FRISS vs Shift Technology: Fraud Detection Head to Head

Insurance fraud costs carriers an estimated $80 billion annually in the United States alone, according to the Coalition Against Insurance Fraud. That figure has made fraud detection one of the highest-priority AI use cases in insurance, and it has pushed two vendors to the front of the conversation: FRISS and Shift Technology.

Both companies are purpose-built for insurance. Both have significant customer bases. But they approach fraud detection differently, they serve different carrier profiles, and they have different strengths depending on whether your primary concern is underwriting fraud, claims fraud, or both. Our team evaluated both platforms during a 2024 vendor selection for a regional P&C carrier processing approximately 800,000 claims per year. This article reflects what we found during that evaluation, supplemented by conversations with carrier teams that have deployed each platform.

Why This Comparison Matters

The fraud detection vendor decision is consequential for three reasons:

1. Long deployment timelines. Both platforms require 6 to 12 months for full deployment including data integration, model calibration, and SIU workflow configuration. Switching vendors after deployment is costly.
2. Data dependency. Fraud detection models need historical claims data for calibration. Once a vendor’s models are trained on your data, the switching cost is not just financial; it is the loss of tuning and calibration work.
3. Regulatory implications. Fraud referral decisions based on AI scores may face regulatory scrutiny. The explainability and documentation capabilities of your fraud platform matter for compliance.

Choosing between FRISS and Shift is not a casual software evaluation. It is a multi-year strategic commitment.

FRISS Overview

FRISS (originally “Fraud, Risk & Compliance”) is a Netherlands-based company founded in 2006 that has built its business on what it calls “Trust Automation.” The company’s core concept is a Trust Automation Score: a 0-to-100 score assigned to policies, claims, and policyholders that indicates the level of trust the carrier should place in each transaction.

Deployment Scale

FRISS reports 300+ implementations globally, with strong presence in Europe, Latin America, and growing deployment in North America. The company’s European roots give it broader multi-country deployment experience than Shift, which is relevant for carriers with international operations.

Product Coverage

FRISS covers three domains:

• Underwriting fraud detection: Scoring policy applications at the point of sale to identify misrepresented risk, potential fraud, and suspicious patterns before the policy is issued.
• Claims fraud detection: Scoring claims at FNOL and throughout the claims lifecycle to flag suspicious claims for SIU investigation.
• SIU case management: Workflow tools for managing fraud investigations, documenting findings, and tracking referrals.

This breadth is FRISS’s key differentiator. Most carriers experience fraud at both the underwriting and claims stages, and FRISS provides a single platform that addresses both.

Trust Automation Score

The Trust Automation Score is a composite score that combines rules-based indicators, statistical anomaly detection, external data enrichment (public records, watchlists, prior claim history), and machine learning models. The score is designed to be explainable: each score comes with a breakdown of contributing factors that SIU investigators and underwriters can use to understand why a transaction was flagged.

FRISS markets the score as a basis for “straight-through processing,” meaning transactions with high trust scores can be processed without human review, while low-trust-score transactions are routed to SIU or underwriting review. In practice, carriers we spoke with use the score more as a prioritization tool than an automation trigger: low scores get reviewed first, but human judgment still drives the final referral decision.

Shift Technology Overview

Shift Technology is a France-based company founded in 2014 that has built its reputation primarily on claims fraud detection. The company has expanded into claims automation and underwriting, but its core strength remains in the claims fraud space.

Deployment Scale

Shift reports 100+ carrier deployments globally, with strong presence in Europe and North America. While the total deployment count is lower than FRISS’s, Shift’s deployments tend to be at larger carriers, including several of the top 20 global insurers.

Product Coverage

Shift offers three main products:

• FORCE (Fraud Detection): The original product. FORCE uses supervised machine learning, network graph analysis, and rules to detect fraud at the claim level.
• SAAM (Subrogation and Automation): Claims automation that identifies claims eligible for straight-through processing and claims with subrogation potential. SAAM is not a fraud product; it is a claims efficiency product.
• Underwriting fraud detection: A newer product that extends Shift’s fraud models to the point of sale.

The combination of FORCE and SAAM gives Shift a claims-focused value proposition that goes beyond fraud detection into broader claims automation. For carriers that want to improve both fraud detection and clean claims processing speed, this combination is Shift’s strongest selling point.

FORCE Scoring Approach

FORCE uses a multi-layered scoring approach:

1. Supervised ML models trained on the carrier’s own historical fraud data (confirmed fraud outcomes) to identify patterns associated with fraudulent claims.
2. Network graph analysis that builds relationship maps across claimants, providers, attorneys, repair shops, and other entities to identify suspicious clustering.
3. Rules engine that captures known fraud indicators and regulatory requirements.
4. Anomaly detection that identifies claims that deviate from expected patterns even if they do not match known fraud schemes.

Each flagged claim includes an explanation of the contributing factors, designed for SIU investigators. Shift emphasizes “decision explainability” as a key differentiator, and in our evaluation, Shift’s explanations were more detailed and investigator-friendly than FRISS’s.

Head-to-Head Comparison

Scoring Approaches Compared

FRISS Trust Automation Score

FRISS’s scoring philosophy is built around the concept of “trust” rather than “fraud.” The Trust Automation Score is assigned to every transaction (policy application, claim, or policyholder record), not just suspicious ones. This means the system produces a score distribution where most transactions score high (trustworthy) and a small percentage score low (warranting review).

The practical advantage of this approach is that it enables straight-through processing decisions: transactions above a certain trust threshold can be auto-approved, while transactions below the threshold are routed to human review. For carriers that want to increase processing speed on clean business while catching suspicious activity, the trust-based framing aligns well with operational goals.

The practical limitation is that the “trust” framing can mask the underlying model complexity. When we asked FRISS to explain how the Trust Automation Score is calculated, the answer involved rules, external data lookups, statistical models, and machine learning, combined into a single composite score. The composite nature makes it harder to understand which component drove a specific score change compared to Shift’s more transparent multi-layer approach.

Shift FORCE Score

Shift’s scoring is more explicitly layered. A flagged claim comes with separate indicators from the ML model, the network analysis, the rules engine, and the anomaly detection module. Each layer contributes independently, and the investigator can see which layers flagged the claim and why.

In our evaluation, SIU investigators preferred Shift’s layered explanations because they could quickly identify whether a flag was driven by a known pattern (rules), a statistical anomaly, or a network relationship. This transparency made triage faster: an investigator could decide whether a flag warranted a phone call, a document review, or a full investigation based on which detection layer triggered the alert.

The limitation is that Shift’s scoring is claims-focused by design. While Shift has expanded into underwriting fraud, its underwriting product was less mature than FRISS’s during our evaluation. Carriers whose primary fraud exposure is at the point of sale may find FRISS’s underwriting capabilities more developed.

Network Analysis: The Key Technical Differentiator

Network analysis (also called link analysis or graph analysis) is the detection of fraud patterns that emerge from relationships between entities rather than from individual claim characteristics. Organized fraud rings, staged accident schemes, and provider mill operations all produce network signatures that individual-claim analysis misses.

FRISS Network Detection

FRISS includes network detection as part of its Trust Automation platform. The system identifies connections between claimants, providers, and other entities across claims and flags suspicious clustering. FRISS’s network analysis is functional and catches common relationship patterns.

Shift Network Graph Analysis

Shift’s network analysis is generally considered deeper and more sophisticated. Shift builds explicit relationship graphs that map connections across claimants, attorneys, medical providers, repair facilities, and witnesses. The system identifies:

• Direct connections: The same attorney appears on multiple suspicious claims.
• Indirect connections: Claimants who do not know each other use the same body shop, which uses the same appraiser, who is connected to a specific attorney.
• Temporal patterns: Clusters of claims filed within a short period that share entity connections.

In our evaluation, Shift’s network analysis flagged a pattern involving three body shops that used the same independent appraiser for supplemental damage estimates. The claims were spread across different claimants and different coverage types, so they would not have been connected by individual-claim analysis. FRISS’s network detection did not flag this pattern in the same test data set.

This is one data point, not a definitive conclusion, but it aligns with what we heard from other carrier teams: Shift’s network analysis is its strongest technical differentiator.

Claims Automation: Beyond Fraud

FRISS is a fraud detection platform. Its automation capabilities are focused on routing transactions based on trust scores: high-trust claims go to automated processing, low-trust claims go to SIU review.

Shift’s SAAM product goes further. SAAM is a claims automation module that identifies:

• Clean claims eligible for straight-through processing (no fraud indicators, complete documentation, clear coverage).
• Claims with subrogation potential (another party may be liable, and recovery is possible).
• Claims requiring specific handling (complex liability, attorney involvement, regulatory notification requirements).

For carriers that see fraud detection and claims automation as related problems (both involve intelligent routing of claims to the right handler), Shift’s combined FORCE + SAAM offering is a stronger value proposition than FRISS’s fraud-only platform. The counter-argument is focus: FRISS does one thing (fraud and trust scoring) and does it across the full policy lifecycle, while Shift combines fraud detection with claims automation in a way that could dilute focus.

Integration Reality

Both platforms integrate with major insurance core systems, but the specifics differ.

FRISS Integrations

FRISS has pre-built integrations with Guidewire, Duck Creek, Sapiens, and TIA. The Sapiens and TIA integrations give FRISS an advantage for carriers running European-origin core systems, which is consistent with FRISS’s stronger European footprint. FRISS also integrates with several external data providers (credit bureaus, public records, industry databases) for data enrichment.

Shift Integrations

Shift integrates with Guidewire, Duck Creek, and Majesco, covering the dominant North American core system vendors. Shift’s API-first architecture supports custom integrations with less common systems, though custom integrations add deployment time and cost.

For carriers running Guidewire (the most common modern core system in North America), both platforms have mature integrations and the choice does not hinge on integration availability.

Deployment Considerations

Data Requirements

Both platforms need historical claims data for model training and calibration. The minimum viable data set depends on the carrier’s line of business and fraud prevalence:

Shift’s higher data requirements reflect its reliance on supervised ML models that need confirmed fraud outcomes for training. FRISS’s approach, which includes more rules-based and data-enrichment-based scoring, can produce useful results with less historical data.

Deployment Timeline

Both platforms typically require 6 to 12 months for full deployment, including:

• Data integration and mapping (2 to 3 months)
• Model training and calibration (2 to 3 months)
• SIU workflow configuration and user training (1 to 2 months)
• Parallel running and validation (1 to 2 months)

Carriers should not expect production-quality fraud detection in the first 6 months. Both vendors can deliver initial scoring within 3 to 4 months, but calibration and tuning continue for 12 to 18 months after go-live.

Ongoing Tuning

Fraud detection is not a set-and-forget deployment. Models degrade over time as fraud patterns evolve, and false positive rates tend to increase without regular recalibration. Both FRISS and Shift provide ongoing model tuning services, but the internal resource commitment is real:

• A dedicated fraud analytics owner (0.5 to 1.0 FTE) to manage alert thresholds, review model performance, and coordinate with SIU
• Quarterly model review sessions with the vendor’s data science team
• Regular SIU feedback loops to improve model accuracy

Carriers that deploy fraud detection without this ongoing commitment consistently report declining value after the first year.

Honest Limitations of Both Platforms

False Positive Rates

Both platforms produce false positives. In our evaluation, initial false positive rates (claims flagged that were not fraud) were 60% to 70% for both platforms. After calibration with carrier-specific data, those rates improved to 40% to 50%. That means roughly half of flagged claims are not fraud after investigation.

This sounds discouraging, but the relevant comparison is not “what percentage of flags are fraud” but “what is the cost of investigating a false positive versus the savings from catching fraud?” If your average confirmed fraud recovery is $25,000 and the average investigation cost (SIU time) for a false positive is $500, a 40% precision rate is still highly cost-effective.

Novel Fraud Schemes

Both platforms are stronger at detecting known fraud patterns than novel schemes. Supervised ML models, by definition, learn from historical fraud; they are less effective against schemes that look nothing like past fraud. FRISS’s rules engine and data enrichment add detection coverage for known indicators. Shift’s anomaly detection adds some novel-scheme capability. Neither platform is reliable at detecting entirely new fraud methodologies with no historical precedent.

Small Carrier Limitations

Both platforms recommend minimum claim volumes (100K to 200K annually) for effective model training. Carriers below these thresholds may not have enough fraud signal in their data for the ML models to learn effectively. For small carriers, simpler rules-based fraud scoring or industry consortium data may be more appropriate than full-platform deployments.

Verdict

Choose FRISS if: your primary fraud exposure spans both underwriting and claims, you operate in multiple countries (especially European markets), you want a single platform for trust scoring across the policy lifecycle, or your data history is limited and you need to start with enrichment-based scoring before building ML models.

Choose Shift Technology if: your primary fraud exposure is in claims, network analysis and organized fraud ring detection are priorities, you want to combine fraud detection with claims automation (SAAM), you have 3+ years of claims data with confirmed fraud labels, or your SIU team values detailed, layered explanations for triage decisions.

For many carriers, the decision comes down to scope. FRISS is the broader platform covering both sides of the policy lifecycle. Shift is the deeper platform on the claims side with stronger network analysis and the added value of claims automation. If you had to pick one word to describe each: FRISS is breadth; Shift is depth.

Both platforms deliver measurable value when properly deployed and maintained. The bigger risk than choosing the “wrong” vendor is underinvesting in the deployment, training, and ongoing tuning that both platforms require to perform at their best.