How is Holmen using data?

Most visibly, to value its forest. Holmen reports its forest at fair value, and its 2025 annual report explains the data behind that number in unusual detail. It combines published price statistics with detailed transaction data on forest property sales. It fits "a regression model based on the parameters location (county or part of a county), volume of standing timber and site quality". It uses three years of transactions and cross-checks three methods, each of which came within 1.2% either side of the weighted value. The standing volume behind it comes from an external sample inventory carried out in 2019, rolled forward with harvest data and estimated growth. The report says little about operational data. AI appears only as a driver of energy demand.

What the usual answer says

Descriptions of Holmen's data use cover forest inventory, planning, production analytics, digital tools and sustainability reporting. The annual report confirms forest inventory and planning. It says little about digital tools. It says a great deal about valuation.

How the forest is valued

The 2025 annual report sets out the method. Valuations are "based on the amounts paid in relation to the properties' volume of standing timber broken down by geographical area". To obtain "a sufficiently large population, three years of aggregated transactions are used". Holmen uses price statistics "published by market participants" from two outside providers, and detailed transaction data from a third, to run the regression property by property.

The report explains why three sources matter. Because valuations combine them, "the risk of valuations being affected by data collection or processing errors is reduced". The result: "The three methods yielded a value within +/- 1.2 per cent of the weighted value at year-end 2025."

It also explains why three years. Using a shorter period would reduce "the quality and reliability of the valuations", because of "the greater impact of individual transactions and the small number of transactions".

The volume behind the value

The other input is how much timber there is. The report states that "the volume of standing timber is based on sample inventories designed to provide the most reliable information possible about this volume". The last of those was "carried out in 2019 by an external party". It had "a mean error rate of 1.4 per cent" for total standing volume. Since then, the change in volume is based on "harvest data and the estimated growth on which the current harvesting plan is based".

So the total rests on one external sample, with each year since adding measured harvest and estimated growth. The company also says it continuously inventories its holdings "to produce data that is used, among other things, to plan silviculture work and harvesting". Its land and forest figures are "Calculated based on Holmen's stand catalogue", its register of stands. The report doesn't say how the stand catalogue and the rolled-forward total compare. That comparison is the natural check on the growth assumptions, and the question an outside reader would ask.

A data risk the company names

In its internal control section, Holmen lists its main sustainability reporting risks as "calculation methodology and dependence on individuals for data generation". That is a data risk named in the company's own words.

A worked check

Say a forest owner's last external inventory was six years ago, with a small error. Each year since, it has subtracted measured harvest and added modelled growth. If modelled growth overstates true growth by an amount equal to 1% of standing volume each year, the roll-forward overstates volume by about 6% after six years, and the valuation with it. If the error is 1% of the growth estimate itself, the drift is far smaller, because annual growth is only a few per cent of volume. Comparing fresh stand measurements with the rolled-forward estimate for the same stands shows which it is. The figures are illustrative.

What it means for other forest owners

Holmen's disclosures describe a valuation built on several outside sources, so one source's errors don't carry through, with enough history to reduce noise and a stated reason for it. The methods are cross-checked and the spread reported. What they leave open is how the continuous stand data and the rolled-forward total are reconciled, which is the check any forest owner using a roll-forward can run on its own data.

The year-end report for 2025 shows the resulting book value of forest assets at each quarter-end.

When it doesn't apply

Owners who carry forest at cost face different tests. This is a reading of what the annual report explains; the report uses "digital" only three times, none about operations, and Holmen's operational data use in forests and mills may go well beyond it. The transaction data it relies on is also richer in some markets than others.

Quarri for forest management is built around how a forest operation runs, from the cruise to the settled account.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.