Dealer directory data is wrong in more places than one
By Jason MurphyPublished 4 min read
By Jason MurphyPublished 4 min read
The short answer
Dealer directory data goes wrong because one dealer is described in many places: the manufacturer's locator, distributor portals, aggregator feeds, map listings and the dealer's own site, each copied from a different moment. No directory is the record. A correction made in one place does not travel to the others, so the error returns.
A dealer is not one listing. It is a set of copies, each maintained by a different party for a different purpose, and each aging at its own rate.
The visible errors are address and phone. The costly ones are structural.
Correcting a downstream copy fixes it until the upstream source refreshes it. Aggregators pull from suppliers who pull from each other, so the same wrong value arrives again from a different direction. Each source has its own claim process, its own owner and its own review cycle, and none of them knows that the others exist.
The deeper problem is that no copy says it is the authority. A directory rarely records where a value came from, who confirmed it or when. Without that, a correction is an opinion competing with other opinions.
An AI assistant asked which dealer carries a line, or where to buy a part near a job site, reads several of these copies and has to choose. It has no way to tell which is current unless the sources carry dates and an order of authority. In practice it weighs agreement, prominence and recency, which favors the copy repeated in the most places, not the copy that is right.
That makes directory errors worse than they were. A wrong listing used to misdirect the person who found it. Now it can become the answer a system gives to everyone who asks.
The fix is not another directory. It is one checked record of the dealer's facts in which every value carries its source, the person who confirmed it and the date, and in which each kind of fact has a named owner.
The useful output of that comparison is a list of disagreements ranked by who can fix them: values the dealer can correct today, values only the manufacturer can correct, and values that are wrong because a feed is stale. That is how a directory problem becomes an operational queue instead of a recurring surprise.
Lehvel builds this record and runs the comparison. The method is described in Business Truth and in Lehvel's product information chain study.
Directories are copies fed by different upstream sources. A downstream correction is overwritten when the upstream value refreshes, and the other copies never saw the change.
None of them. Authority depends on the kind of fact: the manufacturer for authorizations, the distributor for territory, the dealer for hours and stock. A checked record states that order explicitly.
They have no built-in way to tell which is current. Without dates and an authority order on the sources, they tend to favor the value repeated in the most places.
The number of downstream copies that disagree with the record, which facts they disagree on, and who can correct each one.
Labs · Field notes
Labs · Field notes
Research report
A dealer network disagrees with itself by design. The fix is deciding, per fact, who wins.
Dealer and channel data3 min read
One physical location, three identities. Machines count them as three businesses.
Dealer and channel data3 min read
Most of a company's public facts are writable by someone else. The first step is knowing by whom.
What AI says and why3 min read
An assistant shopping for a car asks about one vehicle, not the dealership. None of what it asks is on a homepage.
Dealer and channel data6 min read
Three sources, three kinds of authority. The assistant has to rank them without being told how.
Dealer and channel data3 min read
One email a week: what we asked the AI, what it got wrong, and how we checked.
Our research reaches 17,000+ followers across our channels. Upstream is the weekly version.
Weekly. Unsubscribe anytime. Privacy