Overview
Almost every site in this field organizes itself by platform, or by industry. This one is organized by what a dispute turns on and by where the record lives, because those are the two questions an attorney actually arrives with.
Twelve pages sit under two headings. The disputes divide into fraud, contract and conduct. The evidence divides into platform, measurement and quantum. Six groups, three under each heading, and every page belongs to exactly one. The scheme is deliberate and it is worth a page of explanation, because the organizing choice is the part of a reference site that either helps a reader or quietly wastes an afternoon.
The two questions, in the order they get asked
An attorney with a paid-search matter asks, in this order: what does this dispute turn on, and is that thing recorded anywhere? Everything else — which platform, which vertical, which product — is detail hanging off those two.
The dispute pages answer the first. Each carries a verdict saying whether the account record settles the matter, settles part of it with discovery supplying the rest, or does not reach it at all. The evidence pages answer the second, on a parallel scale: the record is in the account, the record can be reconstructed from several sources, or the number is modeled and never was a record. Two scales, three states each, and on both scales the third state is the one that does the most work. "Not a data question" and "outside the data" are the two conclusions that save a client money, and a structure that cannot express them is a structure built to sell rather than to inform.
Why not by platform
The obvious alternative is a page per advertising platform. It fails for a specific reason: one platform is the record in nearly every dispute in this area, so a platform-shaped site produces several pages that all describe the same account in slightly different words. The change-history window gets written six times. The dispute gets written zero times.
Where platform difference genuinely matters, it is a fact, not a category. The change record on one major platform survives two years and on another six months. That difference is important enough to change a preservation deadline — and it is one row in one table, not six pages of prose. It is handled where it belongs, on the pages about the platform record and about preservation, rather than used to inflate a site map.
There is also a self-inflicted problem with platform-shaped organization that has nothing to do with readers. When several pages address the same subject with the same vocabulary, none of them is decisively the answer to anything. That is a failure state I have watched play out, and this structure exists partly to avoid repeating it.
Why not by industry
Industry organization fails differently. An industry page is the same page with the noun swapped, because the account record does not behave differently by vertical. A change history logs the same fields for a professional-services advertiser as for a retailer. A conversion column blends modeled and observed figures identically in both. The auction mechanics do not change.
Industry does matter in places — lead buying, regulated verticals where what may be advertised is constrained, franchised systems where several advertisers under one brand compete in the same auction. In every one of those, industry is a fact inside a dispute, not a container for it. The franchised-system case is a good illustration: what makes it distinctive is not the industry, it is that the dispute turns on an agreement and a shared fund's terms. That puts it with the other contract matters, next to the agency dispute it actually resembles, rather than in a silo of its own where its real nature would be obscured.
What each group is for
Fraud collects the disputes where the fact in issue is the traffic itself, and where the record is split between the platform and the advertiser's own systems. Contract collects the disputes where an agreement decides the outcome and the account audits the money. Conduct collects the disputes about what was done in an account — including the one where the honest answer is that there is no published standard to measure it against.
Platform collects the records that are exported rather than reconstructed. Measurement collects the numbers that were assembled from several systems and have to be rebuilt before they can be relied on. Quantum collects the money and the preservation work that keeps the money provable, which sit together because a damages figure is only as good as the record still standing behind it.
Where the scheme does not hold
It is a reading aid, not a taxonomy, and there are three things it will not do.
- It does not classify a real matter. Live disputes sit in two groups at once — an agency dispute with an invalid-traffic allegation inside it is both a contract matter and a fraud matter, and it needs both pages. The group is a starting point for finding the right record, not a label for a case.
- It does not rank difficulty or value. A dispute in the group where the data settles it can still fail, and the usual reason is that the window closed before anyone asked. The clock is indifferent to which group a matter is in.
- It does not tell you whether you have a claim. Whether conduct that is plainly visible in an account is actionable is a question for counsel, and I am not an attorney. What the record shows and what follows from it are separate questions, and this site only answers the first.
One last note on why the two headings are not equal in the way they are usually treated. The dispute is what an attorney knows the name of on the first day. The evidence is what determines whether the matter is worth taking further. Most sites in this field write the first and skip the second, which is why they read as advertising. The evidence half here is the longer half on purpose.
Published