The problem
Credit teams rarely lack data. The scarce thing is the hour it takes to turn it into decisions.
Dear credit lead,
Your analysts are not behind because they are slow. They are behind because the work of a portfolio review is assembly: pulling exposure, utilization, payment behavior, and collector notes into one view before judgment can even start.
The data sits in more than one place. The ERP, the warehouse, aging exports, dashboards, collector notes, and analyst memory. Assembling one view of it takes longer than reading it.
A dashboard shows balances. It does not explain what changed. And deterioration is gradual: utilization creeps, payments stretch a few days at a time, and weeks of small movements hide inside averages until the quarter notices.
That assembly is the part worth fixing first. Not the analyst.
Ibrahim, Hearth Advisory
Principles
Warmth is engineered. The ember, not the blaze. Structure protects.
Warmth is engineered
A hearth is masonry before it is comfort. The work starts with the extract and the field mapping, and lets precision produce the calm. Never the other way around.
The ember, not the blaze
One small, intense point of capability inside a calm structure. A review that surfaces the four names that matter beats a dashboard that glows everywhere.
Structure protects
Human in the loop is not a policy page. Every judgment is labeled as analyst inference, and your credit professionals make every decision that follows from it.
The diagnostic
The Credit Portfolio Diagnostic.
This is the first thing Hearth sells. You pick one set of customers and send the export your system already produces. What comes back is a written review, not another dashboard to check.
Where the money actually sits
Not one number for the whole book. Which customers owe you, how old that money is, and whether it is getting older. A single average hides the group that is sliding.
- open AR and past due 30+ / 60+
- aging migration between periods
- concentration and largest exposure changes
- newly deteriorated against improving accounts
Which decisions nobody revisited
A credit limit is a decision someone made on a date. This shows which of those decisions has not been looked at since, and which customers have quietly outgrown them.
- credit limit utilization
- over and under limit accounts
- accounts over limit with no hold decision on file
- last review date against current exposure
Who is paying slower than they used to
The loud problem is a customer who disputes an invoice. The expensive one is a customer who simply starts paying a few days later each month and says nothing about it.
- average days delinquent and weighted days beyond terms
- last payment recency
- week over week AR movement
- remittances beyond terms with no dispute code
Something you can hand to a committee
A short written review your leadership can read in one sitting, plus a page on each customer that matters. No login to maintain, nothing to install.
- executive summary
- account level briefs for the names that matter
- consecutive weeks flagged watchlist
- caveats, confidence, and recommended analyst follow up
Scope
Real questions, not vague transformation.
Every line here is something a credit team already argues about on a Thursday afternoon.
How the review runs
Watch one review get built.
A quarterly re-check on the demonstration book: the analyst's side of the conversation on the left, the checking steps on the right. All of the data is made up on purpose.
Synthetic demonstration data. No real company, customer, or client appears. The review proposes; the analyst decides.
The working desk
Not a mockup. Open it and use it.
The review does not have to arrive as a document. This is a working screening desk built on the same demonstration book: 48 customers, twelve months of history, a flagged review list, a sortable book, and a page on every customer. It runs in your browser with nothing to install and no connection to anything.
Puts the eight names that matter on one screen
It applies a set of review rules to the whole book and shows you what tripped them: over the approved limit, on hold, orders blocked, rating worsened, balance going late, or nobody has looked in a year. Each flag says which rule and why.
- a review list, ranked
- a sortable book of every customer
- a page per customer with aging, trend and history
- a signal feed for what changed outside the ledger
The same desk, pointed at your export
Nothing here is specific to the demonstration book. The rules, the thresholds and the risk scale are placeholders set for the demo, and on a real engagement they would be yours. What stays the same is the discipline: the numbers are computed from the file, and they tie out.
- your fields, mapped and written down
- your rules, not these ones
- your risk scale, not this 1 to 10
- every figure traceable to the export
Open the screening desk Talk through it on a call
Every company, balance and rating in the desk is fictional. It is a demonstration of what gets built, not a record of anyone's book.
Method
Read the file. Study the book. Write the decisions down.
Three steps, a fixed scope, and no access to your systems beyond the file you choose to send.
Read the file you already have
Start from an export you already trust, from NetSuite, Snowflake, or whatever your system produces. Work out exactly what each column means, then check the totals. Every number has to tie back to that file.
Study the book, not just the balances
Who owes what, how old it is getting, who is pressing against their limit, and who has started paying later. Groups and changes over time, because a single average hides the group that is sliding.
Write the decisions down
A short summary for leadership, a page on each customer that matters, what I am unsure about and why, and what I would look at next. Your team makes every actual decision.
| Bexar Flowline | utilization 112.5%, over limit 12 months, last review 2024-11-15 | FLAG |
| Harlow Marine | remittances arriving beyond terms, no dispute codes | WATCH |
| Portfolio | past due 61+ at 10.1%, up 2.7 pts since March | REVIEW |
| Exceptions | 5 accounts over approved limits, 3 with no hold on file | OPEN |
| Reconciliation | every published figure tied to the source extract | TIES |
Drafted by script, reviewed by an analyst. Synthetic demonstration data. Figures reconcile to the demonstration dataset.
Boundaries
Careful where it counts. Clear about what it is not.
The limits are part of the product. Stating them plainly is how a credit buyer decides whether to spend fifteen minutes.
Not automated underwriting
No credit decision is made by software, and none is promised. Every judgment is labeled as analyst inference.
Not a staff replacement
This is analyst leverage: faster review cycles and cleaner escalation narratives for the people you already trust.
Not a data platform
No system access is required beyond the extracts you choose to share. There is nothing to install and no login to maintain.
Not a promise of outcomes
No savings figure, no ROI claim, no detection guarantee. The deliverable is a reconciled review and the decisions it puts in front of you.
The demonstration portfolio is fully synthetic and labeled as such. Every figure in it reconciles to its source data by script, and the report fails to build if it does not. That is how the demonstration pack is built, and it is how client work gets built.
no client names appear on this page, because none are claimedWhat it surfaces
What a finding looks like.
Three from the demonstration book, in the order they would reach a committee. Every number is made up on purpose and ties back to the demonstration data.
The average hid it
Money more than sixty days late reached 10.1% of the book in June, up 2.7 points since March. Inside one group of customers it went from 8.8% to 17.6%, while the headline number for the whole book barely moved. That is how a problem stays invisible for a quarter.
Synthetic demonstration data.
This one is about your process, not your customer
Five customers are over the limit you approved. For three of them there is no decision on file about it either way. The worst is using 112.5% of its limit and has been over for twelve straight months, against a review last dated 2024-11-15.
Synthetic demonstration data.
The quiet one
Payments started arriving late with no complaint attached. Nobody is disputing an invoice, the money is simply slower. That is the one worth a phone call while the conversation is still an easy one to have.
Synthetic demonstration data.
Other work
CreditIntel is the first product. It is not the whole practice.
Hearth Advisory is the practice. CreditIntel is what it leads with, because credit and receivables is the work I know best and can prove fastest. The way of working underneath it is not specific to credit.
The same four moves, a different subject
Take the data a team already has. Work out what the fields actually mean. Check the numbers so people can trust them. Turn the result into something a person can act on. Credit is where I apply it first, not the only place it applies.
Work that is still assembled by hand
Usually it looks like one of these:
- a report someone rebuilds by hand every month
- data living in two systems that must be stitched together first
- a review that only happens when one person remembers
- numbers nobody fully trusts, so meetings argue about the data
Plain, boring, and yours
Pipelines and reporting built with SQL, Python, and the systems you already run, such as NetSuite and Snowflake. Nothing exotic, nothing you need me to keep running for you.
Whether it fits is a conversation
I will not tell you in advance that your workflow can be automated. For industrial teams I can look at where finance handoffs create avoidable delay or risk, but that is a discovery conversation, not a promise made before I have seen anything. If it is not a fit, I will say so.
Who runs it
The work is mine to answer for.
I am Ibrahim. I work in credit, receivables, and portfolio risk. The job I have done for years is taking messy finance data and turning it into a clear view of who owes what, how they are paying, and what has changed, with NetSuite, Snowflake, SQL and Python doing the heavy lifting.
Hearth is an owner led practice, which means when you hire it you get me. There is no account team and no handoff to someone junior. That is a limit on how much I can take on, and it is also the point.
Where slower payment or a few large customers could matter to the finance chief, the review says so plainly. That is decision support for leadership, not an attempt to run your treasury.
The demonstration book is made up on purpose. I would rather show you a clean invented example than imply a client I do not have.
- Someone who has done credit and receivables work for a living, not a generalist reading about it
- Numbers that tie, or a report that does not ship
- The same format every time, so the second review is comparable to the first
- A person signing off on every judgment, by design
- A fixed scope written down in advance, not an open ended engagement
Start here
Start with one portfolio that is hard to explain.
The Credit Portfolio Diagnostic is a fixed scope review of a selected portfolio extract. You get a reconciled picture of the book and the decisions it puts in front of you.
See the format before anything else
Fifteen minutes on screen, walking the demonstration portfolio. It is synthetic and labeled as such. If the format is not useful you will know by minute five.
Request the sample diagnostic Book a 15-minute walkthrough Questions or a walkthrough: hello@hearthadvisory.com