The work never got less: Jevons' paradox comes to deal-making

AI will not automate away the work of selling a company. It will raise the bar for diligence on every deal.

As the cost of analysing data falls, buyers will scrutinise more information, faster. Advisors will do deeper analysis on every mandate, and what counts as a "ready" asset will change. The advantage will move from producing a polished Information Memorandum to building an investment case that holds up when the buyer's AI checks it, and that protects the valuation.

It sounds contradictory that automation leads to more work. Economics has a name for it: Jevons' paradox.

Recently, I sat down with a group of bankers with decades of sell-side experience. I brought up Jevons' paradox, and the conversation turned to whether 25 years of Datasite, Mergermarket and DealCloud had actually made selling a company less work.

They gave the same answer, almost word for word: "No. The work never got less. The diligence got better."

I expect AI in M&A to follow the same pattern.

A paradox from 1865

In 1865, the English economist William Stanley Jevons published The Coal Question. Britain ran on coal, and many assumed that more efficient steam engines would reduce coal consumption.

Jevons argued the opposite. Better engines made mechanical power cheaper, so it became worth using them in more mines, mills and railways. Each engine used less coal, but there were so many more engines that total coal use went up.

Efficiency does not just reduce the cost of doing the same work. It makes new uses worthwhile. When something becomes cheaper, demand often rises enough to offset the savings. That is the question AI now raises for boutique advisory firms.

When efficiency raises the bar

We have seen this before. Spreadsheets made financial modelling much easier. They did not eliminate the modeller. They made it practical for buyers to ask for more scenarios and stress tests.

We see it in other knowledge work too. One early study of accountants found that AI adoption was associated with supporting 55% more clients and closing the month 7.5 days earlier. Meanwhile, the cost of running AI models at a given quality level has been falling by roughly ten times a year.

Deal-making has been through several waves of technology. Paper data rooms went virtual. Financial databases put market data in the hands of every analyst. Yet the work grew. Research covering more than 900 deals found that due diligence before announcement took over 200 days on average, with data rooms holding thousands of files.

As analysis became cheaper, specialist workstreams such as commercial, tax, IT and cybersecurity were added. Buyers asked more questions because they could. The standard for thoroughness went up. I see AI speeding up this cycle in three ways for boutique partners.

1. The diligence arms race: AI vs. AI

Imagine producing an IM, a financial databook and several operating scenarios in a fraction of the time. At first, you save hours. But the buyer has the same technology.

Buyers will use AI to work through your data room, read your contracts and check your adjusted EBITDA against the raw ledgers for anomalies. It is no longer only associates reading your documents. It is the buyer's AI checking yours.

Clients will expect deeper analysis upfront for the same fee. The result is more work per deal to make sure your story holds up.

2. The end of sampling

Diligence has always relied on sampling. A buyer's legal team might review the top 20 customer contracts. A financial team might check a few months of transaction-level data to verify gross margins. Checking everything simply cost too much.

AI removes the reason to sample. When software can read, compare and flag 500 contracts or hundreds of thousands of ledger entries in minutes, buyers will check all of the target's data.

For the sell-side advisor, the bar for a "clean" company just went up. You can no longer rely on averages or samples to cover operational issues. You need to know what is in the 395th contract before the buyer's AI flags it as a dealbreaker.

3. Verification becomes the bottleneck

AI can write a plausible market overview or a coherent financial story in seconds. But in M&A, plausible is not enough. Numbers have to reconcile, and material claims need evidence behind them. Deal certainty depends on facts, not fluency.

Before buyers commit time and money to an LOI, they need to know that the data behind the IM will hold up in legal and financial diligence. Generating text has become cheap. Verifying AI output has not. Resolving conflicting evidence, deciding what is material and putting a partner's reputation behind a conclusion still need human judgment.

When polish is no longer a signal

George Akerlof showed how markets break down when buyers cannot tell good assets from bad ones. Michael Spence showed how credible signals help them tell the difference.

For a long time, a carefully crafted 80-page IM signalled an advisor's rigour and competence. AI makes a polished document much cheaper and faster to produce, so it says less about quality.

The answer is not a 150-page IM. The answer is verifiability.

Imagine an IM where every metric links to its source, every adjustment is clearly explained, and every market claim is backed by credible evidence. The buyer's AI can confirm your claims quickly instead of raising red flags that stall the process. The advisor who helps the buyer reach a well-founded conclusion faster adds real value, even if the document cost less to draft.

The partner's imperative

Saving associate hours on drafting improves a boutique's economics. Each mandate costs less to run, your team can take on more mandates, and margins improve. But your firm earns its success fee at closing, and that is where the bigger risk sits. Buyers will use AI to find anomalies, drag out diligence and chip away at the valuation.

"How many hours can AI save me?" is only half the question.

The other half is how you protect your win rate and success fee when buyers expect solid evidence across all of a company's data.

The best advisors will use AI to find weaknesses before the buyer's AI does, and build an investment story that holds up under scrutiny. Better technology never removed the work in deal-making. It changed what good work looks like.

At Mindro, we help advisors build verified, well-evidenced IMs. Not to generate more pages, but to meet the new standard of deal certainty: material buyers can trust quickly, grounded in evidence you can stand behind. When intelligence is cheap, the value goes to the advisor who can show they are right.

(Fun fact: Jev, the decision engine we wrote about in our previous blog post, is named after this paradox.)