Goodwood House builds knowledge-graph and inference tools for risk that is spread across systems and not yet measured, and tests them against pre-registered criteria before they are relied upon.
Whether the subject is a supply chain, a sensor network or an eradication campaign, the method begins in the same way, with the evidence assembled into a formally structured knowledge graph and every claim checked against its source, so that the quantities a decision-maker needs, which are confidence, exposure and early warning rather than raw data, can be computed from it. Where a warning system already exists, the same machinery can be used to test independently whether its signal is ready to be acted upon.
An exposure that no single record shows, made explicit by traversing the graph.
What we do
The signal usually exists, but it is scattered.
Most analytical failures are failures of integration rather than of data collection. The signal is present but distributed across systems, formats and organisational boundaries, and what is missing is the machinery to bring it together with its provenance intact and to reason over it in a way that others can check. That machinery is what Goodwood House builds.
The core of the practice is a family of knowledge-graph and network-inference tools that have been developed and tested against commercial, government and scientific problems. The foundation is the Ontology Auto-Tuner, a refinement engine for OWL and SHACL knowledge graphs demonstrated at a scale of 1.4 million triples, and around it sit applications in supply-chain exposure mapping, provenance-assured data verification, the independent assessment of early-warning signals and inference of absence for field campaigns.
Fixed-price audits
A set scope, a set price and a written report.
The quickest way to put these tools to work on your own problem is a fixed-price audit. Each one has a defined scope and a written report at the end, so you know exactly what you are buying before any work starts, and every finding in the report can be traced back to the evidence behind it.
Supply chain exposure
Most firms know their direct suppliers well, yet the failures that stop production usually sit two or three tiers further back, where several apparently independent suppliers turn out to depend on the same plant, region or component maker. These audits use DependMap to map that hidden structure from your supplier list and bill of materials, and rank the points where a single disruption would reach you through more than one route.
Supplier Exposure Snapshot
£2,950
One product line, up to 50 direct suppliers, 10 working days
Tier 2 and tier 3 mapping built from your supplier list and public sources
The ten most concentrated dependencies, ranked by how many of your routes pass through them
Geographic and single-source exposure summary
Written report with the evidence behind every link
Product feeds carry errors that are hard to see from inside the business, such as missing attributes, contradictory sizes, prices that disagree between channels and identifiers that fail marketplace checks, and each one costs listings, search ranking or returns. These audits use the CatalogueScore engine to check every record against the rules your channels actually enforce and rank the faults by their likely commercial effect.
Feed Health Check
£1,250
One feed, up to 10,000 products, 5 working days
Validation of every record against the target channel specification, for example Google Merchant Center or a marketplace
Error and warning counts by attribute and by category
The twenty faults most likely to be costing you listings or sales
Short written report with worked examples from your own data
A free thirty-minute scoping call, then you send a supplier list or a feed export under a mutual NDA. No access to your systems is needed.
Fixed prices
The price shown is the price you pay, with no VAT added as the company is not VAT-registered. Half is invoiced on order and half on delivery.
Your data
Data is used only for your audit and is deleted on completion unless you ask us to keep it. Larger scopes are quoted at a fixed price.
How we work
The discipline matters as much as the tooling.
Clients see the same practices whatever the domain. They were adopted for settings, from government security programmes to peer-reviewed science, in which an error that goes unnoticed costs far more than one that is found and stated.
Pre-registered analysis
Analyses are pre-registered before data is touched, so that a positive result means what it appears to mean.
Null results reported
Negative findings are reported as plainly as positive ones, because a method whose failures are hidden cannot be relied upon where it succeeds.
Provenance-verified claims
Every claim, including machine-generated content, is verified against a source before it is used.
Readiness stated as assessed
Technology readiness is given as it has been demonstrated rather than as it is hoped to be.
Clean closure
Directions that the evidence does not support are closed and recorded rather than kept alive for appearances.
Capability
Inference, graphs and provenance.
The practice's core work applies knowledge-graph engines, network inference and provenance-assured verification to problems where the risk is real and distributed but has not yet been measured.
Foundation engineOntology Auto-Tuner
Refining knowledge graphs until they can carry weight.
A knowledge graph is only as useful as its consistency, and at real-world scale consistency does not survive contact with real data unless something enforces it. The Ontology Auto-Tuner is an automated refinement system for OWL ontologies and SHACL constraint shapes that detects structural weaknesses, proposes corrections and validates the result.
The Auto-Tuner is the common ancestor of everything else on this page. It allows a supply chain, a sensor estate or an evidence base to be represented formally enough that inference over it can be checked, and it is available both as background IP within consulting engagements and as the basis of licensed applications.
An early-warning pathway is only as useful as the evidence that it would sound when it mattered and stay silent when it did not. Published methods are usually assessed by their authors in terms of skill, whereas the decision-maker's question is narrower and harder, since it concerns whether a particular signal, on particular data and under stated uncertainty, is ready to be acted upon. Goodwood House offers that assessment as an independent layer in which observations, model outputs and forcing scenarios are bound into a provenance-carrying evidence graph, each warning pathway is tested against pre-registered positive controls, negative controls and stationary nulls, and the outcome is a traceable readiness judgement rather than a single headline number.
The layer assesses the detectors it evaluates rather than competing with them, so it can be applied to physics-based and statistical pathways alike. It draws on the Ontology Auto-Tuner for integration and on PRISM for claim-level provenance. To date it has been applied in full to one pathway, CascadeWatch, whose evaluation was frozen as a hash-verified audit package that a third party can reproduce figure by figure.
Pre-registeredprotocols locked before data access
Hash-verifiedfrozen, reproducible audit package
Supply-chain intelligenceDependMap
Finding the supplier you did not know you depended on.
DependMap maps multi-tier supply-chain dependency risk. A bill of materials is ingested, the supplier network is traversed beyond the first tier, and exposures that are invisible in any single procurement record become explicit, the canonical example being indirect rare-earth exposure that surfaces only at the second and third tier of a Western manufacturer's network.
The method was tested blind against the fire at Renesas Electronics' Naka semiconductor plant in March 2021, which disrupted vehicle production across the motor industry. A representative automotive-semiconductor network of twenty nodes, six chipmakers, six tier-one suppliers and eight carmakers, was built from publicly known supplier relationships alone, with no market-share figures and nothing to suggest that any node was larger or more fragile than another. Every node was then ranked by structural systemic risk before the outcome was revealed. Renesas came out first, and although it held no direct contract with any carmaker it could reach all eight through the tier-one suppliers it feeds. The test demonstrates the logic of the method against a known case and does not claim to have modelled Renesas's actual contract book or to have predicted the fire itself.
DependMap complements an ERP system rather than replacing it, because its job is to discover the shared sub-tier structure that a bill of materials does not record. The tool progressed through the Topcoder and Wazoku "Innovation Builders, AI Disruption" challenge, winning Stages 1 and 2 and reaching the Stage 3 vendor showcase, and the working system comprises a live risk dashboard, an event feed and a manual simulation environment. It is the engine behind the supply chain exposure audits offered above.
01Renesas the plant that caught fire
02NXP
03Infineon
04STMicroelectronics
05Texas Instruments
Blind structural ranking of a representative 20-node network, built from publicly known supplier relationships with no market-share data. Renesas, with no direct contract with any carmaker, ranked first and reaches all eight carmakers through the tier-one suppliers it feeds (six are labelled).
Finding the faults in a product feed before a marketplace does.
CatalogueScore applies the same verification engine to retail product data. Each record in a feed is represented in a knowledge graph and checked against SHACL shapes that encode the rules a sales channel enforces, together with consistency rules that compare the same product across channels, so that faults are found, explained and ranked rather than simply counted.
Because the rules are written in an open W3C standard, they can be handed over at the end of an engagement and rerun by the client's own team. It is the engine behind the product feed audits offered above.
Provenance-assured verification for evidence you have to defend.
PRISM is a data model and enforcement layer for organisations that must know not only what their data says but where every statement came from and how well grounded it is. Built on the W3C PROV-O and Web Annotation standards with SHACL enforcement, it scores material along two independent axes, the assurance of the source and the grounding of the claim, so that downstream reasoning can be weighted accordingly.
PRISM was shortlisted to the final three in a UK government intelligence and security challenge, and the evaluators' feedback has since been folded back into the design. It is the practice's answer to a question that generative AI has made urgent, namely how an organisation reasons over material it did not author and cannot fully trust.
Final 3UK government security challenge
Climate early warningCascadeWatch
A network view of Atlantic early warning, tested to a pre-registered standard.
CascadeWatch asked whether destabilisation of the Atlantic overturning circulation and the subpolar gyre could be detected in advance from sea-surface data using a network statistic, the leading eigenvalue of a correlation matrix across physically distinct ocean regions, rather than the single-site indicators that dominate the literature. Every analysis was pre-registered before data was touched, with the predicted outcomes and their interpretations locked in advance.
The observational record showed a multidecadal strengthening of coupling across the Atlantic sea-surface network that passed its pre-registered tests. Model-world testing on the HadGEM3 freshwater-hosing experiments, which is the only setting in which a warning can be checked against a known collapse, then established where a surface coupling signal can be relied upon, and in particular that it needs to be paired with other evidence to separate the approach to a tipping point from a forced retreat that carries no such risk.
The work produced a reusable evaluation protocol of pre-registration, positive and negative controls and stationary nulls, together with a frozen, hash-verified audit package that a third party can reproduce figure by figure. That protocol now underpins the practice's signal-readiness assessment for early-warning pathways more generally. The work was presented at an invited seminar to Professor Tim Lenton's tipping-points group at the University of Exeter in September 2026.
Decade means of the network statistic from ERSSTv5 sea-surface data across three Atlantic regions, with a sketched line between them. The rise out of the early-century minimum carries the positive trend.
Eradication inferenceEradication-confidence engine and planning tools
Proof of absence, and the difference between clearing a problem and merely holding it down.
The hardest question in any eradication campaign is not how to remove the target but how to know, defensibly, that removal is complete. Absence of detection is not detection of absence, and a campaign that cannot tell the difference either stops too early or pays for suppression indefinitely.
Goodwood House approaches this as an inference task. A stage-structured population model is combined with multi-modal detection data in a provenance-aware framework that generalises published inference-of-absence methods, and the output is a running, quantified confidence that eradication has been achieved, which separates a genuine path to zero from permanent suppression. The approach was developed for the invasive brown tree snake problem on Guam and validated in simulation against published field data.
The same thinking was then applied to an existing public-domain planning model for the problem. Working entirely from the published release, the practice reproduced that model independently from its equations and validated the reproduction against the model's own worked example, which it matched exactly on cost, coverage and the principal outcome probabilities, and in doing so found points at which results depended on the spreadsheet software evaluating the model rather than on the model itself. The single-file model was rebuilt as a transparent planning tool that runs in a browser with no specialist software, and was then extended, again using only published data, so that it shows whether a plan is on a path to permanent clearance or is settling at a plateau where every animal removed is replaced from the surrounding landscape. A published behavioural dataset was incorporated so that the physical placement of control tools informs the forecast. A companion tool reimplements a published model of native lizard reintroduction and links a suppression plan's residual snake density to the probability that the lizards re-establish.
Schematic of the planning tool's main output, which shows whether a control plan is heading for permanent clearance or settling where every animal removed is replaced.
Both tools are independent work built from public releases and are not endorsed by the original model authors.
Validatedagainst the public model's own worked example
Alongside the named tools, Goodwood House has developed proposals for UK government innovation programmes, including HMGCC Co-Creation and ARIA, on problems ranging from the visual comprehension of complex sensor estates to thermal management for compact electronics. Details of programme work are not published here.
Selected challenge work
A real problem, a hard deadline and an expert evaluation.
The inference practice sits on a broad engineering foundation, and Goodwood House keeps a deliberate sideline in open-innovation challenges across the physical sciences, because the format puts ideas in front of evaluators who know the field. The work has produced stage wins and shortlistings, and each submission is described here as it was described to the evaluators.
Pharmaceutical
Synthesis routes for a development-stage analgesic, including a Grignard and dehydration route costed with conservative cost-of-goods modelling rather than best-case assumptions.
Analytical chemistry
A submitted approach to freeze-thaw stability of serum calcium assays, an analytical chemistry problem in a regulated context.
Structural materials
A concept for a low-viscosity fluid that forms discrete load-bearing microparticles in place, addressing an in-situ proppant generation problem.
Polymer applications
New application development for a commercial polyolefin elastomer family for a major petrochemical seeker.
Thermal management
A passive phase-change thermal buffer for a high-power portable radio, developed for a UK government co-creation challenge and assessed by its panel as mature and compliant with every essential requirement.
Humanitarian tooling
An offline-first damage-reporting web application with formal data validation, built for a UNDP crisis-mapping challenge and taken through to the evaluation phase from a field of about eighty entries.
About
Who we are.
Goodwood House Ltd is the independent research and engineering practice of Michael Eccleston, whose prior career reached SVP and CIO level in global freight logistics, following earlier engineering work that included project management of a nuclear decommissioning robot and maintenance-management systems for UK water utilities. The company operates as a specialist subcontractor and consultant, taking problems from concept through to validated demonstrator, and acts as an independent assessor where the question is whether an existing analysis or warning can be relied upon.
The practice has worked on open innovation platforms including Wazoku, InnoCentive and Topcoder over a sustained period, with stage wins and shortlistings across intelligence, environmental, pharmaceutical and industrial domains.
Company
Goodwood House Ltd
Registered
England and Wales
Prior career
SVP and CIO, global freight logistics
Earlier engineering
Nuclear decommissioning robotics, water-utility maintenance systems
Platforms
Wazoku, InnoCentive and Topcoder
If your organisation has a risk it can sense but cannot yet quantify, or a warning it cannot yet verify.
The simplest place to start is one of the fixed-price audits above, beginning with a free scoping call. Beyond those, we take on a small number of engagements at a time, typically scoped as a feasibility study, a demonstrator build, an independent assessment, or a licensed application of an existing engine with knowledge transfer.