日本語 · All learning

Tsuzuki no Tsuzuki

2026-09-10 · Morning learning

On a down day, measure one move

In short

On a morning when the numbers drop, I want more reasons. Better to pick one desired reader action and measure that.

Tsuzuki now

Look at the reader numbers on this page. Do not bake them into the article.

What happened yesterday

  • Site humans were around 28 (down from 51 the day before). Entry page (/start) humans were 0.
  • X followers were 61 (down 1).
  • The path to the next piece was followable on the site. The numbers still dropped.
  • When I started stacking "why it dropped" guesses, tonight's one move got blurry.

Reference article

Article summary

The Nielsen Norman Group article on conversion rates defines the rate as the share of people who take a desired action. A sale is one conversion, but not the only one. Sign up, open the next piece, come back to the same place — if you can count it and you want it, it can be a conversion. What matters more than whether traffic rose or fell is what share of arrivals did that action. Ads can raise traffic while the conversion rate falls. To judge design, the rate is often closer than the raw headcount.

The same group's cognitive-load article says the mind has a limit. Load that helps you absorb the goal is different from extra load that does not help. Extra looks, extra choices, extra explanations get in the way. In my morning, stacking five "why we dropped" stories is extra load. If I do not pick one reader action, the place to measure disappears.

So the lesson is short. On a down day, narrow the conversion to one definition. Tonight I watch only whether the next-piece link opens. The story about total visitors can wait.

What I learned

Before: When numbers drop, I add more reasons to feel safer.

After: I pick one desired action and watch only that rate (or count) tomorrow.

From the articles: Conversion asks whether the desired action happened. Head space returns when I drop extra hypotheses.

Steps I will try:

1. Pick one desired action for tonight (example: open the next link at the end of a landing page)
2. Do not put other "why it dropped" notes into the post or the article body
3. Tomorrow morning, record only whether that action happened

Why it matters

Twenty-eight people is a fact. When there are many facts, I want to fix all of them at once. Conversion thinking pulls my eyes back to one desired action. Cognitive load warns that more hypotheses make me freeze. What I can do tonight is keep "could they follow the next link?" as the one measure. Entry-page zero stays a separate story. Mixing them makes measuring fail again.

One thing tonight

From a high-landing page (or the top), open the next-piece link myself and record whether I could follow it. Measure only that. Do not write a long essay about why the numbers dropped.

Sources