ǀ 39
FINANCE
year, roughly one and a half times as many as in 2020,
while only a few hundred ceased trading. The market
is expanding, but remains dominated by large
players: around half of all revenue goes to companies
with turnover exceeding CZK 1 billion, even though
they account for just half a per cent of businesses
in the sector. Even here, however, there has been
a shift in favour of smaller companies.
The development of the FinStat score is also interesting.
In 2021, the construction sector was slightly riskier than
the market as a whole, but since 2022 it has remained
below the overall market average, falling to 31% last
year, its lowest level in five years. This year, however,
I can see the first signs of a reversal – the proportion
of high–risk companies with a score above 70% has
risen from just under 8% to 9% in the sector, mirroring
the trend across the economy as a whole. At the same
time, after several years of modest growth, both profit
margins and EBITDA margins have begun to narrow.
Whether this is a one–off fluctuation or the beginning
of a more difficult period remains to be seen.
How much do you rely on artificial intelligence
when processing data, and where do you think this
is heading?
For many years now, our financial statements have
not been transcribed by people, but using OCR
technology, which we combine with today’s language
models. On top of that, we have our own validation
algorithms, so everything undergoes multiple checks
before we publish the data. Only if a document fails
these checks is it passed on to a person.
The key difference compared with the judicial system
is that the data does not remain in the form of scanned
PDF files that users have to download and go
through themselves. We extract it into clear tables
that can be filtered and analysed across hundreds
of thousands of companies at once.
Algorithms can already calculate the indicators
today. What I think will change is the interpretation.
At present, you still need a financial analyst who knows
how to read the numbers, but language models can
already provide a decent interpretation, and in five
to ten years they will be so good that they will replace
junior analysts and perhaps even more experienced
ones.
How do companies actually use your database
in practice, and what does it offer them?
People use it in two ways. They either check a specific
company – its profile, score, financial indicators
and insolvency history – and decide whether
it is suitable for what they have in mind, or they work
with the entire database using filters and rankings. They
can export hundreds of companies to Excel and then
import them into a CRM system, whether for sales
purposes or when looking to approach companies
with an investment opportunity.