The hardest question in this whole course: did that work? Visibility data is noisy, seasonal, and slow, which makes it easy to convince yourself of almost anything. This lesson is about not doing that.
Why it is hard
- Everything fluctuates. Rankings, views, and AI answers move without any change from you.
- Effects lag. Directory corrections propagate over months, so this month work may show up in three months numbers.
- Seasonality is real. Psychiatric enquiry patterns vary through the year, and holidays distort short comparisons.
- Confounding is normal. A competitor closed, a payer changed its network, a referrer retired.
- Small numbers are unstable. A practice with twelve enquiries a month cannot detect a ten percent change.
- Motivated reasoning. You want the work to have worked, and there is always a metric that rose.
The four requirements
Before claiming a change worked, you need all four:
- A baseline captured before the change, with a date. Without this you have nothing, and it is the most common omission.
- A dated record of exactly what you changed.
- Enough time. Allow at least two to three months for listing and directory work.
- A comparison that controls for season. Same period last year, not just last month.
If any is missing, describe what you observed rather than asserting causation.
Change one thing at a time
The discipline that makes attribution possible at all. If you change your primary category, rename the practice, correct four directories, and add twenty reviews in the same month, you cannot know what caused any subsequent movement.
Practical compromise, since you also want to get work done:
- Do the foundational work in a batch, since you need all of it and speed matters more than attribution. Record it as a batch.
- After the foundation is in place, make significant single changes discretely, at least a month apart.
- Never change the practice name and the primary category in the same month, as chapter three notes.
Distinguishing signal from noise
Rules of thumb rather than statistics:
- One month of movement is noise. Two or three consecutive months in the same direction is a trend.
- Small percentage changes on small numbers are noise. Look for changes large enough to be visible without squinting.
- A step change coinciding with a platform update is probably the platform, not you. Chapter nine profile lesson covers the definition changes.
- Movement across several independent metrics is more credible than movement in one. Enquiries, calls, and direction requests rising together is real; one rising alone is suspicious.
- Movement in the absence of any change from you is a reminder of how much noise exists.
When nothing changed
This happens, and it is worth reporting plainly. Common and legitimate reasons:
- Not enough time has passed, particularly for aggregator propagation.
- The change was too small to matter.
- You fixed something that was not the constraint. If your listings were already accurate, correcting them further changes little.
- The constraint is elsewhere: missed calls, long waits, or a poor conversion from enquiry to evaluation.
- The market moved against you at the same time.
The fourth is the most common and most actionable. Practices frequently invest in visibility while missing a large share of inbound calls, which no amount of visibility will fix.
The honest posture
Say what you know, say what you do not, and be willing to conclude that a change had no measurable effect. That conclusion is valuable: it stops you repeating it and frees the effort for something that does work.
A practice that measures this way accumulates real knowledge about its own market over a couple of years. One that claims a win every month accumulates nothing. Chapter ten turns all of this into checklists you can run.