Point Predictive review
Point Predictive is a Income & Asset Verification product. MortgageTechReview scores Point Predictive 2.1 out of 5.0, ranking Point Predictive #31 of the 35 products tracked in Income & Asset Verification Software, as of August 11, 2026. Scores on MortgageTechReview are weighted across five axes and are never paid for or influenced by a vendor relationship.
Point Predictive builds fraud risk models from a consortium data repository, and its centre of gravity is auto lending. MortgagePass is the mortgage product, a score that rank-orders applications by the odds of material fraud. It claims up to half of applications route to lighter review, and up to sixty percent fewer false positives. The buyer is a lender with enough volume that triage changes staffing rather than just flagging files. What decides it: will you trust a mortgage score from a company whose proven data depth is auto. Disclosure is the limit, with no integration, no pricing, no ownership detail and no mortgage results beyond one announcement.
How Point Predictive compares to Model Match
Ranked first in VOI/VOAModel Match currently scores highest in VOI/VOA, so every other product in the category is compared against it here. That is a ranking on our published rubric rather than a recommendation, and it changes when the scores change. Category Leader
| Axis | Point Predictive | Model Match |
|---|---|---|
| Production impact | 2.3 | 4.9 |
| Functionality & depth | 2.3 | 4.8 |
| Integrations & ecosystem | 1.6 | 4.3 |
| Adoption & support | 2.1 | 4.9 |
| Return on spend | 2.1 | 4.9 |
| Overall | 2.1 | 4.8 |
Point Predictive wins 0 of 5 axes against Model Match, on the weight profile published for this category. Full head-to-head →
Where it wins
- MortgagePass rank-orders applications by fraud propensity, enabling risk-based triage over binary flags
- Draws on application, servicing, real estate and behavioural data plus its consortium repository
- Published paystub fraud research, finding one in ten submitted paystubs fake, grounds the modelling
- A separate mortgage model validation service exists for defending an incumbent model
Where it falls short
- No origination system or point of sale integration is named anywhere
- The half-of-applications and sixty percent claims carry no third-party validation
- Consortium depth is strongest in auto, with no equivalent mortgage scale disclosed
- Ownership and funding are undisclosed, and no pricing appears anywhere
Why it scores 2.1
Scored on the Verification & Data weight profile. The number shows where it sits in this category. It rests on evidence anyone can check, including the vendor's own record. The weights →
Production impact
25% of scoreTriage scoring changes capacity when it works. Routing half a pipeline to lighter review frees underwriting hours no workflow tool recovers. That figure is the vendor’s own, and so is the sixty percent cut in false positives; neither carries third-party validation. One named adopter, Stearns, is more disclosure than several rivals manage and still not a body of evidence. The consortium depth behind the model is auto, not mortgage. A mechanism, not a result.
Functionality and depth
20% of scoreMortgagePass is a score, not a system. It produces a number and a rank order, with no investigation workflow, no audit trails and no condition clearing. That is a real gap against FraudGuard or ICE Fraud Monitor, which surface the underlying data and manage exceptions to resolution. IncomePass, BorrowerCheck and a model validation service extend the line. The analytics read strong. The operational surround a fraud team actually works in is missing.
Integrations and ecosystem
20% of scoreNothing is named. No origination system, no point of sale vendor, no data partner and no delivery mechanism for the mortgage product. A score has to land at one exact decision point in underwriting, and nothing published tells you how it gets there. Assume an API build, staff it, and add that to the quote. This silence is the biggest gap in the record and the reason the number sits near the bottom.
Adoption and support
10% of scoreA score is easy to consume and hard to trust. Your risk team has to accept the cut points, watch for drift, defend the model to investors and answer auditors. That is analytics work, not training. Point Predictive publishes nothing on implementation timelines or ongoing model governance support, and ownership and funding are undisclosed too. Those are exactly the questions that decide whether a fraud score survives its first bad quarter.
Return on spend
25% of scoreThe economics only work at scale. Below a few thousand applications a year a fraud score does not save enough hours to cover the contract, and the risk-avoided argument is too diffuse to underwrite. Above that line a real cut in manual review is worth money, but only if the false positive claim survives your portfolio rather than a consortium average built mostly on auto lending. No pricing appears anywhere.
On price. Not published. There is no pricing page and no per-application rate. No tier structure appears either. Ask whether the price runs per application scored or per funded loan. Ask too whether a retrospective validation on your historical loans is chargeable. That test is the only way to verify the accuracy claims before signing.
