> ## Documentation Index
> Fetch the complete documentation index at: https://docs.thedatacity.com/llms.txt
> Use this file to discover all available pages before exploring further.
# SPV identification
> What The Data City means by a special purpose vehicle (SPV), how likely SPVs are identified, and how to use the IsLikelySPV field.
## Overview
The Data City flags companies that are **likely special purpose vehicles (SPVs)** on UK company records. The result appears as the boolean field `IsLikelySPV`.
This page explains what we mean by an SPV, how the flag is produced at a high level, and how you should use it in analysis. For the field definition, see [`IsLikelySPV`](/data-dictionary/classified-company#is-likely-spv) in the data dictionary.
This is an **operational definition**, not a legal or accounting classification. Use `IsLikelySPV` to improve economic analysis — for example when adjusting employee or turnover totals — not as a definitive statement about a company's legal status.
## What we mean by an SPV
An SPV is a **UK-registered company whose primary role is to hold assets, pass through financing, or sit within a corporate group** — not to employ staff and conduct ordinary commercial operations in its own right.
Such companies are often created to:
* Isolate risk or ring-fence property or debt
* Facilitate securitisation or private-equity structures
* Act as intermediate holding entities
They may file accounts and have directors, but their economic footprint in terms of employees and turnover is often minimal, absent, or atypical over long periods. They frequently appear alongside related companies in wider corporate structures.
In practice, we treat a company as an SPV when it **looks structurally like a vehicle** — in its role within a group and its governance — and **behaves financially like a non-trading or dormant entity**, while often still holding balance-sheet assets.
## How we identify likely SPVs
We use a **machine-learning model** trained on verified examples and applied to UK company-register data, enriched with The Data City's proprietary company dataset.
The model does **not** rely on a single rule. It weighs several types of evidence together, in much the same way an analyst would review a company from multiple angles:
How the company sits within a corporate group, its governance, and characteristics of its registered identity.
Patterns in reported employees, turnover, and filing history over time — including long periods of little or no trading activity.
Whether the company holds assets or financing on its balance sheet despite limited operating activity.
Whether the company appears to operate as a distinct trading brand rather than a passive structural entity.
Companies that align across these dimensions are more likely to receive `IsLikelySPV = true`. Companies with sustained trading activity, meaningful employment, and the profile of an independent operating business are less likely to be flagged.
After scoring, we apply **additional quality checks** — including rules that distinguish passive structural vehicles from legitimate group-head or holding companies with real economic activity. These checks reduce false positives among active parent entities without treating every ultimate parent as an operating business.
The field is refreshed **quarterly** as part of our standard data updates.
`IsLikelySPV` is a **probabilistic estimate**, not ground truth. Borderline cases exist, and the flag should be interpreted alongside other company attributes — especially group structure, financials, and sector context.
Specific model design, training data, scoring thresholds, and post-processing rules are **proprietary to The Data City** and are not published in full.
## What the flag means
| Value | Meaning |
| ------- | ----------------------------------------------------------------- |
| `true` | The company is **likely** an SPV under our operational definition |
| `false` | The company is **unlikely** to be an SPV |
A `false` value does not guarantee that a company is an active trading business. A `true` value does not mean the company has no legitimate purpose within a group structure.
## How to use it
Common applications include:
* **Excluding structural vehicles** from sector counts, growth analysis, or employment totals where SPVs would distort the picture
* **Identifying project- or asset-specific entities** within a wider corporate network
* **Cleaning company lists** before delivery or downstream modelling
Filter or export on `IsLikelySPV` when you need to separate likely operating businesses from likely structural vehicles. Combine the flag with other filters — such as company category, group structure, and financial estimates — for more precise results.
## Limitations
* The flag reflects patterns observed in UK register and platform data at the time of model training. Unusual corporate structures or recently changed companies may be misclassified.
* Coverage and accuracy depend on the quality of filed accounts and group-relationship data available for each company.
* Some legitimate holding companies and group parents can resemble SPVs on register data alone; our post-processing reduces but does not eliminate these edge cases.
* The flag is intended for **population-level analysis**. For individual companies where the classification matters materially, review the underlying filings and group context.
## Related fields
* [`IsLikelyDistinctBrand`](/our-data/proprietary-data/what-are-distinct-brands) — identifies meaningful operating companies within a group
* [Distinct brands and subsidiaries](/our-data/using-our-data/what-are-and-how-to-remove-subsidiary-companies) — guidance on analysing companies within corporate groups