How much time consultants lose to admin work depends entirely on whose number you are willing to trust.

Search how much time consultants spend on admin and the first page reads like consensus: six to eight hours a week, fifteen to twenty-five per cent of billable time, numbers repeated confidently across a dozen near-identical articles. Click through looking for the underlying study and there usually isn't one. The figure is asserted, then repeated by the next site, then treated as established fact by the third, until a specific-sounding number has circulated widely enough to feel true regardless of whether anyone ever measured it. One methodologically described survey does exist in this space, published rather than merely asserted, and its numbers are less dramatic and more useful than the ones quoted everywhere else. Everything around it deserves a second look.

"Admin" means different things, which changes the number

Part of why this category produces so many incompatible figures is that nobody agrees on what's being measured. Does admin include invoicing and chasing late payment. Does it include the meeting-note write-up after a call, or only the scheduling before it. Does proposal-writing count as admin or as unpaid sales work, a completely different bucket with a completely different implied fix. Two surveys asking "how much time do you spend on admin" without defining the term are not measuring the same thing, even if both report a number, and stacking their results together to form a consensus range is statistically meaningless regardless of how confidently the range gets repeated.

This matters beyond pedantry, because the definition a survey uses quietly decides the answer before a single respondent is asked anything. A narrow definition, invoicing and scheduling only, produces a small number. A broad one, anything that isn't billable delivery work, produces a large one, and neither is wrong. They're answering different questions wearing the same headline.

Why round numbers travel so well

The impulse behind the unsourced figures is not malicious. Admin overhead is a real, widely felt problem, and a specific number is more persuasive and more shareable than an honest "it varies enormously and nobody has measured it properly." A writer facing an under-researched topic on a deadline reaches for a plausible figure, because the alternative is publishing an article that admits it doesn't know. That is editorial pressure rather than conspiracy, and it explains why the same handful of numbers keep reappearing with new bylines attached.

Rigorous research does sit adjacent to this question without being aimed at it. Gloria Mark's work on workplace interruptions, conducted over years of logged office behaviour, is credible and widely cited; it was not measuring solo consultants. Asana's annual Anatomy of Work report surveys a large cross-section of knowledge workers on this kind of overhead question with real methodology behind it, aimed at an employed workforce rather than the independent-practice population the search is about. That work deserves crediting. Serious research exists near this topic. None of it was pointed at the person typing the phrase.

Where the specific figures stop holding up

Look closely at the numbers actually circulating and the same three problems recur across almost every article repeating them.

A round number with no methodology is a guess wearing a citation.

"Six to eight hours a week" and "fifteen to twenty-five per cent of billable time" are suspiciously tidy ranges for a figure that would, if real, require tracking actual consultants' actual time in actual detail. Neither range is accompanied by a sample size, a data-collection method, or a link to underlying data, anywhere it was found repeated across this category.

General knowledge-worker research isn't the same population as a solo consultant.

An employee's admin overhead, expense reports routed through a finance department, status updates into a shared tracker someone else built, is a structurally different problem from a solo consultant's, who is simultaneously the person doing the delivery work and the entire back office. Borrowing a statistic from employed-workforce research and relabelling it for consultants is a category error dressed up as a citation.

Nobody has surveyed the population this question is actually about.

Independent consultants, specifically, running their own client relationships without a team, are a narrow and genuinely under-studied population. Market researchers have real commercial reasons to study large employers and enterprise software buyers; a fragmented population of solo practitioners, individually small and collectively hard to sample, is expensive to reach and doesn't obviously pay for the research to be commissioned. The absence of good data is the honest answer here, and it should be stated as one rather than papered over with a plausible-sounding number.

A statistic repeated by ten websites with no shared source is not ten confirmations. It is one guess, copied nine times.

The bar real research on this question would have to clear

A credible answer would need a clear definition of admin, a large enough sample of independent consultants specifically, and either direct time-tracking data or self-report collected carefully enough to account for the well-known tendency to underestimate low-attention, fragmented work. That's a genuinely expensive study to run properly, which is precisely why almost nobody has run one, and why the gap gets filled with assertion instead of an honest "this hasn't been measured well yet." Naming that bar is useful even without clearing it, because it makes clear why one survey with a described method is worth more than a dozen without one, regardless of how the two compare in headline size.

What the one real survey actually found

Smallpdf's 2026 freelancer survey is the exception worth taking seriously, because it discloses its method: 397 self-employed freelancers surveyed directly about how they spend their working time, with the resulting figures published rather than merely asserted. Its headline finding is that respondents lose an average of 204 hours a year to administrative work, close to a full working month, once invoicing, scheduling, client communication and general business upkeep are added together. That figure is more conservative than the "six to eight hours a week" claims circulating elsewhere, and more trustworthy for being conservative, because it comes with a described sample rather than a confident assertion.

The same survey found that seventy per cent of respondents now use AI tools for some part of their admin work. The more interesting number sits beside it: forty-eight per cent said the burden hasn't meaningfully reduced despite that adoption, and almost nobody quoting the seventy per cent bothers to mention it. A statistic that flatters the AI-adoption narrative travels further than one complicating it.

A single survey, however well described, can only do so much. Three hundred and ninety-seven respondents is a useful sample rather than a population-level census, and the true figure for any individual practice could sit well above or below 204 hours depending on client mix and business model. Treating one good survey as gospel repeats the mistake of treating an uncited blog post as gospel, with better manners about it.

Why the burden isn't falling even as AI adoption rises

That forty-eight per cent deserves more attention than it gets, because it cuts against the easy assumption that AI adoption alone solves an overhead problem. A tool that drafts an email faster, or summarises a document, removes friction from one task in isolation. It doesn't remove the reconciliation between tasks, the re-entry of the same fact into three different places, or the fact that nothing kept track of the obligation in the first place. Speeding up a single step in a fragmented process leaves the fragmentation intact; the same broken handoffs happen faster. That is a plausible explanation for why adoption and relief decoupled the way the survey found.

Where the time consultants spend on admin actually goes

Without a rigorous, consultant-specific time-and-motion study, the honest move is to describe the shape of the loss rather than assign it a number that doesn't exist yet. Reconciling what was discussed on a call against what a client's own tracker says. Assembling a status report from facts scattered across three tools because nothing generates it automatically. Chasing down which of several loosely related notes actually contains the decision a client is now asking about. None of it is dramatic on any single day. Together it is the describable texture of admin overhead that a category desperate for a number has mostly skipped past.

The time consultants spend on admin is easy to miss in the moment

Researchers aren't the only ones without a good number. Each individual instance is small enough to feel unworthy of tracking. Two minutes checking whether a decision was already logged somewhere. Ninety seconds re-typing a note into a second tool. None of it registers as a task, so none of it gets counted the way a billable hour does, even though the same person would notice at once if a client asked them to spend that time on something explicit. The invisibility follows from the work never being attached to anything that would have made it visible.

What Callisto can and cannot claim here

Callisto has not run its own survey of consultants' admin hours, and implying otherwise would be the fabricated authority this piece argues against. The honest claim is structural rather than statistical. The connected-graph model removes nameable re-entry points: a completed priority updates the project, the person and the request it came from in one action, and a meeting-note segment reaches the project, the account and the person it concerns in a single save, rather than the same fact being typed in three times. Whether that adds up to two hours a week or eight for any given consultant is not a number Callisto can claim to know.

Getting started

For one week, without changing any workflow or trying to fix anything yet, keep a plain running note of every moment spent re-typing the same fact into a second place: a decision from a call copied into a tracker, a status re-assembled from memory for a report. At the end of the week the number will be specific and personal, and more useful than anything six other articles copied from each other, because it will be the one figure in this category that is true.

That number is also the one worth pointing at whichever tool in the current stack is responsible for the most re-entry, since reducing the count that matters starts with knowing which one is actually costing the time.