By Ted Souder, Co-Founder & CEO, Quoin

Every industry has an eighty percent. It is what the current way of working produces.

A company runs the way it was built to run. That is what an operating model is for, and most of the time it is a strength. But it means change never happens on its own. Somebody has to identify the opportunity, build the case, win the budget, sell it to teams who did not ask for it, put it into practice, measure it, and adjust. That is a program, and it competes for attention against everything already on the list. This is as true of an eight-person advisory firm as it is of a global bank. The scale changes; the pattern does not.

When that program does not happen, or when nobody spots the need for it in the first place, the pattern is predictable. New technology gets absorbed into the existing workflow instead of reshaping it. Capital gets allocated against last year's baseline. Teams optimize inside the process they already have. Everyone works with the tools and the knowledge already in the building, and the output of all that is about eighty percent.

That is not a criticism. It is how organizations behave, and it is the most understandable thing in the world.

McKinsey's global AI survey, published last November across nearly 2,000 respondents in 105 countries, found 88% using AI regularly in at least one business function. 39% could attribute any enterprise-level EBIT impact to it. About 6% cleared what McKinsey calls high performer, attributing 5% or more of EBIT to AI.

They ran it again this year and published in August. Adoption climbed again. EBIT attribution slipped to 37%. The high performers were unchanged at about 6%.

In finance, you hear it like this: "We can already get eighty percent of the way there with the tools we have. It's good enough."

It is almost never said defensively. It is said the way people state something obvious.

If anything, the new technology has made eighty percent more comfortable to sit in. Most of what has been added to the research process in the last few years made the old way faster without changing it. So the same answer arrives sooner, which feels like progress, and the structure underneath it never moves.

You rarely get criticized for producing the memo everybody else would have produced. Which raises the question: Eighty percent of what?

Almost always, it is eighty percent of the exact same memo they already knew how to write.

The trap of consensus

If you are getting eighty percent of the way there, you have reached the baseline. You have the same facts, the same summaries, and the same high-level overviews as every other firm looking at the same asset.

Eighty percent is not where greatness happens. In this industry, the edge lives in the final twenty percent. The insight that wins a deal or changes a strategy is rarely the part everyone else also found.

Not all of it is worth having. Some of that last twenty percent is noise, and knowing which is which is a judgment call. That judgment is what you pay experienced people for. Most firms never get to apply it, because capacity makes the call first.

Look at where the gains have actually gone. Mercer surveyed 131 asset managers early this year. 55% have AI running inside at least one strategy's investment process, so adoption is not the problem. 69% cite enhanced operational efficiency as a measurable benefit. 8% cite improved returns. That is an industry that bought the technology and kept the process.

That is the trap. A good enough tool gets you to consensus faster without changing what your team is able to attempt. They are still doing the deep work one name at a time, one document at a time, by hand, and that is exactly where the final twenty percent is buried. The research tools of the last few years have mostly made reading faster. They have not changed how much research a team can actually run.

I watched this happen once already

I spent twenty years at Google, most of it sitting with business leaders while they worked out whether the shift to search actually mattered to them.

In the early days, most of them wanted one thing: Show my ad when somebody types my name, or types the thing I sell. That was reasonable, and it worked. It was also the same thing they had always done, in a new place.

Then a smaller group realized they could reach people who might be interested but were not looking for them yet. That meant rebuilding how they thought about budget, about measurement, about who a customer even was. It was uncomfortable. Those were the people who ended up reaping the benefits, and they were not smarter than anybody else. They were more comfortable being uncomfortable.

Nothing forced that choice on them. No quarter arrived where the old way suddenly stopped working. That is why these gaps stay open for years. They close when somebody decides to be uncomfortable before they have to be.

Habits older than the tools

Look at how research gets done inside most firms today. Nearly every habit goes back to a physical limit on what one person could carry.

An analyst may cover thirty names. Real diligence still happens one name at a time, because coverage scales and depth does not.

You screen fifteen comparables in minutes and then do genuine work on three, because three is what fits into the week.

Your memo has the same sections it had in 2015, and for good reason, since the committee has to read across deals. It also means the work gets shaped to fit the document.

You commission the market map instead of building it, because building it in-house costs weeks you do not have.

And when the same name comes back around nine months later, you pull the model and the old memo and rebuild most of the rest, because the thinking behind last year's work is not somewhere you can query.

Together these habits are the structure of a business built to run at human speed. That worked fine when the world moved at human speed. It does not work now. News, geopolitics, technology, and the economy all land on the same desk in the same afternoon, feeding the same decision. Work that used to get a week now gets a day. This is what I mean by the Great Compression.

You cannot solve a structural problem with a tooling upgrade. To reach that final twenty percent, you have to change the shape of the work.

Run all fifteen names properly, each worked independently and then compared, with every claim carrying its source. The reviewer stops assembling the work and starts interrogating it. Outside data connects in rather than sitting in a system somebody has to remember to check. And the template stops being the constraint, so an analyst pulls what the moment calls for, an investment memo, an executive summary, a thematic breakdown, all off the same underlying work.

The point of automating the eighty percent is not saving time. It is freeing your people to work the twenty percent where the firm's value actually lives.

What do you own at the end of the year?

Add up what your firm spent on research last year. The subscriptions, the data, the people, the outside diligence providers, the hours nobody bills for.

Then ask what you own now.

For most firms, it is a folder of documents that were true on the day they were written. The models get maintained. The reasoning does not. Why you passed, what you believed at the time, what you got wrong and when you found out, none of that sits anywhere the firm can question later. It lives in the heads of the people who did the work, and it leaves when they do.

That is the thing worth changing, and it is not a software decision. It is a different idea about what research is.

Research should be a living body of knowledge the firm owns and keeps building. Every report you run feeds it. Every outside document feeds it too, your diligence files, your sponsor notes, your market maps, all in one place. Instead of starting from zero, your team is asking questions of everything the firm has already learned.

It keeps working when you step away, watching the themes you care about and telling you when the landscape shifts. It is worth more every month, because everything you have already done is still in there, still working, and still informing the next decision.

Most serious firms will end up with one place where their research and their analysis live and compound. Some will start building it now. The rest will keep producing documents, and will spend the back half of this decade trying to catch firms that never start from zero.

So stop asking whether your research is good enough. Ask what you own at the end of the year.

Eighty percent is not bad work. It is a number about how your firm is built, not about how good your people are. Change the way it is built and the number moves on its own.

We built Quoin for the firms that go first.

Sources: McKinsey, The State of AI: Global Survey 2025, published 5 November 2025 (1,993 respondents, 105 countries), and The State of AI: Global Survey 2026, published 25 August 2026 (1,719 respondents, 97 countries, fielded 4 May to 8 June 2026). Mercer, AI in Asset Management Survey, fielded February and March 2026 (131 managers).