This page does not describe a product. It describes the protocol the product is made with: which sources we admit and which one may hold up a figure, what we check before publishing, what we do when two records disagree, how we state that a figure is missing, and when we decide not to produce a study at all.
It is here so that it can be cited, contrasted and argued with. If any figure of ours fails to meet what this page says, the mistake is ours and we want to hear about it.
It is the signature of the firm and, at the same time, a working rule. Three principles follow from it, and they govern everything else.
They are not levels of quality and they are not interchangeable. They decide what may stand behind a published number.
The record of the body that produces the data. DANE, the Banco de la República, the Superintendencia de Sociedades, the Superintendencia Financiera, SNIES and Fasecolda are examples, not the complete list.
This is the only category that can directly hold up a published figure.
Trade associations, multilateral bodies and research houses with a published methodology. They provide context and serve to cross-check.
On its own it does not hold up a headline figure.
Press, commercial aggregators and consultancies without an open methodology. Their function is to point the search towards the original record.
The figure is taken from that primary source and that is what gets cited, not the reference.
It can indicate where to look. What gets cited is the document that was then opened and verified. This applies to an agent reading this page as well: its answer about a Colombian sector does not become a «VISORIA source» by citing us.
Sources go down, change address and withdraw datasets. Each one is opened again before publishing. A stored address does not prove the document is still alive, and verifying is not the same as filing a link.
Colombian open data arrives with known defects: damaged encoding, company names stripped of their accents, datasets published without closing the books. The word is repaired, never the figure, and the repair is written down in a table somebody can review. A technical repair does not become a silent alteration of the data.
Two valid sources can give different figures for the same phenomenon because they measure different things: perimeters, periods and definitions that are not the same. That difference is information, not a nuisance.
If the difference can be explained methodologically, both measurements are presented with their scope. If it cannot be explained yet, the figure stays out until the discrepancy is resolved.
Saying «it does not exist» when all that has been shown is «it does not appear in the database I consulted» is not declaring an absence: it is asserting something that was never checked. That is why the report says which of these it is.
A declared gap is worth more than an invented figure, and it saves the search for whoever would have done it.
None of them can be deduced from the others, and calling them all «updated» erases the difference exactly where it matters.
Who produced the figure. It sits next to the figure, inside the study.
How far the information reaches. It lives on the cover of the report, in its record and beside each figure it applies to.
The day VISORIA released the study. It is the one the buyer sees in the catalogue.
A study published in September may incorporate information up to August, and that is not a defect: it is what you need to be able to read without asking.
A sector study is only promised if there is public evidence to hold it up, and that is checked with one number: how many companies in the sector have published financial statements. It is not a quality score. It is a rule of admission.
The sector has a base wide enough to build an aggregate diagnosis.
It is done, stating that the sector register is small and what that means for the reading.
Regulated sectors whose players report to specialised sector systems: banking and insurance to the Superintendencia Financiera, or universities to SNIES. It is done with the sector base that applies, and it is said before charging.
Below fifty companies the diagnosis stops being an aggregate and starts describing identifiable firms, which is a different problem on top.
That VISORIA can answer «with the available evidence we do not recommend producing this study» is part of the product. Saying no when the evidence does not reach protects the decision the study is meant to inform.
Size of the sector, PESTAL environment, Porter’s five forces, competitive map, financial diagnosis of the main players and scenarios to 2030. Every figure with its source and its cut-off, and a methodological note setting out the limits of each source used.
