The public results ledger

We publish the results either way.

A firm that only shows a curated case-study reel is choosing what you see. We keep a public, dated ledger of what actually happened, including the runs that underperformed. Publishing the misses is the costly, hard-to-fake signal, and it's the most literal expression of what this brand is.

Entries · newest first

Published study · Legal

Who AI cites for "best lawyer" queries

The desk's first public study, run on our own market rather than a client's. We sampled the AI citations a "best lawyer" search returns and classified every source.

Citations sampled236
Firms' own sites67%
Earned pickupTrade press
Baseline · Taqtics own domain

Our own starting line, frozen in public

Before we claim a thing, we publish our own before-picture: per-platform citation presence, mention rate and link rate as two numbers, human-origin demand separated from raw counts. As a cold-start newsroom our own numbers are thin, and we show them thin rather than dress them up.

Per-platform readRecorded
Two-number cutRecorded
Lift claimedNone
Template · how an engagement will appear

Including the ones that miss

When engagements run, each enters here with its holdout-measured read, whether it beat the control or not. This slot is the template, not a result. We won't quietly delete a run that underperformed, because deleting it is exactly the manufactured-authority move we sell against.

Measured againstHoldout
Outcome shownEither way
AnonymizedWhere required

The honest read is the close.

If a ledger that publishes its misses reads as more trustworthy than a wall of five-star quotes, that's the whole idea.

Sources & method

  1. Holdout-verified reads: measure lift over a pre-registered control, the one number attribution can't fake. Aggregated public research.
  2. Legal citation study, n=236, 67% to firms' own sites; earned pickup Search Engine Roundtable, Jul 6, 2026. Full study.
  3. Publish-the-results-either-way ledger, baselines and misses included. Aggregated public research.