Batch Similar Work Across Companies, Not One at a Time

Do not finish one company before starting the next. Do all the same type of work across every company in one sitting, then move to the next type. That is the single change that cut my switching cost the most. When you run a portfolio, the expensive thing is not the work. It is the reload. Every time you swap from one company's context to another, your brain pays a tax to page in the details. Batching by task type instead of by company means you reload once per type instead of once per company.

Batch by task, not by company

Say I have twelve companies that each need an invoice reviewed, a support queue checked, and a blog post approved. The intuitive path is to open company one, do all three, close it, open company two. Feels tidy. It is the worst possible order. I just made myself reload twelve full contexts and switch task types thirty-six times.

The batched path: review all twelve invoices in one pass, then check all twelve support queues, then approve all twelve posts. Now I reload the task context three times, not thirty-six. Inside each batch the work is nearly identical, so I stay in one mental mode. Invoice brain. Support brain. Editorial brain. I explained why this reload is so costly in the context switching tax a solo operator pays.

Why context switching is the real cost

The switch is not free and it is not fast. Studies put the recovery cost of a hard context switch in the range of minutes, and my own experience is worse than the studies. When I jump from a legal question in one company to a pricing question in another, the first several minutes in the new context are garbage. I am not thinking about the new problem yet. I am still unloading the old one.

Batching keeps me inside one problem shape long enough to get good at it for that session. By the eighth invoice I am fast, because I have not left invoice mode. The same principle drives why I keep one metric per company instead of a sprawling dashboard per venture. Fewer shapes to hold means faster passes.

What batches well and what does not

Not everything batches. The rule is simple: work batches when the task type is the same and the company context is shallow.

Routine, repeatable, low-context work is perfect. Invoice review, email triage, social approvals, metric checks, dependency updates. These have a fixed shape and you can rip through them.

Deep, company-specific work does not batch. A strategy decision for one venture needs that venture's full context loaded, and there is no shortcut. Those get their own dedicated block, one company at a time, the opposite discipline. The skill is knowing which is which. If a task needs me to think hard about one company's specific situation, it is not batch work. If it is the same motion repeated across companies, it is.

Tools that make batching possible

Batching only works if the same task looks the same across every company. That is a systems problem. If each company runs different software, invoice review in company one looks nothing like company two, and the batch falls apart because you keep re-learning the interface. This is why I standardize hard. One billing view, one support surface, one deploy path. I use a shared operations layer at reflexware.com so that the same task genuinely is the same task everywhere, which is the precondition for batching to save time instead of just reshuffling it.

Automation extends the idea. The best batch is the one you do not do at all because an agent handles the whole pass and only surfaces the exceptions. I push routine batches into automated workflows and review the outliers, which turns a twelve-company pass into a three-exception pass.

How to start

Look at your week and label each recurring task by type. Then group your calendar by type instead of by company. Pick one afternoon for all financial reviews, one morning for all approvals, one slot for all metric checks. Resist the urge to close out a company just because you are already in it. Leave it half done and move to the next company's version of the same task. It feels wrong for about a week. Then it feels obvious. Your focus stops shattering into thirty pieces a day, and the portfolio starts to feel like one job with repeating parts instead of twenty jobs fighting for your attention.