01
Establish what is true
We preserve the starting condition and identify its source, definition, date range, and limits. If a baseline does not exist, we say so instead of constructing one after the fact.
Public evidence standard, version 1.0
We establish what is true, correct what is broken, and verify the result before we call it marketing success.
This is the standard Sequoia now applies to new and revised assessments, case studies, and client reporting. It keeps technical work, platform activity, leads, and jobs booked from being blended into one flattering number.
Published August 29, 2026. Current version 1.0.
The operating method
01
We preserve the starting condition and identify its source, definition, date range, and limits. If a baseline does not exist, we say so instead of constructing one after the fact.
02
We name the defect, record the work completed, and separate our correction from strengths the business already had.
03
We test the corrected condition against a stated acceptance check. A page loading, a form arriving, and a job being booked are different results and remain separate.
04
We label later observations by their actual stage and state what they do not establish. Work completed is not automatically a lead, and a lead is not automatically a job booked.
Proof status
Every published proof record receives one of these labels. The limitation travels with the claim so readers can see what the evidence supports and where it stops. Status describes what has been established. Evidence class identifies where the fact came from. Neither label replaces the other.
A starting condition was checked against a dated public, platform, account, or operating record.
That Sequoia created the condition or that it will remain unchanged.
A named defect was corrected and the change was recorded.
That the correction changed traffic, inquiries, qualified leads, or jobs booked.
The corrected condition was tested against a stated acceptance check.
That the verified condition caused a later business outcome.
A dated measurement was observed after work was completed.
That Sequoia caused the movement unless a defensible causal design is stated.
The operating business confirmed an inquiry, qualified lead, or job booked under a stated definition.
That a prior marketing action caused the outcome unless identity and source are joined.
The evidence needed to support the next claim does not exist, cannot be obtained, or has not passed review.
Anything beyond the stated evidence gap.
Evidence class
Sequoia read the result from a named platform report, system record, or retained artifact.
Publication rule: Publish the metric definition, date range, read date, source, and limitation together.
The operating business confirmed the fact and Sequoia did not independently read the underlying system.
Publication rule: Attribute the statement to the client and do not present it as independently measured.
The condition was real before Sequoia became involved.
Publication rule: Show it as starting context, never as an outcome Sequoia produced.
A condition was visible on a public surface at a recorded point in time.
Publication rule: State the method and observation date, and do not imply account access or persistence.
A fact was assembled from partial records or later recollection.
Publication rule: Label it as reconstructed and keep it out of headline proof or result tiles.
The necessary evidence was not retained, cannot be accessed, or does not exist.
Publication rule: Name the gap. Do not replace missing evidence with a platform estimate or a stronger narrative.
The outcome ladder
Progress can move through these stages. We do not skip a stage or promote a record because the later label would sound more impressive.
If the required evidence for a stage is unavailable, the record stays Not established and does not enter the ladder.
Causation is a separate claim.
A result observed after our work is reported as observed after. We use causal language only when the comparison design supports it, such as a valid pre-period, holdout, or controlled panel.
AI search uses a channel-specific expansion of platform activity: Mentioned, Recommended, Primary recommendation, Cited, and Referral visit. Inquiry, Qualified lead, and Job booked remain distinct stages. The AI page applies Sequoia’s own pipeline qualification rule as a business-specific instance of the general rule above. The AI search measurement method defines those additional observations.
Before publication
Our refusal list
See the standard applied
Our current case-study library is being migrated to verified starting conditions, corrections, checks, and evidence limits as records and permissions are confirmed.
View current case studiesSee how we separate mentions, recommendations, citations, visits, inquiries, and jobs booked.
Read the AI methodStart with a public-surface assessment that documents the current condition before recommendations are made.
Request an assessmentVersion 1.0 published the four-step operating method, six proof statuses, six evidence classes, outcome ladder, causation rule, publication requirements, and refusal list.