The Resolution Layer · Case Study in BriefEKOM
Outdoor power equipment.
A cold read of the powered-equipment range on a manufacturer's own storefront — chainsaws, blowers, trimmers and brushcutters, where a buyer chooses a machine by its specification table rather than its looks. One of a set of anonymized EKOM analyses — what surfaced, and where it leads.
Lead finding · specification values no machine could have
A chainsaw at ten times its sibling's power speed, and zeros where values belong.
One gas saw carries an inserted zero on its power speed and a dropped decimal on its oil tank; its sibling, with the same engine, publishes the right values. Elsewhere the range records a zero-volt battery, a spark plug with no gap, a brushcutter rated silent with handles that do not vibrate, and an oil pump that delivers nothing. Each zero sits in a field comparable products fill — and one, a zero-length bar, was enough to start the storefront's own bar-length filter at zero.
Also surfaced
One machine, two answers
Battery machines are typically listed twice, as a bare tool and with battery and charger. On five models the two listings disagree — a brushless motor on one and a brushed motor on the other, two voltages for one battery system, vibration figures on one listing only.
The figures the biggest machines leave out
The two most powerful backpack blowers publish no sound, vibration or emissions figure, where their smaller siblings publish all of them. Five products in the range carry no sound data at all.
Two words for one fuel
Every gas leaf blower names its fuel with the British word; every other gas machine uses the American one. The filters split the same way, and a query in the American word returns no leaf blowers.
A list nobody prunes
Of 61 promotions the product pages read, 54 had already ended — the oldest nearly two years before the read. An expired offer renders exactly as no offer does, so nothing ever flags it.
How EKOM read it. The engine read the structured data the manufacturer embeds in its own product pages — no file, no credentials, no schema — archived it on the day, and re-checked the headline findings live. It profiled what each field is for from how the range itself uses it, then held every machine against its siblings: the bundle against its bare tool, the saw against the one with the same engine. It learned the range's own conventions and reported where they stopped.
From diagnosis to channel-ready. Most of the 30 findings collapse into six upstream mechanisms — each packaging variant kept as its own record, missing values written as zero, one product family loaded on a path of its own, units and formats typed by hand, one record transcribed badly, and a promotions list nothing prunes. Fixing the findings fixes the findings; making a bundle inherit its specification from the bare tool fixes every bundle launched after. Then the same intelligence holds the line at intake, so the next zero never reaches a live page.