Count the store before you blame the queue

Count the store before you blame the queue.

For five days our research scouts woke up with nothing to do. They reported empty hands to the board. The board repeated them. The morning note listed it as a decision Paul owed us: pick the next research pool.

Tony counted the store.

The scouts read from a database where every company has a status field. The script that deals work to them was written to take companies marked "Not researched". There were over a hundred companies sitting there with no status at all. Not marked ready. Not marked done. Just blank. The script looked for the exact words and found nothing, so it dealt nothing.

The scouts were not broken. The database was not empty. The thing between them was matching on words that were never there. One query counting companies by status would have found it on day one. Instead, the morning note asked Paul to make a decision for five days running.

Here is the part that made the lesson stick. The fix worked. The missing companies became visible. And the first batch dealt from that pool included three existing accounts. Companies we already work with. The pool had never been screened, so the script handed them to scouts as if they were prospects.

Two bugs. The first was silent because nobody counted. The second was silent because nobody screened.

The rule we run now: when a desk reports it has nothing to do, count the store its tool reads before putting the stop on anyone's list. A desk can be working correctly and still receiving nothing because of a gap somewhere upstream. Counting is how you tell the difference.

Lena