Embedded financing, credit decisioning, fraud prevention and collections for fintechs, payment providers, marketplaces, lenders and insurers. Designed and deployed by a team that has built them at scale.
“Liquidity Lab helped us optimise our DSO in record time. They also allowed us to converge as a Board on the right blend of factoring, credit insurance and in-house collections.”
“Liquidity Lab helped us to roll out a new counterparty credit risk framework. The team’s deep technical knowledge combined with their hands-on practical experience made them a great partner for us.”
Firms launching or scaling a B2B finance proposition
B2B payment providers and fintechs adding pay-later or trade credit. Marketplaces and platforms embedding financing at checkout. Lenders and banks building B2B propositions. Credit insurers, factors and receivables funds digitising origination, underwriting and collections.
Built by our team
100k+
B2B clients and over 500 million euros financed at the B2B pay-later provider our cofounders built and ran.
Fraud reduction delivered
100x
Fraud rate cut from 5% to 0.07% at a B2B fintech under active attack, with 98% of buyers approved friction-free.
Conversion from instant decisions
2x
Sales conversion doubled when trade credit decisions moved from a 48-hour manual check to a real-time engine. Buyers offered terms spent 48% more, with a 3.6x higher repeat purchase rate.
Why work with us
We have built and run these products ourselves
Our cofounders built Hokodo. Our data lead ran fraud analytics globally at Klarna and was Chief Risk Officer at Hokodo. Our technologist was CTO of Hokodo and of the Allianz Trade Digital Agency. We are consultants, not a vendor: we do not sell software, lend or take balance sheet risk, so our advice on what to build, what to buy and who to fund with stays unbiased.
Design
Credit policy, underwriting rules, limits and pricing grounded in real payment behaviour data.
Funding and risk transfer options: credit insurance, forward flow, factoring and credit funds from our network.
Operating model, controls and the business case your board and funders will sign off.
Deliverable: a product and risk design ready to build
Deploy
Senior engineers who ship the decision engine, fraud stack and collections agents into your platform.
Models and agents that keep learning from your own data after we leave.
Your risk and data teams trained and in control of every model and rule.
Deliverable: live components, governed and owned by your teams
What we build with you
Six components, each proven in production
Embedded financing
Credit policy, underwriting rules, pricing, funding structure and risk transfer, plus the operating set-up needed to launch a trade credit or pay-later product at checkout.
Credit decision engines
Machine learning scorecards combining bureau data with actual payment behaviour. Dynamic limits, real-time exposure monitoring and automated limit adjustments instead of static, quarterly reviews.
Fraud prevention
Three-layer defence with risk-based friction: device fingerprinting, behavioural scoring and step-up identity verification reserved for the high-risk minority of buyers.
Agentic collections
Dunning strategies, AI agents that handle inbound queries and outbound calls, management of third-party collections agencies and automated remittance reconciliation.
AI workflow automation
AR and AP automation for your own business: cash application, reconciliation between your ledger and your partners, and contract-to-pay integrity on the supplier side.
Build versus buy and vendor selection
Unbiased mapping of the vendor landscape, target architecture, RFP support and integration design. We publish the AI collections vendor map and take no vendor commissions.
Credit decision engines compared with the traditional approach
Credit decision engines compared with the traditional approach
Liquidity Lab
Traditional approach
Data source
External bureau data plus actual payment behaviour
External bureau data only
Update frequency
Real-time exposure monitoring with automated data feeds
Quarterly, based on 6 to 18 month old accounts
Credit limits
Dynamic, auto-adjusting to payment behaviour
Static, manually reviewed
Early warning
Ongoing score refreshes and automated limit adjustments
You find out when they stop paying
Focus
Fix the root cause of long DSO before selling the receivables
Factoring implemented to accelerate cash, overdues still lead to costly financing
How an engagement runs
From design to live components in one to two quarters
01
Diagnostic and design
1 to 4 weeks
Product, risk and data review on your real transaction data. Target architecture, credit and fraud policy, funding options and a business case with a sized opportunity.
02
Pilot
4 to 8 weeks
Deploy a first model or agent on a limited segment and measure it against your current process: approval rate, fraud rate, recovery rate, cost per case.
03
Scale and embed
4 to 12 weeks
Roll out across the portfolio, integrate with your platform and partners, train your risk and data teams, and hand over governed components with named owners and full documentation.
Trusted advisors
We are software agnostic. We do not sell a product and never take a commission from vendors or funders we introduce, so our recommendations are unbiased. We help you navigate build versus buy decisions.
We focus on outcomes. We know our trade and we are comfortable putting our fees at risk. We get paid when results are delivered.
What makes us different
Three things traditional consulting does not do
01
We generate value fast
Within three weeks we start implementing changes. You will see measurable results within six weeks. Traditional programmes take 12 to 15 months to deliver results.
02
Our engineers bridge the PowerPoint gap
Most value creation plans fail between the recommendation deck and the implementation. We build the credit models, AI agents and data pipelines ourselves. They run in your systems under your controls, owned by the organisation rather than any one person, and keep learning from your data after we leave, so the value does not walk out of the door with the consultants.
03
Prevention, not just cure
Factoring and collections treat symptoms. We also fix root causes: who gets payment terms, on what limits, and how exposure is monitored, so overdues do not arise in the first place. A healthier ageing balance means 5% to 15% lower DSO and cheaper financing if you factor receivables.
Results
Numbers from real engagements
Anonymised at our clients' request. References available on the call.
Fraud prevention
B2B fintech onboarding 4,000 new buyers a month
0.07%
fraud rate, down from 5%
98%
of buyers approved with zero friction
> €600k
of losses stopped within weeks
Real-time, risk-based verification at point of sale: device fingerprinting, behavioural scoring and step-up identity checks only for high-risk buyers. Adopted as the standard fraud stack for future merchant roll-outs.
Automated credit decisions
B2B services provider, trade credit offered at point of sale
2x
sales conversion
+48%
average spend
3.6x
repeat purchase rate
4 to 5x
customer lifetime value
Instant trade credit approval shown to sales reps during the first order call. 85% of customers are now offered credit immediately, up from a third via a 48-hour manual process. Conversion doubled from 15% to 30%.
Agentic collections
Wholesaler serving 10,000+ retailers a month
90%
lower cost per case
70%
collection rate on small debts, up from 45%
8.5 / 10
customer satisfaction
AI agents answer customer emails and texts, place outbound calls and handle invoice copies, proofs of delivery and small disputes. On cases above 400 euros they matched the human team at a tenth of the cost.
Questions financial firms ask us
Frequently asked questions
How do you build a B2B credit decision engine?
We start from your data: bureau records, financial statements where available and, most importantly, actual payment behaviour across your book. We build machine learning scorecards on that combined dataset, define limit and pricing rules, and set up real-time exposure monitoring with automated limit adjustments. A first model is typically live on a pilot segment within six weeks, then scaled across the portfolio.
What fraud controls does a B2B BNPL or trade credit product need?
Three layers, applied in proportion to risk: device fingerprinting and network signals at onboarding, behavioural scoring on the order and buyer profile, and step-up identity verification only for the high-risk minority. Done well, over 98% of good buyers pass with zero friction. At one B2B fintech onboarding 4,000 buyers a month, this cut the fraud rate from 5% to 0.07% within weeks.
How long does it take to launch an embedded financing product?
A diagnostic and design phase of one to four weeks defines the credit policy, underwriting, pricing, funding and operating model. A pilot on a limited segment follows and runs for four to eight weeks, then roll-out and integration take four to twelve weeks. From a standing start, a controlled launch is realistic within one to two quarters, depending on funding arrangements and integration complexity.
Should we build or buy our collections and decisioning stack?
It depends on volumes, differentiation and your engineering capacity, and we have no stake in the answer: we sell no software and take no vendor commissions. We map the vendor landscape (we publish an AI collections vendor map), define the target architecture, and run the selection if buying makes sense. Where building wins, our engineers deliver the models and agents alongside your team.
Can AI really handle collections conversations?
For the routine 80% of an AR inbox, yes. Requests for invoice copies, payment promises, proofs of delivery, statement queries and small disputes follow deterministic playbooks. Tool-calling agents resolve these instantly, with full audit trails, and escalate anything that needs judgement. At a wholesaler serving 10,000+ retailers, agents matched human recovery on cases above 400 euros at a tenth of the cost and lifted the collection rate on small debts from 45% to 70%.
Factoring, credit insurance or in-house collections: which is right for us?
Usually a blend, and the right mix depends on your customer concentration, margins, growth plans and cost of capital. Factoring accelerates cash but costs more when your ageing is poor. Credit insurance protects against large losses. Strong in-house or AI-assisted collections improve both. We have no product to sell, so we model the options on your own receivables data and help your board converge on the right combination, as we did with Markit.
How do you work with our existing risk and data teams?
As an extension of them. Your risk team owns the policy and your data team owns the platform. We bring specialist credit, fraud and collections experience and senior engineers who ship. We work in your repositories and tooling where possible, document everything, and train your people so that models and agents keep improving after the engagement ends.
Do you take any balance sheet or origination risk yourselves?
No. Liquidity Lab is a pure consulting firm. We do not lend, insure or factor, and we do not sell software. Where funding or risk transfer is needed, we introduce and structure arrangements with our network of credit insurers, factors, credit funds and brokers, and we help you negotiate terms. Our fees are consulting fees, and in the pilot and roll-out phases they are paid only when agreed results are delivered.
Contact
Let's build your B2B finance stack.
Book a 45-minute introductory call. We will come with questions about your situation and leave you with a first view of the opportunity.
Product and risk review scoped in the first call
Build versus buy view on your current stack
Introductions to our funding and risk transfer network