What research has Mindflow actually published?
Five things. Two are compilations of public data, built from the NOAA storm record and labelled as compilations rather than research of ours. The other three are the measurement method, a real client report, and the claim checks we ran before publishing figures other people assert without checking.
The two compilations exist because the underlying data are public, complete and checkable. Both were mapped with titles the data did not support, and both were retitled rather than published as briefed. Neither contains a marketing figure, because the storm record contains none.
| Published item | What it is |
|---|---|
| The Texas hail belt, measured → | Ten years of NOAA hail records for all 254 Texas counties, counted as distinct hail days. The belt turns out to be two cores rather than one corridor, and per square mile Texas ranks last of five plains states. |
| Florida storm season, from the record → | Ten years of NOAA storm records for Florida. The damaging-weather peak is June and July rather than September, and the page carries no dollar figures because the source field will not support them. |
| Claim checks → | Claims we tried to source before publishing them. Three of six failed verification and became corrections. |
| The X/12 Protocol, published in full → | The complete measurement method behind AI Share of Answer: question selection, surfaces, run counts, scoring and known limitations. |
| The monthly report, in full → | What a client actually receives, with the numbers left in and the competitor set held constant. |
Why is the benchmark series still empty?
A published benchmark needs a dataset behind it, gathered on a stated method, at a size that makes the number mean something. Public weather records clear that bar, which is why the two compilations above exist. What the entries below need is measurement of our own, and what we hold is client runs under the X/12 Protocol: small, self-selected and confidential.
So rather than dress up a handful of client runs as an industry study, these sit here as unbuilt. Each entry says what it would take.
| Benchmark | Status | What it would take |
|---|---|---|
| Share of Answer benchmarks by trade and metro | NOT COLLECTED | Requires quarterly X/12 runs across a defined trade and metro matrix. We run the protocol for clients; we have not yet run it at the scale a published benchmark would need. |
| State of Home Services Search, annual survey | NOT COLLECTED | Requires a surveyed sample of roughly 300 to 500 business owners. No survey has been fielded, so there is nothing to report. |
| Review velocity benchmarks by trade | NOT COLLECTED | Requires collected review data across a defined set of businesses over time. Our review velocity calculator does the arithmetic on your own numbers in the meantime. |
| Map-pack volatility notes | NOT COLLECTED | Requires tracked position data over a sustained period across a fixed keyword and location set. |
| Local visibility benchmarks by market | NOT COLLECTED | Planned for Atlanta, Charlotte, Tampa, Georgia fencing and Florida review velocity. Each needs its own measured dataset rather than a public-data compilation. |
What would have to be true before any of these ships?
A stated sample size and how it was drawn. A method published before the result. A date on every figure. Named surfaces rather than a blended score. And a limitations section written by us rather than inferred by you.
That is the same standard the X/12 Protocol already meets, and it is the reason this page is short. Most published marketing research in this category would fail it, which is precisely why another one would add nothing.
The protocol itself, including its limitations, sits on the methodology page.
Rankings are never guaranteed. Anything we could not trace to a primary source is absent from this page, not estimated. The audit runs the same six layers described on the pricing page, and Share of Answer is scored quarterly.
Fair questions about this page
Why are the benchmarks empty?
Because we have not collected the data behind them. A published benchmark needs a stated sample, a method published before the result and a date on every figure. Client measurements are small, self-selected and confidential, and dressing them up as an industry study would fail our own standard.
When will the benchmarks be published?
When the underlying data exists. We are not putting a date on it, because a cadence promise on a page about verification would be the wrong thing to make.
What can I use in the meantime?
The X/12 Protocol is published in full, a real sample report is available with the numbers left in, and the claim checks cover the figures most often quoted without a source.
A measurement of your own beats a benchmark
A free Visibility Check runs twelve real buying questions across Google and AI, for your business. You keep the findings.
Get my free Visibility Check