- What does it mean to sort twice?
- Why does the second cut feel wrong?
- The worked example
- Where the second cut breaks
- From the finding to the action
- How to run it
- What to do with the answer
- What not to do with it
Compounding Pareto is a phrase people use without ever having run the arithmetic, so it is worth defining precisely before anything else. It means applying the same sort a second time, to the population that survived the first one. If twenty percent of inputs drive eighty percent of output, and the same relationship holds inside that surviving twenty percent, then four percent of inputs drive sixty-four percent of output.
Four percent and sixty-four percent. Those two numbers are the entire subject, and almost nobody has ever looked at their own version of them. This assumes you have built the customer and product profitability picture, because the second cut runs on the same data as the first.
What does it mean to sort twice?
It means treating the winners of the first sort as a full population and ranking them against each other, rather than as a block that has already been dealt with. Nearly every organization stops after one pass, and the reason is structural: the first cut produces a list of things to fix and a list of things to protect, and the protected list feels finished.
It is not finished. The protected list contains the same distribution as the original population, because the underlying mechanism that produced the concentration did not stop operating at the threshold you drew. Your best segment has a best quartile inside it, and that quartile is carrying the segment the way the segment carries the company.
Why does the second cut feel wrong?
Because you already told these people they were the winners. The first cut is emotionally easy: the tail is obviously the tail, everybody in the room suspected it, and the conversation is about accounts and products nobody personally champions.
The second cut asks the same room to rank things it has just finished blessing. The product families that survived. The accounts you decided to build the business around. The salespeople who own them. Now somebody has to say that inside the winners, a small group is carrying most of it, and the rest are riding on a reputation they did not individually earn.
That is an uncomfortable meeting and it is why the cut is unclaimed. It is also why it is worth doing, because an advantage that requires a difficult conversation is one your competitors are not having either.
The first cut is easy because the tail is obviously the tail. The second cut asks you to rank the winners you just finished blessing. That is why nobody runs it.
The worked example
Take a business with 500 customers where the first sort establishes that 100 of them produce eighty percent of gross profit. That is a normal, unremarkable finding and most businesses stop here, reorganize coverage around the hundred, and consider the exercise complete.
Now run the same sort inside those hundred. If the concentration holds, roughly twenty of them produce sixty-four percent of total gross profit. Twenty relationships, out of five hundred, carrying nearly two thirds of the money. The method I use for this is 80/20 Squared, and the practical value is not the arithmetic, it is that twenty is a number a management team can actually act on.
What changes when you know the twenty. You can name them in a meeting. You can assign an executive to each. You can build the product roadmap around what they need rather than around what the aggregate asks for. You can model, precisely, what happens if one leaves, which is a risk conversation most boards want and few businesses can support with data.
A hundred accounts is a segment. Twenty accounts is a list of people, and the difference between a segment and a list of people is the difference between a strategy nobody executes and a set of assignments somebody owns by Friday.
The same arithmetic runs on product families and it is frequently more actionable there, because products have no feelings and no account manager defending them. Take the families that cleared the first cut, rank them by gross profit after allocated complexity cost, and the top group usually turns out to be a handful of items carrying an entire catalogue. That finding points directly at where engineering hours and capital should go next year, which is a decision most businesses make on the basis of who asked loudest. For a company-scale version of that cut, see how LEGO cut its complexity and saved the company.
Where the second cut breaks
Three places, and I would check all three before acting on the output, because the second cut is considerably more fragile than the first and the failures are not obvious.
It breaks when the concentration does not actually hold. The eighty twenty relationship is an observation about a distribution, not a law, and it does not automatically repeat at every level. In some businesses the top hundred are genuinely flat, with the hundredth account producing nearly as much as the tenth. Run the arithmetic before assuming the shape. If the surviving population is flat, the second cut has told you something important and the answer is that there is no four percent.
It breaks when the population is too small. A business with forty customers does not have a meaningful second cut, because the surviving eight are a list you already knew and the sample is too thin for the distribution to mean anything. Below roughly a hundred accounts or a hundred product families, this is arithmetic theatre.
And it breaks when acting on it would destroy the base that generates the winners. This is the one that actually hurts. If your twenty best accounts all came from a segment you only reach through a broad channel, concentrating resource on the twenty and withdrawing from the channel removes the mechanism that produced them. The second cut tells you where the profit is now. It says nothing about where the next twenty come from, and businesses that have optimized hard on this have occasionally optimized themselves out of their own pipeline.
The second cut tells you where the profit is now. It says nothing about where the next twenty come from, and that distinction has ended more than one good strategy.
From the finding to the action
Everything above is about whether the second cut applies to your business and what it means if it does. What follows assumes the concentration held, the population is large enough, and you have checked that the winners are not being generated by something you are about to stop funding.
A fourth failure mode, less common but worth checking: the second cut can be an artifact of a single period. An account that appears in your top group because of one unusually large project year is not structurally concentrated, it is temporarily large. Run the second cut on three years rather than one and compare the lists. Names appearing in all three are the real group. Names appearing once are a project, and building a strategy on them is how businesses end up over-invested in a customer who has finished buying.
How to run it
Five steps, and they run on data you already produced for the first cut. This is an afternoon of work, not a project, which is another reason the neglect is hard to justify.
- Isolate the survivors as a standalone population. Export the accounts or product families that cleared the first threshold into their own file, with their gross profit after allocated cost to serve. Do not carry the rest forward.
- Sort and cumulate again, from zero. Rank by gross profit descending and build a running total as a percentage of the company total, not as a percentage of the survivors. That distinction matters: you want to know what these accounts are worth to the business, not to each other.
- Check the shape before you read the list. Look at whether the curve is steep or flat. A steep curve means the concentration held and the second cut is real. A flat one means it did not, and that is a finding rather than a failure.
- Name the top group out loud. Write down the actual accounts. Not a segment label, not a tier, the names. The exercise fails the moment somebody converts the list back into a category.
- Test each one for durability. For every name, answer why they are there and whether it persists: a technical position, a qualification barrier, a contract, a relationship with one person. A concentration built on four individual relationships is a different risk from one built on four designed-in components.
Before any of that, run the shape check on three years rather than one, as described above. The five steps take an afternoon. Running them on a single year and acting on the result is the most common way this exercise produces a confident wrong answer, and the extra two years of data cost you nothing beyond a longer export.
What to do with the answer
Assign, protect and model, in that order. Assign a named senior person to each of the top group, with a standing obligation rather than an account plan. Protect the specific things that make each one profitable, which are usually process characteristics rather than commercial terms.
Then model the loss of each one individually. Not as a risk register entry, as an arithmetic exercise: what happens to gross profit, to plant loading, to the overhead absorption on everything else. In most businesses the second cut is the first time anyone has quantified that, and the number is frequently large enough to change how the company thinks about capacity, about credit, and about how many of these relationships it can afford to have.
The other use, and the one people miss, is prospecting. Take the five durability characteristics that the top group shares and use them as the screen for new business. Most commercial organizations prospect on size. Screening on the profile of your own proven best accounts is a materially better filter and it costs nothing to adopt.
There is a second-order use worth mentioning because it costs nothing. Once you know the twenty, look at what your organization currently spends on them relative to everything else: coverage hours, engineering time, executive attention, entertainment budget. In most businesses the allocation is closer to flat than anyone expects, because coverage was set by territory or by revenue rather than by profit. Correcting that allocation is free, requires no customer conversation, and is usually the single largest practical consequence of having run the second cut at all.
What not to do with it
Do not cut the rest. The population between your second cut and your first cut is profitable, it absorbs overhead, and a meaningful share of it is where your next top-group account comes from. The second cut is a resource allocation instrument, not a rationalization instrument, and confusing the two is the most expensive error available here.
Do not publish the list broadly. A named top group circulating in an organization changes behaviour in ways you did not intend: coverage quietly drains from everything else, and people outside the list start managing their careers rather than their accounts.
Do not run it annually and expect it to move. It will not, much, and that stability is the point. If your top group churns substantially year over year, the concentration was never structural and you were reading noise. The related question of how a structural sort holds over decades is covered in the mechanics of how Illinois Tool Works ran this, and the answer there is the same as it is here: the sort survives because the structure around it changed, not because the analysis was good.
The second cut is a resource allocation instrument, not a rationalization instrument. Cutting the population between your two thresholds removes where the next winners come from.
Monday morning: take the file you built for the first sort, delete every row below the threshold, and re-cumulate from zero.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B+ shareholder value across five turnarounds at Berkshire Hathaway, Illinois Tool Works, and Whirlpool. He is VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.