Wealth DeskplusEnterprise Wealth Tech
Pricing & Licensing

What Credit Decisioning Software Costs a Lender Making 250,000 Decisions a Year

A three-year hypothetical model for a mid-size consumer and SME lender. The software is the smaller cost: data, integration and model upkeep take all-in spend to about $3.75 per decision.

By the Wealth Deskplus editorial team · November 2026 · 8 min read

Why the license quote is usually under 25% of the real bill

The vendor quote on your desk covers the smallest part of what a mid-size lender will spend on a decision engine. In most credit decisioning software pricing negotiations, buyers argue over the platform subscription and the per-decision rate. Meanwhile bureau data, integration work and model upkeep cost roughly three to four times as much in the model below.

This article builds a hypothetical three-year cost model for a lender making 250,000 credit decisions a year. It splits the budget into five lines: platform subscription, per-decision fees, bureau and alternative data, integration, and ongoing model maintenance. In the mid case, the three-year total is about $3.1 million, and software fees make up roughly $644,000 of it. Every figure is a typical 2026 market range in USD. Real numbers vary by contract, region and vendor, so confirm them with vendors before they go into a business case.

Advertisement

The hypothetical lender behind the numbers

Say a lender has a book that's mostly unsecured consumer loans plus a growing small-business line. It runs automated decisions on new applications and doesn't yet use the engine for collections or account management. The assumptions:

  • Volume: 250,000 decisions in year one (200,000 consumer, 50,000 SME), growing 10% a year to about 302,500 in year three.
  • Bureau pull rate: 85% of consumer and 90% of SME decisions trigger a paid bureau call. The rest are declined at pre-screen or reuse a recent pull.
  • Alternative data: 40% of applicants go through open-banking transaction analysis, and every application gets a fraud and identity check.
  • Deployment: vendor-hosted SaaS with production, test and development environments.
  • Team: an in-house credit risk function that already owns its scorecards and wants to change strategies without raising vendor change requests.

At 250,000 decisions a year, a lender is too small for enterprise volume discounts and too big for starter tiers. That in-between position is why pricing for this profile is so negotiable.

Platform subscription: what the base fee actually buys

At this volume, mid-market decision engine pricing usually arrives as an annual subscription somewhere between $90,000 and $400,000. The model uses $180,000 with a 5% annual uplift. Where you land depends less on the logo and more on scope: how many products, how many environments, whether case management and champion-challenger testing are included, and which support tier you buy.

The large vendors don't publish rate cards. FICO Platform pricing and Provenir pricing are quote-based and modular, and the bill grows as you add decisioning, data orchestration, analytics workbenches or fraud components. Experian PowerCurve, Zoot and GDS Link are sold on similar terms. Newer cloud-native engines such as Taktile often position themselves at the lower, usage-based end, though that varies by deal. All of these sell internationally, but local partner pricing and data-residency hosting can change the quote.

Ask every vendor to price the same scope: products, environments, named strategy-editor users and the uptime SLA. Without that, a lower quote usually just means less is included.

Per-decision pricing and the volume-band trap

Most contracts bundle a decision allowance into the subscription and charge overage above it. Per-decision rates beyond the band typically run $0.05 to $0.50: lower for simple rule calls, higher where the vendor also orchestrates several data calls. The model assumes 150,000 decisions are included and the rest are charged at $0.20, so overage is $20,000 in year one. With growth, that rises to $25,000 and then $30,500.

The trap is how a "decision" is defined. Some contracts count every call to the engine, so a pre-qualification, a full application and a re-decision after a document upload can add up to three billable events for one customer. Under that definition, this lender could be billed for 400,000 to 500,000 events a year instead of 250,000, and the overage line would multiply.

  • Get the billable-event definition in writing.
  • Ask whether test and sandbox calls count toward the allowance.
  • Negotiate tiered overage that gets cheaper as volume grows, not a flat rate.

Bureau and alternative-data costs outgrow the software

This is the biggest line in the model, and vendor proposals usually leave it out. You typically contract with bureaus directly. The decision platform charges for the connector, and some vendors add a pass-through margin on data they resell.

Data callAssumed unit cost (USD)Year-one volumeYear-one cost
Consumer bureau report$1.20170,000$204,000
SME commercial report$4.0045,000$180,000
Open-banking analysis$0.60100,000$60,000
Fraud and identity check$0.25250,000$62,500
Total$506,500

Unit costs vary widely by market. Consumer reports can cost well under a dollar in some markets and several dollars in others, and commercial reports with director and payment data can run past $10. Open-banking costs depend on whether your region has a regulated API framework (the UK, EU, Brazil, India and Saudi Arabia each have one in a different form) or relies on aggregators. The real lever is waterfall design: call the cheapest data first and pay for expensive reports only when the decision is still uncertain. A well-built waterfall can cut this line noticeably, and for most lenders it's the largest saving available in the whole model.

Integration work in year one

Implementation is where credit risk software cost estimates most often slip. Vendor professional services for a two-product rollout commonly run $60,000 to $250,000, and internal effort costs about the same or more. That covers engineers wiring the engine into your loan origination system, data teams mapping bureau attributes, and risk analysts rebuilding and testing existing strategies.

The mid case puts year one at $280,000: $150,000 in vendor services, $120,000 in internal time (two engineers and an analyst for about six months), and $10,000 in bureau connector setup. Years two and three each carry $30,000 for new products and connectors. Timelines commonly run 4 to 9 months. Older core systems and multi-country data-residency requirements push you toward the long end.

Model maintenance and governance after go-live

Credit models degrade as applicant mix shifts, bureau scores get re-based and economic conditions change. Budget for monitoring, recalibration and independent validation. The model uses $120,000 in year one (models go live partway through the year), then $160,000 and $165,000. That covers roughly one full-time modeler, monitoring tools and periodic outside validation.

Supervisors expect this work whichever vendor you pick. Examples include the US Fed and OCC's SR 11-7 model risk guidance, the UK PRA's SS1/23, and the EBA's guidelines on loan origination and monitoring. The EU AI Act treats creditworthiness assessment of individuals as high-risk. Regulators in India, the Gulf and Southeast Asia are also raising expectations on explainability and fair treatment. How these apply to your institution is a question for qualified compliance and legal advisers. They add cost either way, and your engine needs audit trails, reason codes and version control built in.

Three-year total and the cost per decision

Cost line (hypothetical, USD)Year 1Year 2Year 3Typical annual range
Platform subscription$180,000$189,000$199,000$90k–$400k
Per-decision overage$20,000$25,000$31,000$0–$100k
Bureau and alternative data$507,000$557,000$613,000$250k–$900k
Integration$280,000$30,000$30,000$120k–$500k (Y1)
Model maintenance$120,000$160,000$165,000$80k–$300k
Total$1,107,000$961,000$1,038,000

Across about 827,500 decisions over three years, that's roughly $3.75 per decision all-in. Take out data and it's about $1.73. Software fees alone come to about $0.78. Vendors compete on that last number, yet it's the smallest of the three. Two platforms whose subscriptions differ by $50,000 a year can end up far apart on total cost if one supports cheaper data waterfalls or lets your analysts change strategies without paid services.

Red flags to clear before you sign

  • Uncapped renewal uplifts. Cap them at a fixed percentage or tie them to an inflation index. Steep uncapped uplifts can erase a first-year discount within a renewal or two.
  • Data markups you can't see. If the vendor resells bureau data, ask for its margin to be disclosed or keep your direct bureau contracts.
  • Paid change requests. If adding a new cut-off or rule needs vendor services, your maintenance cost will rise every quarter.
  • Exit terms. Make sure you can export strategies, decision logs and audit history in a usable format.
  • Hosting region. Confirm where data is processed if you lend under GDPR, India's DPDP Act, Saudi Arabia's PDPL or similar laws.

Decision rule: if a vendor's software fees come to less than about 30% of your modeled three-year total, compare finalists on data orchestration and self-service configuration, not the per-decision rate. Rebuild this model with your own volumes and bureau contracts, then send every shortlisted vendor the same scope sheet.

Quick answers

Do FICO and Provenir publish their pricing?

No. FICO Platform pricing and Provenir pricing are quote-based and depend on which modules you take, your decision volume, how many environments you need and your hosting region. Most other established vendors work the same way. Ask each one to price an identical scope so you can compare them like for like.

Is per-decision pricing better than a flat subscription?

Per-decision pricing suits lenders with uncertain or seasonal volume because cost tracks usage. A flat subscription with a generous allowance is usually cheaper once volume is predictable. Either way, the definition of a billable decision matters more than the headline rate, because pre-qualification and re-decision calls can double the count.

Should bureau data costs be part of a decision engine business case?

Yes. You'll pay for bureau data under any platform, but platforms differ in how well they support cheaper data waterfalls, caching and conditional calls. In a model like this one, data is the largest cost line, so a platform that trims data spend by even 10% to 15% can save more than the gap between two subscription quotes.

Written by the Wealth Deskplus editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.