Online event
Our CSR Webinars
×
Check out our webinars, replays, and upcoming sessions to start taking action on your CSR strategy.
Sign up for the next sessionCheck out our webinars, replays, and upcoming sessions to start taking action on your CSR strategy.
Sign up for the next sessionDenys
Denys is a construction company specializing in, among other things, the installation of water network infrastructure, in France and internationally.
Client
The context: clients who ask for the carbon impact of every construction site
Denys needs to measure the carbon impact of its construction sites, in particular to meet the requirements of its clients. A single site involves a very large number of items: materials, machinery, vehicles, travel, energy, and time spent by the teams. The calculation covers several hundred, even several thousand lines to analyze.
Until now, it was done manually, from an estimating file and a file tracking actual expenses taken from the ERP. Assigning an emission factor to each line is time-consuming, and accuracy suffered as a result, for lack of the resources to process everything.
Against this backdrop, the main objectives were set:
- Produce a site’s carbon calculation faster, at the end of the project.
- Gain accuracy by processing every line, not just a portion of them.
- Provide results that clients can use directly.
- Have a traceable method that can be repeated from one site to the next.
Our approach: “method first, automation second”
Denys entrusted Altopi with designing this tool starting in September 2025. Rather than stacking a tool on top of an existing calculation, we first set out to understand and secure Denys’ method, then built a tool tailored to it.
1) Defining the calculation method
Objective: know precisely what is calculated, with which data and which assumptions.
We:
- analyzed the existing calculation files (estimating phase and execution phase);
- identified the difficulties specific to large construction sites: manual assignment of emission factors, accounting codes that are sometimes unreliable, expenses with no purchase order;
- defined the emission categories tracked per project: equipment rental and purchase, services and subcontracting, energy, transport and freight, business travel, commuting, insurance;
- set a data hierarchy: physical data (liters, kg, km) comes first, data reconstructed from a documented assumption comes next, and the amount in euros is used only as a last resort;
- based the method on the GHG Protocol, ISO 14064-1 and ADEME’s Base Empreinte®.
2) Building the tool: from your management data to the impact by category
Objective: turn management data that is hard to use into a clear view of the site’s carbon impact.
The tool is a secure online platform. For each site, the Denys team:
- creates the project and uploads three files: the site’s expenses extracted from the ERP, the list of suppliers, and commuting data;
- launches the calculation: the AI reads each line, assigns it to the right category and matches it with an emission factor from Base Empreinte®;
- views the results by emission category, as charts and a downloadable file.
A pilot phase on real site data validated the principle before roll-out.
3) Keeping the expert in the loop: consistency checks and traceability
Objective: a fast result that can still be defended in front of a client.
The AI suggests, a human validates. Before any result is issued, the tool enforces a check:
- mandatory review of the lines with the highest kgCO₂e-per-euro ratio;
- optional review of the 15 highest-contributing lines;
- for each line, validation (individually or in bulk) or a re-run of the calculation with additional keywords.
The result is only consolidated once consistency has been validated. Each calculation keeps the tool version, the emission factor database version, the source files and the validations performed. The share calculated from physical data and the share calculated from monetary data are shown, and the items that are not quantified are explicitly listed: the result is never presented as more exhaustive than it is.
4) Evolving the tool with the Denys teams
Objective: make the tool fit the company’s day-to-day operations, beyond the pilot.
Once the tool was in use, we evolved it together with the client:
- the calculation of commuting for site teams is built directly into the tool;
- Denys’ own activity codes are reproduced identically in the output;
- the result can be read in two ways: a summary view in the tool, with charts by category, and a detailed Excel export by activity code, which Denys then reworks to suit each of its own clients.
An industrialized carbon calculation: faster and more accurate
- A much faster calculation to produce at the end of a site, where hundreds of lines used to be processed by hand.
- A more accurate assessment, because every line is analyzed. In the pilot, 100% of lines were categorized automatically and more than 90% were matched with an emission factor.
- Results that clients can use directly, with a clear view of the impact by category.
- A tool adopted and adapted to Denys’ daily work: the client’s activity codes, Excel export by code and commuting calculation are deployed at the client.
- A traceable method: each factor applied is kept with its source, version and unit, making it possible to audit the calculation and update it.
- A reusable tool: import your management data, classify it with AI, convert it into impact with recognized factors, then have an expert validate the result. Calculating the carbon impact of a construction site is a first use case; the same approach can be adapted to other environmental reporting indicators.
