Academy / Analytics diagnosis: events, UTMs, and the memo

Reports, booking truth, and the diagnosis memo · 24 min

Diagnosis memo method: symptom → five hypotheses → one fix

In this lesson

  • Turn a symptom into five hypotheses labeled observed, inferred, or unknown
  • Pick one measurement fix (GA4, GTM, UTM, or call tracking) that can ship this week

The memo is a method, not a recap of the CTR table

Lesson 4 taught you which reports to open when Northfield’s CTR jumps. This lesson is the write-up an owner will act on. Structure, every time: (1) Symptom — one sentence copied from what you were given, not a vibe. (2) Five hypotheses that could produce that symptom. (3) Each hypothesis labeled observed (a number or URL you were given or a table you described under a named “if”), inferred (a reasonable mechanism you did not see), or unknown (you would need a login, a week of Search Terms, or a desk export). (4) The one measurement fix you would ship in the next five business days. The fix is a tag, a thank-you URL, a UTM convention, a forwarding number, or a desk join key — not “run better ads” and not “appointments matter more than CTR.” If you only restated lesson 4’s 400/560 table, you did not write a memo.

How to label without performing honesty

Observed: “Sessions are up and booked sales appointments are flat” (Northfield brief). “61 form fills, 22 booked, 9 no-shows, 14 GBP calls to a shared cell” (Cedar). “The GTM trigger is Page Path = /thank-you” — only if the brief or your own stated assumption says so; otherwise that trigger is inferred. Inferred: “Extra clicks are research queries” until Search Terms is open; “the desk is not calling back” until you have speed-to-lead. Unknown: Quality Score, last-click ROAS, which of the 61 forms became jobs, which campaign sourced the 14 rings. A junior marks everything inferred and then recommends a bid change. A professional marks unknown and recommends the measurement that would move one row from unknown to observed.

What “one fix this week” means

Shippable this week: split /thank-you/appointment from /thank-you/newsletter; add form_id to the dataLayer; install click_to_call on every tel:; put a forwarding number on GBP and Search; publish the UTM convention and rebuild the five live URLs; add desk_booked / showed columns and require a call_id. Not shippable as the week’s measurement fix: a new $3,000 budget split, a 15-headline RSA, a brand-awareness video, “educate the team on vanity metrics,” or a conversion rate you invented so the slide has a number. Pick the one fix that would change what next week’s table can say. Tell Dana or Priya the hour cost (half a day in GTM vs a call-tracking vendor) and what remains unknown after it ships.

The memo’s job is to move one unknown to observed. If your “fix” does not change next week’s table, it is a slogan.

Exercise

Practice one labeled row

Pick ONE symptom (Northfield CTR 4.0% → 5.6% with 12 appointments both months, or Cedar’s 14 unlogged GBP calls). Write five hypotheses with observed / inferred / unknown. Circle the one measurement fix you would ship this week. Do not paste the lesson-4 Search Terms table as the whole memo.

Deliverable: Five labeled rows + one circled fix (GA4 / GTM / UTM / call tracking).

Create a free account and verify your email to save progress, quizzes, and portfolio work. Create a free Academy account · Sign in

Reports, booking, and memo check

Same 10,000 impressions. Clicks 400 → 560 (CTR +40%). Desk-logged appointments still 12. Which reports do you open first?
If those extra 160 clicks were research queries, what would the Search Terms table show?
Cedar & Stone booked 22 appointments and had 9 no-shows. What is show rate?
True or false: A website form fill is the same event as a desk-logged appointment.
In a diagnosis memo, how do you label: “Search Terms last Tuesday listed 40 clicks on ‘how to inspect a used car’ ”?

Assignment

Diagnosis memo — five hypotheses, one measurement fix

Write a one-page memo Dana or Priya can act on. Choose ONE symptom (do not mash both businesses into one fog): Northfield’s CTR 4.0% → 5.6% on 10,000 impressions with 12 desk-logged appointments both months, OR Cedar & Stone’s 14 GBP calls to a shared cell plus 61 forms / 22 booked / 9 no-shows. Then: (1) one-sentence symptom; (2) five hypotheses that could produce it; (3) each labeled observed / inferred / unknown; (4) the ONE measurement fix you would ship this week — a concrete GA4 event, GTM trigger, UTM convention, or call-tracking change — plus what remains unknown after it ships. This is not a recap of the lesson-4 Search Terms table. Do not invent conversion rates, Quality Score, ROAS, or DebugView screenshots.

Deliverable: One-page memo: symptom, 5 labeled hypotheses, one shippable GA4/GTM/UTM/call-tracking fix, leftover unknowns.

Write at least a short brief (80+ characters). A mentor reviews for a pass.

Ask the tutor

The tutor will guide you. It will not finish graded work for you.