Why we write the brief in words, not charts

When MarketWatch was twelve months old we had a customer ask us, very politely, whether we could build them a dashboard with thirty-seven competitor metrics on it. Thirty-seven was a specific number — they wanted it because their existing intelligence vendor had thirty-six and they wanted us to be one better. We talked to four other customers that month and everyone had a version of the same ask. We very nearly built it.

What stopped us was a conversation with a brand director at one of our larger Growth-tier accounts. We asked her, casually, when she had last opened the existing intelligence dashboard her team paid £24,000 a year for. She thought about it for a long time and said: "in earnest, December." It was March. The dashboard was great. It just wasn't a habit.

The lesson we took was that the bottleneck in competitive intelligence is not the volume of metrics. It is the volume of attention. The dashboard sits there and waits for someone to come to it. The brief arrives in their inbox, says one specific thing, asks for one specific decision. Charts are wonderful inside the brief. Charts as the brief, instead of the brief, is what doesn't get read.

There is a second reason. Charts cannot lie, but they cannot reason either. A chart can tell you that Bramleigh's price moved from £24 to £19 and a chart can tell you that this is the lowest price in 90 days. It cannot tell you, in a sentence, that this is consistent with their last-Christmas pattern and historically they hold the price for under a fortnight. The reasoning is what makes the signal useful. We pay our analyst pod to put that reasoning in. The Compass language model layer formalises the same reasoning at scale.

We do still build charts, by the way — every brief has a thirty-day price history embedded against any SKU we are flagging, and a sentiment delta sparkline against any review-shift call. Charts are the citation, not the message. Words are the message.