What does it mean that tool bottlenecks?

Sometimes a tool is the constraint and upgrading or replacing it unblocks the whole system.

Example: A data analyst runs a market segmentation query on her 8GB laptop every Monday morning. The query takes 47 minutes. She makes coffee, checks email, gets pulled into a Slack thread, and by the time the query finishes she has lost the analytical thread she was following. On a properly specced machine with 32GB of RAM and an SSD — a $1,200 upgrade — the same query runs in under 3 minutes. She stays in flow, iterates on the results immediately, and produces her weekly report in 90 minutes instead of four hours. For two years, she assumed the bottleneck was the complexity of the analysis or her own speed. It was the laptop. The tool's throughput ceiling had become her throughput ceiling, and she had adapted so completely that she stopped seeing the constraint.

Try this: Pick the single workflow you perform most frequently — the one you do daily or multiple times per week. Time each discrete step of that workflow with a stopwatch or timer. For each step, mark whether you are actively thinking and creating, or waiting for a tool to respond (loading, processing, rendering, syncing, exporting, building, compiling). Calculate two numbers: total active time and total tool-wait time. If tool-wait time exceeds 15% of total workflow time, you have a candidate tool bottleneck. Write down the specific tool, the specific step where the wait occurs, and the duration of the wait. Then estimate: if that wait dropped to near zero, how much faster would the full workflow complete? That delta is the throughput you are currently leaving on the table.