Reading a scrape while it runs: what the numbers mean

A running scrape reports several counters. They reconcile, and knowing how tells you whether a disappointing run is a bad query or just an early one.

The arithmetic

raw found = kept + duplicates + filtered out. Every address the crawler sees ends up in exactly one of those three buckets.

What each one is telling you

• Kept — unique addresses that passed your filters. The number that matters.
• Duplicates — the same address seen again on another page. A high count is normal and healthy; it means you are covering a niche thoroughly.
• Filtered out — addresses removed by your own settings. If this is large, your filter is probably too strict rather than the web being empty.
• Coverage — how much of the selected geography has been swept. Low kept with low coverage simply means it is early.

The most common misreading

A restrictive email-type filter quietly discards most of what is found, and the run looks like a failure. If the filtered count dwarfs the kept count, widen the filter before you change anything else.