Private market transactions are negotiations. From the first page of the information memorandum to the last line of the closing accounts, the entire process is positioning conducted in the language of finance. Everyone in the industry knows this. We just keep calling it "analysis" and "diligence" and "valuation," as if naming it differently changes what it is.
Now, "something" is shifting. The analytical labor that used to fill the space between "we have a deal" and "we have a price" is compressing. And as it compresses, the negotiation underneath becomes harder to ignore.
How analysis works in practice
A sell-side advisor preparing a CIM makes dozens of choices that have nothing to do with accuracy. Which comparable companies end up in the peer group. How the addressable market gets defined. What time period best shows the growth trajectory. Every choice is defensible on its own, and every choice tilts the narrative toward a higher or lower number. That's not a flaw in the process. That's the process.
On the buy-side, it runs in reverse. Financial DD identifies adjustments that shrink normalized earnings. Commercial DD pokes holes in the revenue projections. Legal flags contingent liabilities. Each finding arrives framed as objective analysis, and each one strengthens the buyer's hand at the table. The DD report looks like a stack of facts. It is a stack of facts. But those facts got selected, weighted, and arranged with a specific outcome in mind, and I think most practitioners would admit this. This is the system working as designed. DD exists to give both sides the information they need to negotiate. The strange part is that we've wrapped this process in the language of neutral truth-finding, and at some point, some of us started believing the packaging.
What the numbers show
Take EBITDA adjustments. A seller's QoE report adds back management compensation above market rate, a facility relocation, and an ERP implementation. Total add-backs: four million on a base of twelve. The buyer's advisor accepts the comp adjustment, disputes the relocation because it's happened three times in eight years (not obviously "one-time"), and reclassifies half the ERP spend as maintenance. Same data. Different adjusted EBITDA. The gap between the two numbers is the zone where the deal price actually gets negotiated, and the "right" EBITDA is whichever number both sides end up agreeing to.
Another example is comparable selection. An enterprise software business growing at 15% with 85% recurring revenue. The sell-side banker shows a comp set of high-growth SaaS companies at 12-15x. The buyer's team adds three slower-growth incumbents with similar customer profiles, and the range drops to 9-12x. On sixteen million of adjusted EBITDA, the difference between 10x and 13x is forty-eight million euros. That gap comes from a judgment call about which companies count as "comparable," and in the context of a deal, that judgment call is a negotiation move.
Or take the management presentation. The CEO walks through the strategic plan, talks about market tailwinds, frames a 40% revenue concentration on two clients as "deep strategic partnerships." The buy-side team knows this framing is generous. The sell-side knows the buy-side knows. But the presentation builds a shared narrative that both parties can reference when they sit down to discuss price. The art is in controlling that narrative without contradicting anything in the data room.
What we can learn from Machiavelli
Machiavelli wrote about something similar in a different context. His point was that you need to see the difference between what a situation looks like and how it actually works. A leader who mistakes ceremony for substance gets outmaneuvered by one who reads the mechanics underneath. I keep thinking this applies to deal work more directly than people realize. The ceremony is the DD report, the model, the management deck. The substance is the negotiation happening through and underneath those documents.
The best deal professionals I've watched work get this instinctively. They read a QoE report the way a diplomat reads a position paper: useful information, though information structured to serve someone's interest. They don't present a management case as "the most likely scenario." They present it as the frame within which the price conversation will take place. Their financial model isn't a forecast. It's a negotiating instrument that happens to contain numbers.
At the same time, none of this diminishes the importance of good analysis. Weak analysis produces weak positions. If you can't defend your adjustments or explain why your comparables are more relevant than the other side's, you lose ground at the table. The skill isn't in doing the analysis. It's in doing analysis that holds up as a credible negotiating position while serving a strategic goal.
The fog lifting
For decades, the sheer labor of deal work obscured the negotiation underneath. Three associates and four weeks to build a model, cross-reference a data room, produce a 200-page report. All that production felt like the job. It wasn't. It was the infrastructure around the job.
AI is compressing the production layer. Data extraction, consistency checks, document cross-referencing, first-pass analysis. What used to take weeks now takes hours in some cases. And when you strip the production away, you're left with the judgment layer exposed: which adjustments to push, which comparables to include, how to frame the management case, what to flag and what to let slide. These were always the decisions that determined outcomes. They were just buried under enough work that you could mistake the digging for the treasure.
I don't think this makes the work easier. If anything, it makes it harder. When the analysis arrives pre-packaged, the person reviewing it has to make framing decisions without having built the intuition that comes from manually working through the data. That's a real problem and one I'm not sure anyone has solved well yet.
What the job actually is
If the production layer compresses, an honest question follows: what are deal professionals being paid for?
From what I've seen, they're paid for what they were always paid for, even when it was less visible. Judgment about which findings matter. Relationship capital that gets the other side to take your call. Knowing when to push an adjustment and when to let it go because the relationship is worth more than three hundred thousand euros. Pattern recognition from dozens of deals that tells you this CIM's growth narrative sounds like the last three businesses that fell apart during integration.
These aren't analytical skills. They're negotiation, relationship, and judgment skills. They compound over time, and they resist automation because they depend on context, history, and trust between specific people. A GP who has done twenty deals with the same lender closes terms in a phone call. A new entrant doing the same deal needs three months and two law firms. That gap is accumulated trust, and no software shortens the accumulation.
What changes, and what doesn't
I don't want to make this sound cleaner than it is.
AI-native services can produce better analytical output than many traditional DD providers. We are building one at Axion Lab, so I say this with full awareness of both the threat we represent and the limits we face. Cross-referencing findings across financial, commercial, and legal workstreams. Spotting contradictions between VDD narratives and data room evidence. Running scenario analyses in hours. This is real, and it puts genuine pressure on advisory firms whose value proposition is primarily analytical throughput.
But private markets run on relationships. An M&A advisor with twenty years of deal flow, a reputation for discretion, and personal connections to thirty GPs across Europe has moats that no AI system touches. If a buy-side team has to choose between "better analysis from a firm we've never worked with" and "good analysis from the partner who's been on our last eight deals," they'll pick the known partner. Not because they're irrational, but because transactions involve sharing sensitive information, managing conflicts, and trusting that your advisor won't create problems you can't see coming.
For newcomers, this is the hard part. The analytical edge is real but not sufficient. You still need relationship infrastructure, and that doesn't come with a subscription. For established players, the challenge is different: if clients can get most of the analytical work from AI at a fraction of the cost, the value proposition needs to shift from "we do good analysis" to "we produce good outcomes." That means investing in the negotiation and framing skills that were always the real differentiator but were never the pitch.
Where this goes
Maybe even in a year from now, the analytical layer of deal work will cost a fraction of what it costs today. The models will be faster, the cross-referencing will be automatic, the DD reports will arrive pre-built. Firms that compete primarily on analytical throughput will find that throughput is no longer a differentiator.
What won't be cheaper is the ability to sit across from a counterparty and find the number you can both live with. The GP who knows when to push and when to concede. The advisor who has enough trust with both sides to broker the conversation that gets a stalled deal unstuck. The deal professional who reads a DD report not as a conclusion but as an opening move, and knows how to play the next three moves from there.
The generation entering private markets right now will spend less time building models and more time in rooms where framing and relationships determine outcomes. That's not a loss. The analytical work was never the interesting part. The negotiation always was. We're just going to stop pretending otherwise.
