Senserity Help
Concepts

How adverse media screening works

What adverse media screening checks, how it covers both the company and its key people, and how results feed into the risk score.

Adverse media screening searches news and media sources for negative coverage associated with a company or its directors and persons with significant control (PSCs). It is one of Senserity's paid enrichments, sitting within the Media risk category, and it runs 20 automated tests once the screening data is available.

Why it matters

Public records tell you whether a company is filing on time and paying its debts. They do not tell you whether it has been linked to fraud, sanctioned individuals, regulatory enforcement, or criminal activity. Adverse media screening fills that gap by checking whether the company or its key people have appeared in news coverage for the wrong reasons.

For procurement and compliance teams, adverse media is often a required step in third-party due diligence. Many regulatory frameworks, including anti-money laundering rules and public sector procurement regulations, expect organisations to screen suppliers and partners for reputational risk before entering into contracts.

What it screens

Senserity screens in two stages: the company itself, then the individuals behind it.

Company-level screening

The first stage searches for the company name across commercial news and media feeds. Results are classified into six adverse categories:

Financial crime. Fraud, money laundering, bribery, corruption, embezzlement, and tax evasion. This category carries the highest weight in the risk scoring algorithm.

Organised crime. Links to criminal networks, gang activity, trafficking, and organised criminal enterprises. This also carries the highest weight and is typically considered disqualifying in due diligence.

Terrorism. Connections to terrorist organisations, financing of terrorism, and related activity.

Regulatory enforcement. Fines, sanctions, licence revocations, and formal enforcement actions by regulators.

Violent crime. Convictions or allegations involving violence, assault, or related criminal conduct.

Political exposure. Connections to politically exposed persons (PEPs) or politically sensitive situations. This category carries the lowest weight, as political exposure is a contextual risk rather than an inherently negative signal.

Articles that mention the company but do not meet the threshold for any of these categories are recorded as uncategorised mentions. These appear in the results as neutral context but do not count as adverse hits.

Individual-level screening

The second stage extends the search to the company's directors and PSCs. This matters because a company can have a clean public profile while the people running it carry significant personal risk.

Individual screening uses the same six categories and the same classification logic as the company-level screen. Each person is screened separately, and results are shown in a table with their name, role, hit count, risk level, and match confidence.

Both company-level and individual-level screening run together as part of the same enrichment. Whenever the screening data is stale and a re-enrichment occurs, both the company and its directors and PSCs are screened in a single pass.

How results are scored

Senserity calculates an adverse media risk score using a weighted algorithm. Each adverse category has a weight reflecting its seriousness:

CategoryWeight
Financial crime10
Organised crime10
Terrorism8
Regulatory enforcement7
Violent crime5
Political exposure1

In addition to the category weights, each individual article is assigned a relevance weighting based on how directly it relates to the entity being screened. Articles with weak or tangential connections to the company or person (such as passing mentions or loosely related coverage) receive the lowest weightings, and the two lowest weighting tiers are excluded from scoring entirely. This means that only articles with a meaningful connection contribute to the risk score, reducing noise from incidental mentions.

The scoring uses a logarithmic scale, so the first few hits in a category have a larger impact than additional ones. This reflects the reality that one confirmed fraud article is significantly more concerning than the difference between ten and eleven articles.

The raw score maps to four risk levels:

Risk levelScore threshold
High50 or above
Medium30 to 49
Low15 to 29
NoneBelow 15

Company-level and individual-level scores are calculated independently. The combined risk score takes the higher of the two. A company with a clean entity profile but a high-risk director still registers as high combined risk.

Match confidence and false positives

Name matching in media screening is inherently imperfect. A search for "Smith Engineering" might return results about a different company with a similar name. A search for a director called "James Taylor" might pick up articles about any number of unrelated people.

Senserity addresses this with a match confidence assessment. At the company level, it evaluates how distinctive the company name is. At the individual level, it scores name uniqueness based on factors such as common first names, common surnames, and whether the person has a multi-part name. Confidence bands are High (70 or above), Medium (40 to 69), and Low (below 40).

Low-confidence results are flagged for manual review. They are included in the results, but the platform highlights that the articles may not relate to the company or person in question. This is particularly important for individuals with common names where false positives are more likely.

What each test covers

The Media category runs 20 tests in total. The key ones are:

Screening status. Whether a screening has been performed and how recently. Screening data older than 30 days is flagged as potentially stale, and data older than 180 days is flagged as requiring a refresh.

Hit detection. Whether the company or any individuals have returned categorised adverse hits, with a count and breakdown by category.

Category-specific checks. Separate tests for financial crime, organised crime, terrorism, regulatory enforcement, violent crime, and political exposure, each surfacing article headlines and source links as evidence.

Risk assessment. The computed risk score and risk level, at both company and individual level, plus a combined score.

Match confidence. An assessment of name uniqueness and the likelihood of false positives, with reasoning explaining the factors behind the score.

Review status. Whether individual-level findings have been reviewed, dismissed, or escalated by your team.

Evidence availability. Whether source articles are available to back up each finding, which matters for audit trails and compliance documentation.

Severity and Red Flags

Financial crime, organised crime, and terrorism findings are classified as Critical severity. A confirmed hit in any of these categories can trigger a Red Flag, which adds a bonus to the company's overall risk score. Specifically, recent financial crime or terrorism media exposure triggers a Tier 3 Red Flag, adding 20 points to the overall score on top of the category-level impact.

Regulatory and violent crime findings are classified as High severity. Political exposure is Low severity. These contribute to the Media category score and the overall risk grade, but they do not trigger Red Flags on their own.

Credits and availability

Adverse media screening is a paid enrichment that consumes 2 credits per company. It is available from the Trial tier upwards. Screening runs as part of the company's enrichment cycle, either on the default monthly schedule or at whatever frequency you have configured for the company (weekly or daily). Each enrichment run that includes adverse media consumes credits.

The screening covers both the company and its individuals in a single enrichment run. You do not pay separately for individual-level screening.

See How enrichment works for more on how paid enrichments are scheduled, and The credit system explained for the full credit cost breakdown.

Last updated .

On this page