Data Foundation Assessment That Leaves One Automation Running
Count where one person’s last four weeks went, then see what a four-week diagnostic would build first.
A data foundation assessment checks whether your records can be trusted before anything is built on them. The Data Foundation Diagnostic does that in four weeks, by counting where the days go and leaving one working automation in production. KORE1 runs it with consultant Khurram Tehseen and then staffs whoever keeps the result alive, and of the people we place, 92% haven’t left a year later.
Last updated: October 1, 2026
Most data assessments end in a score. Governance a two, architecture a three. Sound familiar? Then comes a roadmap with eighteen months on it, and at the end of the first month nothing new is running.
This one doesn’t. It ends with something switched on. It reads an ordinary month of the people who do the work, finds the days that went to hunting for records, re-keying them and making two systems agree, and builds one automation against the biggest block. The method is Khurram’s, it sits underneath KORE1’s private credit data operations consulting and our data work for life and annuity carriers, and the people who own the result afterward come through KORE1’s data staffing practice.
Count One Month Before Any Data Foundation Assessment
Pick the person who’d get the call if a lender, an LP or an auditor asked a hard question tomorrow, take their last four weeks, which is twenty working days, and set how many of those days went to each line. Whatever is left is the job they were hired for. The sheet opens on an example month, not a benchmark, so change every line.
Days out of the last twenty
- Finding itWhich file, which system, which version is current 3
- Keying itDocuments and statements typed into workbooks 5
- Reconciling itMaking two systems agree about one record 4
- Answering for itOne-off questions from lenders, LPs and auditors 3
Left for the job itself 5 of 20
How often the biggest line comes round
What a diagnostic could count on
- FFinding it3
- KKeying it5
- RReconciling it4
- AAnswering for it3
- Hand work15
- JThe job itself5
- Working days20
- Left for the job25%
Keying it, 5 days
Your largest hand line. The first build is usually one document type read into checked fields, so nobody types the same page twice.
Yes, if a month-end falls inside
It gets scored on last month, which already finished. The live run is the next cycle, so start the four weeks where they cover it.
Ready, with a gap
The person and the access are enough to begin. With nobody to name an owner, the automation ships and then has no one to keep it.
Hand work 15 of 20 days, 5 left for the job. Aim the automation at: Keying it.
Nothing leaves this page. The sheet only does arithmetic on your own days, so it never prints a saving or an industry average. A real diagnostic replaces your estimate with a count taken on your files.
A count is dull to look at. Strokes on a sheet, one per day, next to the paper those days were spent on.
Why This Data Foundation Assessment Counts Days Instead of Scoring Maturity
Maturity models have their place. The EDM Council’s DCAM, the one most financial firms reach for, arranges data management into eight components, 34 capabilities and 101 sub-capabilities in its third version, each with evidence to collect and criteria for scoring. When a board or a regulator wants a rating, that’s the right instrument. Use it.
A rating won’t tell you what to build on Monday, though. It says governance is weak. It doesn’t say that your senior analyst lost four days last month matching the administrator’s borrower list against the loan system by hand, or that those four days come back every month.
Finance teams know the feeling. In the 2025 FP&A Trends Survey, 46% of FP&A time still went to collecting and validating data rather than analysis, and nobody in those teams needed a maturity score to notice. What they tend to lack is a count taken on their own files, by line, that somebody outside the team wrote down while the work was happening, instead of reconstructing it from memory a month afterward.
So the diagnostic counts. Old method. Industrial engineers call it a time study, and it works on a credit operations desk the same way it works on a factory floor.
The Figures Worth Knowing Before You Start
- 4 weeks First count to one automation in production The fixed length of the Data Foundation Diagnostic
- 46% FP&A time spent collecting and validating data 2025 FP&A Trends Survey, reported July 2025
- 101 Sub-capabilities a full DCAM v3 review scores EDM Council, DCAM framework page
- 92% KORE1 placements in the same job after a year Across every KORE1 desk, measured at twelve months
The survey figure covers planning teams in general and isn’t a private credit number. We haven’t found a published one for credit operations, which is part of why Khurram is interviewing 25 operators for the Private Credit AI Readiness Benchmark. Until that lands, the only count worth trusting is your own.
Inside the Four Weeks of a Data Foundation Diagnostic
Four working weeks. Five days each. The boundaries shift a little from firm to firm. Two things don’t move, the end date and the one automation running on it.
-
Week 1
Count
We sit with whoever does the work and log where one finished cycle’s days went, line by line, on the files they actually used.
-
Week 2
Trace
One borrower, policy or claim gets followed through every system that holds it, and we write down which source wins each field.
-
Week 3
Build
One automation gets built against the biggest hand line, then scored on a cycle that already finished, so there’s an answer key.
-
Week 4
Run
It goes into production with a named owner. You get the written read and the order of what to build next.
Week four is a handover. A few pages, one thing running, and a name next to it.
What the Diagnostic Hands Back in Week Four
Three things. None is a slide deck. First, the read. A written count of where the days went, by line and by record, with the files it was taken on listed beside it, so that anyone who disagrees with a number can go and check the same files.
Second, the automation itself, in production, with its rules written in plain language and its score against the finished cycle attached. Third, the order. Which build comes next, which one after that, and what each needs before it can start.
Then comes the part most assessments skip. Someone has to own it. Khurram defines a data foundation as a single canonical record for each borrower, policy or claim, plus rules for which of its fields can be trusted. A record like that goes stale the moment nobody is answerable for it.
Two seats usually take it on. A governance analyst keeps the definitions and the rules, and data governance analyst staffing is a search we run often. An engineer keeps the pipelines running, hired through data warehouse engineer staffing, often as contract staff for the first few months. And where data has no senior owner at all, a fractional chief data officer usually comes first.
What the One Automation Usually Turns Out to Be
It depends on which hand line is longest. Four answers cover most firms we see.
-
Finding it
An authority table
One row per field naming the system that wins, so the answer stops living in one person’s head.
-
Keying it
One document, read once
A single document type read into checked fields, the mechanic behind document-to-decision automation.
-
Reconciling it
A nightly tie-out
Two systems compared nightly once the same borrower is matched across them, with only the breaks sent to a person.
-
Answering for it
One question, answered once
A recurring investor question answered from the record, which is where LP reporting data work begins.
Is the month-end the real trouble? The Close Replay diagnostic reads a close stage by stage instead. And a dashboard is never the first build, for reasons Khurram gives in his piece on build order. Short on hands for the build itself? Data engineering staff augmentation adds them by the month.
Common Questions
What is a data foundation assessment?
A data foundation assessment is a check on whether the records a business runs on are complete, consistent and owned well enough to build on. Some score maturity against a framework. The Data Foundation Diagnostic counts where working days go and proves the finding by shipping one automation.
How long does a data foundation assessment take?
Four weeks for this one. That covers the count, the trace of one record through your systems, and one automation built, scored and running. Framework-based reviews often run longer because they score every capability, and rebuilding a whole data foundation is a separate, longer piece of work that the diagnostic only puts in order.
How is this different from a data maturity assessment?
The question is different. A maturity assessment asks how developed your data management is and returns a rating by capability. This asks where the days went last month and returns a count, plus one thing built. Plenty of firms need both.
What counts as the one working automation?
Something that runs in production on your real files without a person retyping anything, and that has an owner. A prototype doesn’t count. Neither does a demo on sample data. It has to be scored first against a cycle your team already finished, which is how anyone knows it works.
What does the diagnostic need from our team?
Mostly one person’s time. The person who does the work has to be reachable for part of each week, somebody has to grant read access to the systems, and a decision-maker has to name who owns the automation once it’s live. A finished cycle with its files kept helps a lot.
Is the diagnostic only for private credit and insurance firms?
No. That’s just where it was built. Khurram developed it across private credit funds and life and annuity carriers, where borrowers, policies and claims live in several systems at once, and it carries over to any operation that runs on recurring records and still does a good share of the work by hand.
What happens after the four weeks?
Usually a hire or two. The automation needs an owner, and the next builds need people, so most firms add a governance analyst and an engineer on contract first and decide about permanent seats a few months later. KORE1 fills both, and across IT roles our searches average 17 days to fill.
Name the Person Whose Month You’d Count
Tell us who does the work and roughly where their last four weeks went. A guess is fine. You’ll hear plainly whether a diagnostic would find enough to justify four weeks, and if the real gap is a seat, you’ll hear that instead.
Set Up a Diagnostic Call →Want the method from its author? Khurram Tehseen is on LinkedIn.
