A study published in 2012 by the McKinsey Global Institute, in the report The Social Economy, measured where an office worker's week actually goes. The result: about 28% of the time is spent managing email, and nearly 20% looking for internal information or tracking down the colleague who knows the answer. More than a decade later, with the number of work tools multiplied many times over, the picture hasn't improved — it has gotten worse.
Research from Qatalog, done together with Cornell University, found that the average worker loses 59 minutes every day just searching for information scattered across different apps and systems. Nearly half of respondents (45%) said switching between so many tools hurts their own output, and 43% reported spending excessive time on that switching every single day.
For anyone deciding to invest in company-wide AI, that number matters more than it looks. An assistant connected to the company doesn't organize knowledge on its own — it inherits exactly the mess that already exists. If the current information is scattered across email, chat, shared folders, and whoever happens to remember it, the AI will search all of that at once, with no way to tell which version is the one that counts.
One question, ten places to search
Ask someone at a mid-size company to answer a simple question — what's the current expense policy, or which contract template to use with a new supplier — and watch where they go to find the answer. They open email and dig for an old attachment. They scroll a team chat back months looking for one message. They open the shared drive and find three files with similar names and different dates. If none of that works, they message a longtime colleague.
That routine isn't a discipline problem — it's the expected outcome of how work tools pile up over time. Asana's Anatomy of Work Index, covered by the trade outlet CIO Dive, found that the typical American professional moves through 13 different apps and switches between them 30 times a day. Each app holds a piece of what the company knows, and none of them was built to say which piece is current.
At growing Latin American companies, this tends to be even more informal: the policy that started as a text document migrates into a message pinned in the team's chat group, then becomes a spoken explanation repeated at every onboarding session. Each new channel doesn't replace the last one — it just adds one more place the answer might be hiding.
The cost doesn't show up on an expense report, but it shows up on the clock: the 59 minutes a day measured by Qatalog and Cornell add up to nearly five hours a week of pure searching, before counting the time it takes to refocus after every app switch. Multiplied across a whole team, that's work time that never turns into delivered work.
The shortcut that multiplies the mess
The most common reaction is to try fixing this with one more tool: a smarter internal search engine, a new "definitive" shared folder, or a request that everyone document things better. None of the three works, because none answers the question that actually matters: among the copies that exist, which one is current, and who can confirm that?
A faster search engine finds everything that's already scattered — including the two-year-old version, which shows up right next to the current one with no warning at all. A new folder just becomes one more place on the list, because the old folders stay active and nobody migrates what was already there on their own. And asking people to document more, without saying who owns each document and where it's supposed to live, produces more versions, not fewer.
The most expensive side effect is duplication. Research from M-Files, surveying more than 1,500 office workers and published in 2018, found that 83% of respondents had recreated a document the company already had, simply because they couldn't find it. Every recreated version is one more copy competing for the title of "the official one" — the exact problem the new tool was supposed to solve, now bigger.
Connecting an AI assistant to that environment without organizing it first has the same effect, only faster: the answer can come from any one of the copies, with no indication of which, and nobody notices until the wrong decision has already been made.
What has to be in place
Fixing dispersion isn't about finding the right software — it's about deliberately deciding where official knowledge lives and who is accountable for it. An environment like that has six verifiable mechanisms.
One place recognized as the official source. Every document has a declared owner, a current version marked as such, and a scope — the policy belongs to finance, the manual belongs to support — instead of existing "somewhere" with no one responsible for it.
Corporate identity, not a personal folder. Access to knowledge comes from the same company network login, not a personal account in whatever app someone picked. Anyone who leaves the company loses access at the same moment, without depending on someone remembering to revoke a random folder.
Access based on each person's role. Someone in sales sees the current sales material; someone in legal sees what belongs to legal. Nobody has to "discover" where to look, because the system already surfaces what's relevant to their role.
Human approval based on risk. Critical documents — a credit policy, a pricing table, a compliance standard — can require two-step review and approval before becoming an official reference, with the person who writes it separated from the person who approves it, and exceptions audited.
An audit trail. When a document became current, who approved it, and when it was consulted — all of that gets recorded. In a dispute over an answer, that's the difference between reconstructing the history in minutes and not being able to reconstruct it at all.
Reuse with permission, not loose copies. Once the right answer exists in a trustworthy place, it can be reused by other people and teams within the allowed scope, instead of becoming one more file copied into someone's personal folder.
This is how Skyller was designed: knowledge with an owner, a current version, and a scope by area, access based on each person's role, and two-step approval for anything critical.
Less time searching, more time deciding
The most visible gain is the time that stops turning into searching. If the 59 minutes a day measured by Qatalog and Cornell turn into real work time, even partially, that's nearly an hour per person, every day, given back to the company.
But the bigger gain shows up somewhere else: in the quality of the decision. When two people from different departments pull from the same current source, they arrive at the same answer — and stop wasting time arguing about which version is right before even getting to the actual problem. Someone new on the team finds the approved material without having to ask "where is this" to every longtime colleague.
That also changes what kind of work AI can actually do. A question about a policy or a process stops depending on whichever copy the assistant happened to find first, and instead comes from one recognized place, with the option to show where the answer came from. At Skyller, this is the foundation of how corporate knowledge gets organized before any answer is generated.
The compounding effect is what matters most to leadership: knowledge that today lives scattered across email, chat, and whatever people happen to remember becomes a company asset, instead of a lucky accumulation left over from whoever thought to save it.
Signs your knowledge is scattered
Before deciding what to do about it, it's worth checking whether this problem already exists in your company. Five signs tend to show up first:
- A simple question takes three channels to answer. If finding the vacation policy means opening email, searching chat, and still asking someone, the knowledge doesn't have a place — it has several candidates for one.
- Two people give different answers to the same question. That usually means each one consulted a different copy, and neither knew the other one existed.
- Nobody knows, off the top of their head, which version of an important document is current. If the answer is "I think it's the last one I got by email," the company is relying on luck, not a process.
- A document gets recreated from scratch because nobody could find the original. Every recreation is a new copy competing for the official spot — and one more chance for an error the next time someone consults it.
- The right answer lives only in one specific person's head. When that person is on vacation, on leave, or leaves the company, the question simply stops having an answer.
If two or more of these sound familiar, the problem isn't a missing tool — it's the absence of a recognized place, with an owner and a version, for what the company already knows.
