GridReady, how it works

GridReady turns public planning and grid data into a battery storage readiness score in seconds. This page sets out what the score means, what it reads, and what it does not claim, in plain terms.

01 · Screen

The score comes from a fixed set of rules run over public data, not from opinion. It gives you a first read in seconds, so you know where a site is strong and where it needs work before you commission it or spend on due diligence.

It is decision support, not a permit and not a promise. A high score does not mean permission is granted, and a low score does not mean a site is dead. It means the public record shows more, or less, work still to do.

The public data it reads

The score reads public sources: the national mapping and cadastre services (PDOK and Kadaster), protected-nature and dwelling data, published permit notices, the zoning register, and the network operators' published congestion data. Every result lists exactly which sources it used and their licences, so you can check the record yourself.

One caveat on congestion. The operators' published congestion map is not available to us as a licence-clean national feed, so we self-host our own extract of it. Today that extract covers the three demonstration areas around our worked examples. Screen a site outside them and the congestion layer reports unknown rather than guessing.

What you tell us, and what it is worth

Public data cannot answer everything. No register will tell us whether you hold the land, how far your permit application has actually got, or which studies you have finished, so the form asks you. Those answers feed the score alongside the public record, and the result says which parts came from you.

Anything you tell us is labelled as declared. A declaration counts for less than the same fact shown in a document, which counts for less than a fact a named reviewer has verified. Ticking boxes cannot talk a site up.

What the score is built from

Five criteria feed the score. Each carries a fixed weight, stated alongside it. When a criterion cannot be scored, it drops out of the mean entirely and the remaining weights are rescaled to fill the gap, so a stated weight holds exactly only when every criterion has a score.

Grid feasibility

Whether the site sits in an area the network operator has flagged as congested, in the direction you asked about: drawing power, feeding in, or both. It reads the state of the area. It is not a reservation of capacity, and it does not test your requested power against a published headroom figure, because no operator publishes one per feeding area. Grid feasibility carries 30% of the score, the largest single weight.

Land control

Whether you, or whoever you are working with, control the land for long enough, and over enough of the parcel, to see the project through. Land control carries 20% of the score.

Permit pathway

Whether the zoning already allows this use, and how far the permit application has actually progressed. Permit pathway carries 25% of the score, the second-largest weight.

Appeal exposure

How exposed the project is to a legal challenge once a permit is granted, and how far along the studies are that would head one off. Appeal exposure carries 15% of the score.

Evidence completeness

How much of the readiness case is backed by evidence rather than asserted. A declaration counts, a document counts for more, and a verified record counts for most. Declare nothing and this criterion scores at the bottom, which is honest: nothing has been shown yet. Evidence completeness carries the remaining 10%.

These five weights, and the band cut-offs below, are expert-elicited v0 priors, not yet empirical. Nobody has back-tested them against real permitting outcomes, because that evidence does not exist yet anywhere in this market. The rubric itself is versioned and hashed, and every dossier carries that version and hash. If we revise a weight, the change is visible on every dossier that used the old one, not folded quietly into the same score under the same name.

What the bands mean, and where they cut

Every site lands in one of five bands: READY+ from 85 upward, READY from 70 to 84, CONDITIONAL from 55 to 69, EARLY from 40 to 54, and NOT READY below 40.

What we do not publish, and what you can work out anyway

One layer of the rubric is not published: the jurisdiction adapter, which turns each categorical input into a number for the Netherlands, and prices what a verification action costs. We do not publish that file, its modifier and price tables, or its internal workings. Building the one we have took real work, and building the next one will too. That effort is the part we are not handing over.

We will not pretend it is unknowable, though. A screening shows you its inputs and the sub-score they produced. For a site whose dossier is already published, one of those rows is handed to you directly, no patience required. More generally, anyone patient enough to screen a site while changing one answer at a time can work out what an individual input is worth. That follows from showing our working, and we would rather show it than hide the number that makes the score checkable. If you want a point value, you can find it. What takes the work is the rest: the outcome data behind the weights, the verification, and keeping the whole thing current as the rules change.

The weights and the band cut-offs above are also published as JSON, at /api/methodology, for anyone who wants to check them without reading a web page.

What the score does and does not claim

The score reflects what the public record can see today, nothing more. Where the data does not cover something, that part stays marked as unknown rather than guessed at, because a confident wrong answer is worse than an honest gap.

The score is a fast first read. It does not replace due diligence, legal advice, or a conversation with the grid operator.

02 · Evidence

The deep dossier behind a score quotes its sources verbatim. Every claim links back to the document it came from, so you can follow each statement to where it was found rather than take our word for it.

03 · Countersign

Humans sign. Models never score. The score itself is deterministic: it comes from a fixed, versioned rubric, and the same inputs always produce the same result. Models help read documents and draft the dossier, but they never move the number.

Before anything is called verified, a named person reviews the evidence and countersigns it. The countersign is a human act, not an automated one.

Recorded demonstrations are labelled

Some site examples on GridReady are shown as a recorded replay rather than a live run, mainly to show what a deep dossier looks like without waiting for one to build. Every one of these is labelled as recorded, so you always know whether you are looking at a live result or a canned demonstration.

Privacy

Running a live screening records a small amount of information so we can build an honest benchmark over time. Read our privacy notice for what that means and how long we keep it.