How AI Impacts Valuation: Lessons from Global PE Transactions and Polish Reality
Picture the scene: the board presents a fund with a "comprehensive AI strategy." The dashboard looks impressive, the slide deck is 40 pages long, and the consultant was expensive. A PE partner asks: "Okay, but how much does this add to EBITDA?"
Silence.
It turns out that 36% of PE funds that declare an "AI strategy" don't have a single KPI to measure its impact on company value – according to an FTI Consulting study from December 2024. Not a single one. They have a strategy, a budget, even a team – but they have no idea if it actually works.
And here begins the real problem with valuation. How do you value something you can't measure? During due diligence, the fund will see your AI, nod its head, and then ask about margin improvement, revenue per employee, customer acquisition cost. If AI doesn't translate into any of these numbers – it's not a competitive advantage. It's just a cost.
In Poland, we have an additional twist: the PE market is growing like crazy (50% more transactions in 2024), tech now accounts for 22% of all deals, but publicly available data on how AI affects Polish valuations? Zero. There are no case studies, no benchmarks, not even rumors at industry events. Either no one is measuring it, or those who are – prefer not to talk about it.
"AI washing" in portfolio companies. How funds recognize apparent value?
During due diligence, the fund doesn't ask "do you have AI" - it asks for numbers and checks five things:
- Problem to solve – what specific business problem AI is supposed to address (not "we will improve efficiency" but "we will reduce customer service time from 4h to 30 min")
- Cost vs. outcome – how much you spent, how much revenue or savings it generated (lack of numbers = red flag)
- Governance – who is responsible for AI at the C-level, how do you measure progress, who makes decisions about scaling (if "the IT team handles it" = problem). A fund that sees red flags will lower the valuation or walk away.
- Data infrastructure – do you have organized data - without it, most pilots fail
- Production vs. pilots – are real customers using AI, or have you been "testing for 18 months"
3 due diligence questions that verify real AI solutions
Question 1: "What specific business metrics has AI improved in the last 6 months?"
Real answers: "conversion rate increased from 2.3% to 3.8% thanks to offer personalization" or "reduction of quality control errors from 5% to 0.8%, saving PLN 200k annually" or "customer service time dropped by 40%, we handle 30% more orders without additional staff". Bad answer: "we improved process efficiency" (zero specifics). The fund looks for impact on revenue, margins, CAC, retention or capacity – if you can't show numbers, AI isn't working.
Question 2: "What's your worst-case scenario if AI stopped working tomorrow?"
This checks whether AI is in the critical path of the business or a nice-to-have. If the answer is "we would go back to the previous process" – AI is not essential. If "we would lose the ability to serve 40% of customers" or "our product would stop working" – it's a core business capability. The fund only pays a premium for the latter.
Question 3: "Who on the team actually understands how this works and what will you do if that person leaves?"
If all the knowledge resides in the head of one data scientist who might leave for a competitor – you have a problem. The fund looks for: do you have documented processes, can the operational team manage the system without a data science team, is it a black box or an operationally managed technology?
From Strategy to Implementation: When Investing in Artificial Intelligence Makes Sense
AI makes economic sense only if:
- you see a specific problem that is costing the company money and AI will solve it cheaper than alternatives,
- you can calculate before implementation whether the investment will pay off in 12-18 months,
- you have the people and infrastructure to implement it and measure the results.
If any of these three are missing – don't invest. "We're doing AI because the competition is doing it" is the worst possible reason – you'll have a cost in the P&L, zero measurable effect, and during due diligence, the fund will catch it in the first week.
Tests: before spending 300k on AI, answer three questions: (1) what financial number will improve and by how much (margin? CAC? time to close?), (2) how will we measure this monthly (specific dashboard, not "we will monitor"), (3) who at the C-level takes responsibility for this (not "the IT team"). Lack of concrete answers = you are not ready.
Companies that implement AI "because they have to" end up with a burned budget and a pilot that lasts 18 months. Companies that calculate ROI before starting – either don't start at all, or achieve quick wins in 6 months.
What a PE fund should see in a company: an AI roadmap building value before exit
Year 1-2: Quick wins that will be in EBITDA before exit. Automation of specific processes with measurable savings. This must be in production and show results by year 3 at the latest, so the buyer sees a trend in the last 2-3 financial years. Without this, AI is just a cost with no proven return.
Integration with value creation thesis. If the fund bought the company to "scale sales in new regions," the AI roadmap must show how AI supports this goal – e.g., automating lead scoring, predictive analytics in new markets. If the AI roadmap lives its own life and doesn't support the main value creation plan – it's a hobby project, not a strategic asset.
Plan B if it doesn't work. 95% of AI pilots fail – the fund knows this. That's why it asks: what if this AI project doesn't yield results in a year? Do you have a backup plan? Alternative use cases? Or will you burn 500k and say "it's a long-term investment"? Companies that only have plan A without a plan B – red flag.
Quick wins vs. transformation: which path increases revenue faster
Quick wins are projects completed in 3-6 months: customer support automation (chatbot handles 50% of inquiries), AI in pricing (dynamic pricing based on demand), predictive maintenance (reduction of machine downtime). Low cost (50k-200k), low risk, quickly measurable effect. Problem: value ceiling – you can save 300k annually, but you won't change the business model.
Transformation is redesigning the core business: shifting from reactive to predictive service, real-time product personalization, AI-driven new product development. High cost, 18-24 month timeline, high risk of failure. But potential: change competitive positioning, higher valuation multiple, new revenue streams.
For Private Equity: quick wins almost always win for three reasons. First: timing – the effect is visible before exit. Second: storytelling during sale – "we implemented 5 AI use cases, each yielded ROI, there's momentum" sounds better than "we are in the middle of a 2-year transformation." Third: risk/reward – in a PE portfolio, 5 small wins are better than 1 big experiment that might fail.
Exception: if the buyer is a strategic player who pays for potential (not a financial buyer), transformation can yield a higher multiple. But that's a gamble, not a certainty.
2026 in Private Equity: how investor expectations for technology are changing
No more "AI experiments." AI fragmentation is ending – companies are moving to top-down, AI programs targeting a few high-impact workflows, not 20 pilots with no results. 53% of PE funds are recruiting more digital transformation specialists, 51% are looking for data scientists and AI experts – not to "have AI," but to actually implement it in portfolio companies.
In 2026, the fund expects a concrete AI strategy: the company knows its AI readiness (data quality, infrastructure, team competencies) and has a plan to use AI for EBITDA – not "maybe someday," but "in Q2 we automate X, saving Y." The problem in the Polish market: most companies don't know their AI maturity. They haven't assessed whether their data is even ready for AI, whether the team has the competencies, whether processes can be automated. A fund that sees this knows: the company is not ready for an AI investment, but for a costly learning process.
AI governance is becoming non-negotiable – Private Equity assesses AI risk during due diligence and verifies the level of risk. Ignoring this can lead to regulatory penalties and delays in exit.
For Polish companies, this means: if you plan an exit in 2026-2027, the fund expects AI to already be in production and showing numbers. "We will do it" is not enough – the buyer will compare you with competitors who already have it.
Check if your company is ready for AI (link)
