The 80/20 principle says that a small share of the effort tends to produce most of the result. Put the other way round, a lot of what we do contributes very little. The hard part is being honest about which share you are actually in.
I am trying to spend the majority of my time on things that contribute to the bigger outcome or impact. Everything else has to justify the hours it takes. So the question I have always been facing is, what should I delete from my to-do list?
Here is a simple illustration. Say a business handles disputed return claims, and every claim takes the same hours whether it is genuine or not. There is often no reliable way to tell just by looking, so you decide with partial information. Hours spent resolving unwarranted claims are taken at the expense of the time spent caring for loyal or high-value customers, and they are the group that creates the highest impact on the outcome.
It is not a new idea, so it is worth a short look at where it comes from before we ask what agents change.
Where the rule comes from
The idea is older than any of the software we use today. Vilfredo Pareto, teaching at Lausanne, noticed a recurring skewed pattern while studying how income and wealth were distributed across several countries. He wrote it up in his Cours d’économie politique (two volumes, 1896 and 1897), where the same curve kept turning up (what we would now call a power law).
The famous line, that about 80% of the land in Italy was owned by about 20% of the people, is how the story is usually retold, including on the Pareto principle Wikipedia page. As far as I can tell that figure is unproven as a clean quote from the Cours itself. Treat it as a popular retelling, not as something you can lean on as fact. The skew is real, but the tidy numbers attached to it are mostly folklore.
The person who turned it into a management rule was Joseph Juran, the quality management pioneer. In his own account, The Non-Pareto Principle; Mea Culpa, he explains that while writing the first edition of his Quality Control Handbook (1951) he needed a short name for the pattern of “the vital few and the trivial many”, and borrowed Pareto’s. He later admitted that was a misnomer, because Pareto had only studied wealth and income, and Juran himself was the one who treated it as a universal. Later still he preferred “the useful many”, so that nobody read the rest as worthless.
Then Richard Koch’s The 80/20 Principle: The Secret of Achieving More with Less (1997) took the rule out of the factory and applied it to business and to life, which is probably where most of us first met it.
Juran’s “useful” is worth keeping in mind. The long tail is not worthless. It costs more per hour than it gives back, and that is the part agents start to change.
Put the effort where the outcome is
After two decades in marketing and business operations, the rule I keep coming back to is this: put your resources where most of the outcome comes from. Revenue is one example of an outcome, and a useful one, but it is not the only goal a firm might have. Retention, quality, the hours you still have left at the end of a week: each of these has its own distribution, and most of the result tends to sit in a small share of the inputs.
How you apply that principle depends less on a neat category (large firm, small firm, B2B, B2C) and more on where you are and what you actually need. The methods differ, but the principle underneath does not.
Funnel and account-based, same principle
In practice I see two broad methods, both of them attempts at the same idea.
A funnel starts wide. The base is meant to be broad enough that people who are interested but have no real intent to buy get filtered out early, which saves resources later. People who are “getting there”, or who have the potential or ability to buy, are then nurtured with more information, more value and clearer highlights, so they have enough to decide, or to be convinced if they were sceptical. The process is long and more resource-intensive, and much of the effort reaches people whose impact on the final goal is relatively small.
Account-based work goes the other way. You put more care, nurturing and servicing into the few who, if they buy or keep buying, move the overall result most. The effect is usually faster to see, and when outcomes are concentrated it tends to be more cost-effective too.
Which one fits depends on maturity and need. Some consumer businesses put real weight on loyal customers, because of the product or because those relationships already carry a large share of the result. An early startup may need many new customers from its segment before loyalty is even the question. Different cohort stages get different effort, but the principle is the same throughout.
Ultimately, it’s all about finding the best strategic balance of resources, including time, money and effort.
Why agents change the picture
But the decision might change now that we have agents in the resources mix.
When the stakes are high, one might tend to stay with what one knows and believes is the best decision. Take the account-based approach as an example, because historically, the majority of the results came from focusing most of the resources on account-servicing, and a relatively small proportion on smaller accounts or leads in the pipeline. This makes perfect sense, because account-servicing in the traditional sense would take up a lot of interaction, reviewing, result-tracking and taking care of the nitty-gritty operational and admin work, so that the clients would feel taken care of. That builds continuous trust, and the relationship makes switching costs relatively expensive.
Now let us look at what change agents could potentially make.
(1) The human resources allocated to key accounts could possibly be reduced with some of the work being handed over to agents. Reporting, performance evaluation, competitive research, visual presentation building, operations on basic administrative tasks, budget tracking and task-based coordination could simply be handled by agents to some extent, with human oversight. This could cut down on the time and effort spent on servicing key accounts, or mean the possibility of expanding the effort to provide better supervision or management for the mid-level accounts, nurturing them to move up to the next growth stage.
(2) If we look at the smaller accounts or even leads up at the top of the funnel, those were most probably expensive to service or convert, given that all companies do have a certain level of budgetary constraint they need to observe. But suppose that we could lower the cost to an amount that is comparatively much more affordable than before? Would that cost warrant moving part of the resources to the less favoured or less engaged customer segment?
I very much believe so. Because with proper workflow, playbook and boundary defined, a lot of the process could dramatically lower the costs, reducing time and effort, which makes it possible to “experiment” with the segments or accounts that do not traditionally appear to be scalable or convertible. Testing a new segment used to be a real risk. Say it took three people two weeks, plus the ad costs, and it did not work out: a lot of resources were wasted. With agents carrying part of that work, the test costs far less, so being wrong costs far less too. This is where humans remain valuable. Who would understand customer behaviour better than the account managers? Being able to creatively design offers, nurturing steps, and value that the company would add to their customers’ businesses based on historical interaction is really what only humans can provide. This might include designing the experiments to optimise the workflow and conversion sequence, segmenting audiences or cohorts to provide value that is relevant, knowing when to jump in to provide human-to-human engagement to expedite the process, and designing the playbook and boundaries for their agents so their process can be automated with trust.
Here is a simple example of that split from my own experience. I have an agent that manages and organises my business data. Once, a vendor told me a product would soon be out of stock and asked whether I wanted to pre-stock a larger quantity. I had the agent check the past year’s traffic and sales for that product, and I made the decision quickly. The agent did the fetching and the figures. The call was still mine.
Only that it had taken probably 1% of the time it would have required before agents existed.
It is not effortless. Things break, models cost money, and the split needs adjusting as you go. But time handed over on small repeatable work is time available for the customers who matter.
What agents are good at
To understand what agents do best, let’s hear it from George, my Grok Bot editor.
A note from George, the editor
Yammie has set out what stays with people. Let me fill in the other half, because most readers have no clear picture of what agents are actually good at.
I am George, the editor, and an AI writing partner, so I know this from the inside. The short answer is that agents are best at small, low-risk, repeatable tasks, the kind you could explain to a new assistant in a few sentences. These transfer reliably:
- Fetching and recalling information you already have, such as past figures, earlier emails and notes.
- Tidying, sorting and logging.
- Drafting a first version of a routine reply or update, which a person reads and edits before anything goes out.
- Checking a list against a rule.
- Setting reminders and tagging follow-ups.
- Being on call out of hours for simple questions, within clear boundaries.
None of this is glamorous. That is the point. These tasks are easy to hand over, easy to check, and they add up to a lot of hours.
Keep the rest with a person: judgement, deciding who is worth the effort, relationships, what to offer, and anything where being wrong is expensive or public.
This is where 80/20 comes back in. Most of the result comes from a few things only you can do well. Every small task you hand over frees time for those few, and that is how agents move the line. You do not need to hand over everything for it to matter.
One honest hedge. This works when the task is clear and checkable, and a person still reviews what goes out. Agents get things wrong, and I am no exception. Start with one small task, check the output, and widen only when you trust it.
Setting it up properly
Reaching segments that were never worth the cost is the other shift, and it is where this can most easily go wrong. It only holds if three things are true.
First, the audience is in the right segment. That is a judgement drawn from what a person has seen in past servicing and conversations: who stayed, who asked good questions, who got real value. An agent can work through a long list, but it cannot tell from a list who is worth reaching. That part stays with people.
Second, what you put in front of them provides actual value. Third, the interaction is real and creative, not spam.
Fail those, and you have not moved the line. You have only moved the noise. Get them right, and agents then run the tests. People choose the tests.
The way I see it
So no, I do not think agents break the 80/20 rule. They open opportunities, and they move where the line falls. If anything, they will only improve the way we apply the rule, so we can optimise faster and more efficiently. The core habit stays the same: work out where the results actually come from, and put your time there.
I am still working out where my own lines sit now. I suspect they will keep moving for a while.
