%402x.svg)
For portfolio companies with USD 20 million to 1 billion of revenue.
We audit the company's systems, data, reporting, security and team in six areas, trace one or two real business questions from source to answer, and rate every finding by severity with the effort to fix it. The cost of bringing the data platform and governance up to the market average goes into the deal model before signing. For a target that sells AI, we check what actually runs in production, on what data, and whether its effect is measured.
We review the company's processes and mark the ones data or AI can improve, each with an estimated return and the effort it needs, ranked from the highest return. At one regulated online business, 11 of 29 processes qualified for AI; nine went to plain automation and ten were parked.
We take the highest-return cases into production and leave a working solution in the company's own systems.
We lay out the plan for the whole holding period, help hire the people on the company side and set up the processes, so the company carries it forward on its own.
Shared KPI definitions and one monthly pack for the fund, fed from each company's own systems, so holdings can be compared on the same terms.
Numbers a buyer's diligence can rely on: reconciled to source, with documented definitions, lineage back to the posting and an audit trail.
How the data and AI audit runs, what the fund gets and what it costs
A data and AI due diligence before signing reduces both risks: it shows whether the company can keep up, and it prices the cost of fixing its data into the deal model.
Apollo Global Management hired an operating partner in mid-2021 who now heads its Data, Digital and AI team. Its AI system looks across purchasing contracts and invoices for more than 40 portfolio companies. In one of them, an analysis of 15,000 software purchase agreements delivered a procurement cost reduction of more than 65%.
A mid-market fund with 9 to 12 portfolio companies can run the same logic at its own scale: the approach, the people and the patterns are built once and reused in every company.
Most portfolio companies between USD 20 million and 1 billion of revenue are too big to run on spreadsheets and too small to hire a Big Four firm for every data question.
Before the shift starts, the plant manager sees output against plan, which machine stopped and why, and what failed quality control. It used to wait for month-end.
No shared KPI definitions. Plant results could not be compared, and reading the data took expert knowledge.
One model joining machine sensors, actual production from the ERP and the production plan, with daily, weekly and machine-efficiency reports.
USD 3.26M a year at full maturity, after USD 351k in year one and USD 990k in year two, from material yield, first-pass quality and machine utilisation.
What it means for a portfolio company: with EUR 100 million of revenue, raw materials typically cost EUR 50 to 65 million, so one point of material yield is EUR 500k to 650k a year straight to EBITDA, and about EUR 4 to 5 million of value at an 8x exit multiple.
Client's business case, prepared before the build. The portfolio company example is illustrative.
A recommended price that a sales rep can override with one click is only a suggestion, and the difference leaks out of the margin.
The system suggested a price, and sales reps regularly went below it. Nobody saw how often, by whom or by how much.
Every deviation from the recommended price became visible to everyone who sets prices, and we made sure people used it. We deliberately kept the option to go below the price: visibility and ownership of the result did the work.
EUR 18M of new margin in three years, on about EUR 1M to build the tool and about EUR 1M to roll it out: about 9x.
Margin measured after completion; effort estimated.
Nobody could answer a simple question: why is this bestseller not on the shelf?
Seven tools calculated the same KPIs, each differently. Data was pulled by hand, and finding the cause of a stock-out took days and usually ended without an answer.
One automated flow of sales, warehouse and delivery data in one model, replacing a fifteen-year-old database. A few reports with one definition per KPI, and the cause of every stock-out visible.
EUR 2.6M of margin recovered in three years, on about EUR 400k: ROI 550%. EUR 600k in the first year and EUR 1M a year after.
Business case.
The same variance analysis, repeated by hand in every subsidiary, four times a year.
Analysts in each subsidiary calculated and explained variances by hand, and reviewers checked them. The knowledge of what matters sat in people's heads and left with them.
On the existing data warehouse, playbooks for every line of the statement: algorithms, materiality thresholds and alert words. Institutional knowledge written down and turned into the company's own asset.
Up to 1,552 working days a year: 557 in the cautious scenario, ROI 104% over three years, and 1,552 in the target one, ROI 391%.
Business case scenarios.
The foundation came first, and AI went only where it paid back.
One customer view shared by sales, finance and compliance, refreshed before the day starts, and segmentation in production with a control group, so its effect on results is visible.
We reviewed 29 processes across five teams and priced each one. Only 11 suited AI; nine went to plain automation and ten were parked.
CHF 120k a year saved in customer service, and CHF 230k of regulatory exposure reduced.
Business case.
So we know how this decision looks from the owner's side.
There is a moment every leader knows. You ask a simple question, "What are our margins this quarter?", and two weeks later three departments send three different answers.
I have spent 15 years inside complex organisations across Europe and beyond, and I have seen how much time, money and energy is lost when data is scattered, inconsistent or simply not actionable. I co-founded Astral Forest to make data work for the people running the business.
Today we help CEOs, CFOs and business owners in fintech, manufacturing and professional services to:
One client used to spend 2,000 person-days a year producing a quarterly financial report. After working with us, every employee gets the answer they need in 15 seconds.
If your data is your biggest untapped asset, or your biggest hidden cost, let's talk.
For more than ten years I have led BI and data programmes for large international companies. What decides their success is people: I like building a team around a shared objective and keeping it there until the result shows in the numbers.
My working day mixes several roles:
I have lived ten years abroad, in four countries, so working across cultures and languages comes naturally, and that matters when a client's teams sit in several of them.
We work from Poland for clients on the US East Coast, in London, France, Switzerland and the Netherlands, in Polish, English and French. Our typical client is a group with USD 1 to 5 billion of revenue; a portfolio company gets the same practice at its own scale and price.
An audit, a list of use cases priced by return, and a decision on what comes next.
We reply within one business day.
