How to Batch Owner Work Into Blocks

Why this matters

Every owner has been told to batch their work. Almost nobody is told how long a batch should be, how often it should run, or which tasks actually belong in the same one, so the advice gets implemented as "do all the quotes Friday afternoon" and then abandoned when a customer waits four days for a number and buys from somebody else.

Batching done properly is arithmetic, not attitude. There is a setup cost you pay every time you enter a kind of work, a minimum block length that follows directly from that setup cost, and a maximum cadence that follows from what you have promised customers. Those three numbers decide your block schedule. This is how to derive them from your own work.

Batch by shared setup, not by shared label

The standard advice says group similar tasks. Similar is the wrong test, because it groups by what the task is called rather than by what it costs to start.

The right test is shared setup: what you have to load, open, retrieve, or get your head into before you can do the work at all. Two tasks belong in the same block when starting the second one requires almost nothing that starting the first did not already provide.

Quoting a replacement and quoting a repair look like the same task and are, because both need the price book open, the customer history in front of you, and your head in the frame of what this customer will and will not spend. Approving a discount and reviewing a quote also look similar and are not, because the approval takes ten seconds against a written limit while the review needs the whole quoting context loaded.

Run the test on your own list by asking, for each pair: if I did these back to back, what would I have to go get for the second one? Nothing means same block. Anything substantial means different blocks.

Measure the setup cost, because every rule below depends on it

Time three instances of entering each kind of work. Start the clock when you decide to do it, stop it when you are actually producing rather than gathering. Take the middle of the three, not the average, because one of the three will include finding something that was lost.

Most owners have never measured this and guess low. Quoting setup in a shop with the price book in one place, job history in another, and the customer conversation living in the owner's memory routinely runs into double-digit minutes. That single number explains why quoting one at a time is so expensive and why nobody ever notices.

The minimum viable block

A block must be at least four times the setup cost of the work in it. Setup of 10 minutes means a minimum block of 40 minutes. Below that ratio, setup is more than a fifth of the block and you are paying a tax large enough that batching stops being worth the scheduling effort.

Unit of analysis: minutes of setup for one kind of work, for one person, measured as the median of three timed instances. The ratio is a floor, not a target - a block can be longer.

Why four and not two: at two times setup, half the block is overhead, which is barely better than not batching. At four times, overhead is 25% of the block and each additional instance you add drops it further. The ratio is a practical floor that keeps you from scheduling blocks so short that they cannot repay their own scheduling.

The ceiling, and why longer is not better

Cap a single-kind block at about 120 minutes. Past that, on work requiring sustained judgment, most people's output per instance degrades faster than the saved setup gains, and the last quote in a four-hour quoting session is worse than the first one in the next session would have been.

If the volume does not fit in 120 minutes, add a second block on another day rather than extending. Two 90-minute blocks beat one 180-minute block for the same reason two service calls beat one very long one: the second setup costs you once, and the fatigue costs you on every instance after the first hour.

The ceiling is softer than the floor. Routine work with low judgment content - a paperwork batch, a filing run - tolerates a longer block fine. The 120-minute figure is aimed at work where being sharp on the last item matters.

The cadence rule: batching cannot outrun your response promise

This is the rule that gets left out, and it is the one that turns batching into lost work.

Your batch cadence must be at least as frequent as the shortest response time you have promised for the items in the batch, measured from the worst case rather than the average. Worst case means an item that arrives just after a block ends.

If quotes are promised within one business day, a twice-weekly quoting block cannot deliver, because an item arriving the afternoon after Wednesday's block waits until Friday. Two business days, not one. The average wait would look fine and the promise would still be broken on a third of the volume.

Work the worst case, not the average, every time. Averages hide exactly the customers who are about to buy from somebody else.

When the two rules collide

They will, and the collision is where most batching attempts quietly fail. Frequent cadence pushes toward short blocks; the minimum-block rule forbids short blocks. When both cannot be satisfied:

Cadence wins, and you attack the setup cost instead. The promise is external and costs you sales when broken. The block floor is internal and exists only to protect you from setup overhead - so remove the overhead rather than breaking the promise. Reducing setup lowers the floor and lets the frequent cadence fit.

Setup cost is far more reducible than owners assume, and it is nearly always about pre-staging: the material gathered before you sit down, by somebody else or by a standing template. That is the third path, and it is better than either side of the collision.

Working it through on quoting

An owner handles 9 quotes in a normal week. He times his setup at 12 minutes: pulling the current price book, finding the job history, and reconstructing what the customer actually said on site. The quote itself averages 18 minutes of work once he is in.

One at a time, as he does it now, he pays setup on every one: 9 quotes at 12 plus 18 minutes each is 9 times 30, which is 270 minutes, or 4.5 hours a week.

Batched three per session, three sessions a week, he pays setup three times: each session is 12 plus three times 18, which is 66 minutes, and three sessions is 198 minutes, or 3.3 hours a week. That saves 72 minutes a week, since 270 minus 198 is 72.

Check the floor: setup is 12 minutes, so the minimum block is 48 minutes, and a 66-minute session clears it.

Now check cadence, and this is where it falls apart. He promises quotes within one business day. Sessions on Monday, Wednesday and Friday mean a request arriving Wednesday afternoon waits until Friday, which is two business days. The schedule fails the promise on every request that lands just after a block.

So go daily. Five sessions, 9 quotes, averaging 1.8 quotes per session: each session is 12 plus 1.8 times 18, which is 12 plus 32.4, or 44.4 minutes. Weekly total 222 minutes, 3.7 hours, still better than 4.5.

But now run it against the floor rule, which is the step people skip: the minimum block is 48 minutes and the average session is 44.4 minutes. It fails. Under the rules as written, the daily schedule is not a legitimate batch, it is one-at-a-time work with a calendar entry on it.

Apply the collision rule: cadence wins, attack setup. He asks his office person to pre-stage each quote request as it arrives - job history attached, the customer's own words typed into a note while the call is fresh, current price book link at the top. His measured setup drops from 12 minutes to 6.

Everything recalculates. Minimum block is now four times 6, which is 24 minutes. Daily sessions become 6 plus 1.8 times 18, which is 6 plus 32.4, or 38.4 minutes, comfortably above the 24-minute floor. Weekly total is 5 times 38.4, which is 192 minutes, or 3.2 hours a week. Against the original 270 minutes he saves 78 minutes a week, since 270 minus 192 is 78 - and unlike either earlier version, this one satisfies the promise and the floor at the same time.

Worth naming what actually produced the win. It was not the batching. Batching alone got him from 270 to 198 minutes but broke a customer promise, and the version that kept the promise only worked once the setup cost came down. The pre-staging was the load-bearing change, and he would never have found it by trying harder to concentrate.

The same procedure on his approvals came out shorter. Setup 3 minutes, 30 approvals a week at about 4 minutes each. Three sessions a day averages 2 approvals per session at 3 plus 8, which is 11 minutes, below the 12-minute floor. Two sessions a day, at 3 approvals each, gives 3 plus 12, which is 15 minutes, clears the floor, and still meets his internal promise of answering approvals within half a business day. Ten sessions at 15 minutes is 150 minutes a week against 210 minutes done one at a time, so 60 minutes saved.

Two batches, about 2.3 hours a week recovered, and no new hire.

Placing the blocks

Deriving the blocks is half the job. Where they land decides whether they run.

Put the highest-judgment block where your interruption count is lowest, using a real count from an interruption log rather than an impression. Then put the lowest-judgment block in the noisiest band, because a paperwork batch survives being interrupted and a quoting batch does not.

Do not put a daily block first thing unless your log says the early band is quiet. In most field-service shops the first two hours of the day carry the heaviest interruption load in the week, because that is when crews are rolling and problems surface. That band is for coverage and dispatch, not for concentration.

Anchor each block to an event, not just a time. "After the huddle" gets started far more reliably than "at 7:30," because the huddle actually happens and 7:30 is an aspiration.

What does not batch, and one thing that must never be batched

Three categories stay out of blocks:

• Anything with a hard external clock inside it. A supplier who answers only in a two-hour window sets your time, not your batch.
• Anything where instance two depends on instance one's outcome. Sequenced work is not a batch, it is a chain, and pretending otherwise means the whole block stalls on the first item.
• Anything whose value decays in minutes rather than hours. A customer deciding between you and another shop on the phone is not a batchable item.

And the one that is not a judgment call: a report of a live hazard is never queued into a block. A suspected gas odor, water reaching live electrical, an open panel with people nearby, a fall exposure. Occupants leave the structure immediately, no light switches or appliances touched, no phone used inside, and the call to the utility emergency line is made from outside. That happens the moment the report arrives, whatever block you are in.

References

• See related: Batching vs Switching: The Cost of Interruptions
• See related: The Interruption Log SOP
• See related: How to Design an Owner's Week That Survives Contact
• See related: The Energy Audit: When You Do Your Best Work
• Trade-standard practice for setup reduction and work sequencing