A pre-earnings channel check is designed around compliance and timing before it is designed around questions. The screen decides who may be approached; the quarter decides when the answer is still informative.
What does the compliance screen check?
Whether a prospective participant has access to material non-public information about the subject company — through employment, board position, supplier relationship, or otherwise. The screen runs before any name reaches the analyst.
Sequencing is the point. A conflict identified after an analyst has seen a name is a different problem from one identified before, and the process should be designed so the second never happens.
Designing the sequence properly also protects the analyst rather than only the firm. An analyst who has never seen the name of a rejected candidate cannot be said to have received anything, which is a materially better position than one who saw a name and declined to proceed. That is why the screen belongs with the provider rather than with the desk.
- 01
Clear the screen
Run compliance before any name reaches the analyst, not after.
- 02
Fix the window
Set the field window against the quarter and the restricted period at scoping.
- 03
Build the frame
Cover the channel across its layers rather than going deep on one.
- 04
Read direction
Interpret the result as direction and sequence, never as a number.
When should a check be run?
Late enough in the quarter to reflect it, early enough that participants are not inside a restricted period. That window is narrower than most analysts assume and should be fixed at scoping rather than discovered.
Running early reads a stale period; running late reaches people who cannot speak. Neither produces a usable answer.
Fixing the window at scoping also forces a decision that is otherwise made by default. Checks commissioned late in a quarter run into restriction periods and checks commissioned early read a period the market has already priced, and both failures are avoidable by deciding the date at the same time as the question. It is a scheduling problem more than a research one.
Who can you speak to?
Channel participants, former employees outside any restriction, and industry observers — each having cleared the screen before a call is booked. The constraint is what they know rather than what you intend to ask.
Former employees are the category requiring most care, because restriction periods and continuing obligations vary.
Industry observers are the most underused of the three categories. Trade press, association staff and consultants who serve the sector sit outside any restriction, see across several companies, and are frequently more willing to characterize a quarter than a participant who has to be careful about their own employer. They are weaker on specifics and stronger on direction.
| Direction | Magnitude | |
|---|---|---|
| A check supports | Better or worse than the prior period | Nothing reliable |
| Based on | Consistent accounts across the frame | Estimates participants are guessing at |
| Fails when | The frame covered one layer | Always, if presented as a number |
What can you legitimately ask?
About the market and about the participant's own business. Questions that would require someone to disclose a subject company's non-public information are out of scope however they are phrased.
A well-designed guide makes that boundary obvious to the participant, which protects them as well as you.
How should you read the result?
As direction, not magnitude. A check tells you whether conditions in a period were better or worse than the prior one; it does not produce a number, and treating it as one is how a divergence from consensus gets over-interpreted.
The strongest use is disconfirmation — establishing that a consensus assumption is not supported by what the channel is seeing.
Disconfirmation also has a practical advantage: it is easier to act on with confidence. Establishing that an assumption is not supported requires less evidence than establishing what is true instead, and for a position that rests on consensus being wrong, that is the whole question.