The work was estimated at ten person-days. It finished at sixteen. And on the next project, it gets estimated at ten again. Where workload management is not working, this loop repeats. The cause is not a lack of record-keeping. It is the absence of any mechanism that feeds those records back into the next estimate. This article covers how to decide what you are measuring for, the five reasons estimates and actuals diverge, and the practical steps for improving accuracy over time.
What Workload Management Is: Three Actions Joined Into One Loop
Workload management means estimating the time a piece of work will take, recording the time it actually took, and correcting the next estimate from the difference. It is those three things joined into a single loop, not simply the tallying of hours.
Estimate: before starting, put a number on the time required, in person-days or person-hours.
Record: capture the time actually spent, at a level of detail you can still make sense of later.
Reconcile: identify why the gap appeared and reflect it in the next estimate.
The third one is what most teams are missing. The logging happens, but it ends at a month-end tally that somebody glances at. That is time tallying, not workload management, and it means next quarter will be estimated with exactly the same accuracy as this one. The reverse is also true: as long as the third step runs, accuracy improves even when the records themselves are fairly rough.
For the definitions of units like person-month and person-day, and for judging whether a quote from an outside vendor is reasonable, see how to read an estimate and the pitfalls of person-month pricing. This article stays narrowly on how to handle your own team's effort.
Decide the Purpose First, Because Purpose Sets the Granularity
Teams where workload management does not stick usually started the discussion at the level of detail. That is backwards. Once you decide what you are measuring for, the necessary granularity follows automatically. There are broadly three purposes.
To improve estimate accuracy: you need actuals broken out by phase, recorded per project at the level of design, production, review, and so on.
To understand project cost: a per-project total is enough. There is no need to split by phase.
To level out workload: what you need is not past actuals but future commitments. Seeing how full each person's next month already is will do.
What Is Scheduling? How to Build a Plan That Does Not Fall Apart
If the third is your only purpose, collecting detailed historical actuals is almost entirely unnecessary. In practice, though, teams frequently start with "let's capture it in detail for now" without settling the purpose, the data-entry burden climbs, and the whole thing quietly dies within three months. Pick one to begin with.
Five Reasons Estimates and Actuals Diverge
The cause looks different every time, but in practice it lands in one of five buckets. Identify which one applies to you and the countermeasure is already half decided.
1. Work is missing from the definition
By far the most common cause. The estimate covers only the main task, while review, revision handling, meetings, preparing materials, environment setup, and explaining things to stakeholders are all absent. Added together, this surrounding work often comes to around thirty percent of the main task.
The fix is to close the gaps while you are still identifying the work. When building a WBS, always add one line for having the deliverable checked and one line for fixing it, alongside the line for producing it. That alone narrows the gap substantially.
2. You are assuming full utilization
Are you planning five person-days of work to finish in five business days? Almost nobody in practice can put all eight hours of a day into a single project. Subtract time split across other projects, recurring meetings, unplanned questions, and email, and what is actually available for one task is five to six hours a day.
In other words, realistic utilization sits around seventy to eighty percent. Assume one hundred and you produce the hardest kind of delay to diagnose: the effort estimate is correct, yet the schedule slips anyway. Treat effort and duration as different quantities, and always convert between them through a utilization rate.
3. You have not budgeted for rework
An estimate that assumes everything passes on the first attempt will almost certainly be wrong. A change in creative direction, a misread specification, a rejection at review: these are not accidents, they are work that normally occurs.
There are two ways to budget for it: add a uniform margin to each task, or place a single revision block at the end of the phase. The second is better. With the first, buffer hides inside every task and the work expands to fill it. Kept visible as a block, it can be pulled forward when it goes unused.
4. Waiting time has no agreed treatment
Waiting on a client reply, on sign-off, on assets being supplied. If nobody has decided whether that period counts as effort, the records will vary. Once some people log it and others do not, the totals can no longer be compared.
The rule of thumb is: waiting is excluded from effort and included in duration. Effort is time during which a person was working, and while you are waiting nobody is. The calendar does stretch, though, so the schedule reserves the period. Separating effort from duration pays off here as well.
5. The records are rounded
When you fill in the whole week on Friday, do you adjust the numbers so each day lands on eight hours? Some days ran long and some ended early, yet the record shows a uniform eight across the board. In that state you will never learn which work was heavy.
The fix is to shorten the interval between doing and logging. Writing it up at the end of the week makes memory vague and invites tidying. Move to logging thirty seconds at the end of each day, or at the moment you switch tasks, and the rounding falls away on its own.
Practical Steps for Closing the Gap
Step 1: Standardize the granularity of estimates
Break work down so each task lands between half a person-day and three person-days. Larger than that and the estimate becomes guesswork; smaller and the overhead of tracking exceeds the benefit. A task you want to answer with "maybe five days or so" is a signal that the breakdown is not finished.
Step 2: Convert to calendar days through a utilization rate
Do not turn total effort straight into business days. Twenty person-days of work handled by one person at seventy-five percent utilization takes roughly twenty-seven business days, about five and a half weeks. Whether you insert this conversion decides how realistic the deadline is.
Step 3: Keep logging to three fields
Date, project (or phase), hours. Three is enough. A free-text field for describing the work drops the completion rate, so leave it out at the start. Make the project a dropdown to eliminate typing. Once logging takes more than thirty seconds, it will not survive.
Step 4: Review by variance ratio
Look at the gap as a ratio rather than an absolute number of hours. Actual divided by estimate is your variance ratio. Reviewing only the items above 1.2 keeps the number of things to examine manageable. Try to review everything and the review itself stops happening.
What you are looking for is whether the high-variance items cluster around a particular type of work. Once you can say "review is running at double every time," the fix is simply to double the review estimate from now on.
Step 5: Turn findings into coefficients and feed them back
Whether workload management ends at tallying or turns into improvement is decided by this step. Write the patterns you found into explicit multipliers for the next estimate: "the design phase gets estimated at 1.3x," "work handled by a new team member gets 1.5x."
This is the work of converting rules of thumb held in one person's head into numbers the whole team can use. Once it exists, estimate accuracy no longer drops when the person doing the estimating changes.
Three Things That Help When Logging Does Not Stick
Show the use before asking for the data: people asked to log their hours suspect surveillance first. State up front that the numbers will not be used in performance reviews and that they will be used for future estimates and staffing. Leave it ambiguous and the numbers skew, both low and high.
Send something back: a system that only takes input and returns nothing always decays. Sharing "this phase ran heavy, so we are raising the estimate for it" once a month is enough to make the logging feel worthwhile.
Fix the cutoff: set a day and a time, such as 5pm Friday. "Whenever you get to it" is close to a synonym for "never."
Workload Management and Budget-Versus-Actual Are Not the Same Thing
These get confused, so it is worth separating them. Workload management deals in time; budget-versus-actual management deals in money. They are different, but internal labor cost is calculated by applying a rate to effort, which means the accuracy of your workload management becomes the accuracy of your cost figures.
"We came in under budget but there is no profit" is what happens when only external spend is counted and the hours your own people poured in never entered the calculation. Only once time is converted into money can you see whether a given campaign was actually worth running.
Notes on Managing This in a Spreadsheet
The logging sheet itself is perfectly buildable in a spreadsheet with three columns for date, project, and hours. The problem is that the estimates live in the task tracker, the actuals live in a separate logging sheet, and the budget lives in a finance file. Every reconciliation means copying by hand, and that friction is exactly why the review gets skipped. The reason workload management fails is often not a matter of discipline but of files being separated.
Xtrategy manages campaign schedules alongside budget and KPIs on a single screen, which removes the copying from the job of comparing plan against actual. For managing work at the campaign level, see the marketing campaign management template as well.
Summary
Workload management is estimate, record, and feed back. Without the third, it is only tallying.
Pick one purpose. Estimate accuracy, cost visibility, and workload leveling each need a different level of detail.
Five causes of divergence: missing work, full utilization, unbudgeted rework, waiting time, and rounded records.
Effort and duration are different quantities. Convert between them at seventy to eighty percent utilization.
Three fields, thirty seconds. The entry burden decides whether logging survives.
Review only what exceeds a variance ratio of 1.2, and write the patterns you find into explicit coefficients.
Nobody produces accurate estimates from the start. Record the fact that you were off, identify one reason, and feed it forward. Run that loop three times and the difference in accuracy becomes something you can feel.
The fundamentals of scheduling, plus the four reasons plans fall apart: optimistic estimates, other people's time, datin...
