How RIC counts

Robots in Construction keeps a record of where construction robots have worked on real sites. Every number on this site comes from that record. This page explains what goes into it, what stays out, and how to fix it when we have something wrong.

The record holds 876 deployments of 129 robots, run by 438 organizations in 41 countries. It is updated as new sources are captured.

A deployment is one robot doing real work on one project. One project, one robot, one timeframe. If the same job shows up in a case study, a LinkedIn post and a trade article, it is still one deployment with three sources.

Some things don't count:

• A demo at a trade show or an industry event. The robot isn't on a site doing a job.
• A robot listed in a catalogue, a partner programme or a funding announcement.
• A job post for a robot operator. It tells us a company plans to use robots, not where they used one.

When a source says a company used a robot but doesn't name which model, we hold the record until the model is known. It still counts toward the robot family, on the family page.

Every deployment carries one of three states.

• Production: the robot did real work on a live project, as part of how the job got built.
• Pilot: a trial or first test on a real project. We count trials as pilots.
• Demonstration: the robot was shown working on a real site, without being part of the job.

PRODUCTIONPILOTDEMONSTRATION

Every deployment and every figure carries an evidence tier. It tells you how far to trust it.

• Verified: Published by the client, by the company that ran the robot, or by an independent outlet.
• Probable: Credible, but it rests on the maker's own material or on a source we couldn't fully check.
• Indicative: A trace rather than a statement; we worked it out from the sources around it, or it comes from a passing mention.
• Unconfirmed: A trace we hold but couldn't confirm, always labelled and listed last.

A maker's own case study is the weakest published source, because the maker is selling. We still record it, and we say so.

Some of the best robot jobs can't be named. The client wants it quiet, or the contract says so. Those jobs still happened.

• Documented deployments have a public source. They show as full records with the project, the company and the source.
• Attested deployments were reported to us by the company that ran the robot, or by its maker, without a client or site. They show only as a count: "+6 attested deployments (no client or site named)."

Counts split the two: "20 deployments (14 documented, 6 attested)." We never need a client's name to count a job, and the company that ran it chooses whether its name appears.

Records range from a full case study to a single line on a customer list. We sort them by how much they tell you, not by date.

• Case study: a named project, with a source that is a write-up of that project: a case study, a press article about it, or a post about that project. These come first.
• Project: a named project, with a source that mentions it among others, such as a portfolio page or a post naming several projects.
• Mention: a company and a robot with no project, such as a logo on a maker's customer page. Mentions count in every total. On the page they are grouped into one line, because a buyer learns less from them.

Every record can move up. A company that tells us which project, when and how much the robot did turns a mention into a case study.

Where a source states what the robot did, we show it: area covered, length, points marked, days on the job, and the crew on site. Every figure carries its source and its tier.

• A figure the robot's maker published is labelled maker-stated. A figure from the company that ran the robot or its client is labelled field-measured. We never mix the two on one card.
• We show figures in your units (m² in Europe, ft² in the US). The source's own unit is in the detail.
• We don't publish averages until enough companies have reported for the average to mean something: at least 5 organizations per trade and region. Below that, nothing is averaged.

Most robots run on far more sites than any public record shows. A maker may report hundreds of jobs. We count the ones we can source.

Where the maker publishes its own figure, we show it next to ours, labelled as maker-stated. The gap between the two is the part of the record nobody has published yet. Makers and the companies running their robots can close it by telling us about their jobs.

We don't publish prices, costs, fees, day rates or savings percentages, even when a source states them. Costs depend on the job, the country and the contract. A number lifted from one case study misleads more buyers than it helps.

Every robot gets the same page and the same treatment. A maker that claims its page can keep its specs current and add jobs we haven't found. It can't move its robot up a list, soften a tier or hide a record. A claimed page loses its claimed sections if the maker stops keeping them current.

• Email christof@robotsinconstruction.com.
• Send what's wrong and, if you can, a source. You can stay anonymous.
• Every correction is checked before it goes live. Nothing goes on the site straight from a form.
• If we disagree, we tell you why, and the record notes that it's disputed.

Companies listed in the record can claim their page and correct it directly, through the same check.