Our method

A measurement is only worth the action it makes possible

Installing an analyser above a conveyor takes a few hours. Making its data reliable, readable and useful for a decision takes support that we provide at every site, with the local authority and with the operator. Here is how, at three equipped sites.

SPL Trigironde
Gironde, residue line
SPL Allier Tri
Allier, infeed and residues
GreenRécup’
Val-d’Oise, infeed and 40-180 mm fraction

Three principles

What guides each deployment

Adapting the analysis to your characterisation grid

The measurement uses your reference frameworks (characterisation grid, mesh size, categories), so that its results can be compared with your historical data.

Proving before claiming

Each system is calibrated against cross-checked manual characterisations. Its limits are set out from the start, not discovered along the way.

Remaining a third party

NextWaste does not operate any site and has no capital ties with equipment manufacturers; our commercial partnerships are declared. You remain the owner of your data and choose who to share it with.

Before any system

First the question to answer. Then the system.

Before choosing a sensor or a camera, we clarify the decision expected, the data already available and the level of reliability required. A targeted campaign, monitoring at a few points or a continuous system can meet different needs.

Owner’s advisory assignment

For a local authority preparing a project or a contract: we scope the need, the trials and the validation of the results, without bidding for the procurement that follows. AI owner’s advisory for local authorities

Integration and use of the data

For a site that wants to measure a waste stream: we install a suitable system and support the use of its data. This is the path described in the five steps below.

Five steps

From installing the analyser to the investment decision

These five steps describe a continuous analysis deployment: what NextWaste does, then what actually happened at the sites.

  1. Installing the analyser without disrupting operations

    The analyser is mounted above an existing conveyor, on a simple support, near a power supply and a network connection. We choose the useful measurement points with the site, rather than equipping every line.

    On site

    Trigironde

    A support made by a local craftsman for less than €500, fitted on the residue conveyor in under two hours.

    Allier Tri

    Two measurement points, the infeed and the residue line before the sorting cabin. The internet connection took longer than the installation.

    GreenRécup’

    Two positions: the main infeed, and the 40-180 mm fraction where uncaptured recyclables hide.

  2. Measuring with the site’s grid, not ours

    The analyser identifies each visible object and classifies it according to the characterisation grid the site already uses. During calibration, time-stamped manual characterisations are compared with its readings over the same period, until the gap is reduced to a minimum.

    On site

    Trigironde

    “Our approach is to know which recoverable materials are in our residues, by asking NextWaste to follow the Citeo standard exactly.”

    Olivier Guilmois, Managing Director

    Fines are defined according to the site’s mesh size, under 5 cm. Calibration was completed in two to three weeks.

  3. Building a shared reading with the operator

    Two people do not always classify an object in the same way, and manual characterisation has its own variability. Before automating, we bring the local authority and the operator together to agree, category by category, on the same reference points.

    On site

    Allier Tri

    “We went through the whole classification grid together; we took a short afternoon with the operator to agree on where each object goes.”

    Jackie Renaud, Managing Director

    Two months, cross-checked characterisations included, between installation and usable data.

  4. Making the data readable day to day

    A site is not meant to become a data analysis department. At the start, we produce regular reports that isolate the useful signals, and we help interpret them. Recurring drifts then become simple alerts.

    On site

    Allier Tri

    Almost daily reports revealed fluctuations at the infeed that were invisible in monthly averages. Above a throughput threshold, an SMS alert notifies the SPL and the operator.

    GreenRécup’

    Time-stamping and automated reporting: a characterisation per waste supplier when their deliveries can be tracked to the line, and production teams alerted to changes in composition.

  5. Informing the investment decision

    Measurement helps decide: it can justify equipment as well as postpone its purchase. When an investment takes shape, we document the return assumptions, presented as such until they are measured, and study coupling the analyser with extraction equipment.

    On site

    Trigironde

    Coupling the analyser with an ejector, which would send packaging back to optical sorting, is under study.

    Allier Tri

    15 to 20% recoverable materials in the residues, mainly fibre-based: no immediate investment on this line.

    GreenRécup’

    The 40-180 mm fraction quantifies the recyclables lost, to decide between settings, infeed procedures and investment.

Two sorting facilities, one question

Should you invest in the residue line?

Trigironde and Allier Tri have been monitoring their line continuously for several months. The measurement is not used to rank the sites: it gives each of them evidence to decide. That is what a measurement should do: it must also be able to conclude against investing.

SPL Trigironde

Saint-Denis-de-Pile, Gironde, 34,000 t a year

“There are always investments to make in a sorting facility; you have to choose the most relevant ones.”

Olivier Guilmois, Managing Director

What the measurement provides

The waste stream, family by family

The materials present on the residue line, identified upstream of manual sorting: evidence for comparing recovery options.

Where the site stands

No investment has been decided. Additional recovery options are under study, and their value remains to be confirmed.

Read the Trigironde case study

SPL Allier Tri

Chézy, Allier, 20,000 t a year

“Perhaps we should not invest in this line straight away.”

Jackie Renaud, Managing Director

What the measurement showed

15 to 20%

recoverable materials in the residues, mainly fibre-based: too much, but not enough to justify equipment on this line alone.

Where the site stands

Adjustments will instead be sought upstream, in the next operating contract.

Read the Allier Tri case study

Methodological requirements

What we make explicit in every project

  • What the system provides: identification of visible objects, counting, surface area, volume, mass estimate or weighing.
  • The categories used, fines, unknowns and how ambiguous objects are handled.
  • The difference between calibration and validation: where relevant, a check on observations that were not used to tune the system.
  • A reference suited to the need, and checks for drift over time.
  • Comparisons on the same fractions, periods and sorting conventions, with their limits of representativeness.
  • Continuous coverage over time does not mean identifying every object or all of the mass.
  • Attributing waste to a supplier requires real traceability: time-stamping alone may not be enough where there is mixing, intermediate storage or a variable transit time.

A more frequent measurement is not automatically a more accurate one. We distinguish what the system observes, what it estimates and what must be confirmed by a reference. Results are assessed under the real conditions of your site.

Reading guide

The limits, set out from the start

What continuous analysis does not do

  • It identifies the objects visible on the conveyor. A camera does not measure the moisture of incoming waste.
  • Accuracy is not the same for an object count as for a mass estimate; that is the purpose of cross-calibration.
  • An alert only has an effect if the organisation can respond to it, with staff available at the right time.
  • Results depend on the configuration of each line and on the incoming waste stream. No figure can be transferred as is from one site to another.

Let’s talk about your site.

Thirty minutes to scope your challenges and see which measurement points make sense at your site.

Book a meeting

Testimonials gathered at the NextWaste webinar “IA en centre de tri : retours d’expériences de l’analyse continue” (AI in sorting facilities: feedback on continuous analysis), held in French on 17 September 2026; the quotes on this page are translated from French. NextWaste article of 30 March 2026 for GreenRécup’.