Sorting Ten Weeks of Signals Into the Four Pattern Types

Marketing Academy · Field Work●Mental Models

All projects·From: Pattern Recognition: The Marketer's Meta-Skill

MiniAudit· 25 minutes

Zomato

Objective: Given a 10-row log of surprising weekly marketing signals from Zomato's growth team, classify each into temporal, behavioral, language, or structural pattern types and flag which ones already meet the 3+ occurrence bar for a pattern hypothesis.

You're the growth marketing analyst at Zomato reviewing the team's pattern journal after a chaotic quarter of festival promos, app updates, and competitor moves.

Classify each logged signal by pattern type, then flag which type has enough repeat occurrences to justify a test.

Before you start

What you'll need

Free path (everything below is enough to finish)

Google Sheets

FreeTag and count signals by pattern type

Free, no account friction, sorts and filters instantly

The process

1 step

Step 01 of 01

Classifying signals into the four pattern types before scoring them

The lesson splits every marketing signal into temporal, behavioral, language, or structural pattern types before any of them earn a test budget.

Given this 10-row signal log, which pattern type has the most independent occurrences and is the safest one to act on first?

Google Sheets— Import signal-log.csv, add a Pattern Type column, filter and count by type.

Procedure

Import signal-log.csv, freeze the header row
Tag each of the 10 rows with one of the four pattern types
Filter and count occurrences per type
Flag any type with 3 or fewer occurrences as not yet test-ready

Sample output

SIGNAL LOG (10 rows, tagged)
  Row 2: Ad CTR drops every week 3 of a campaign -> TEMPORAL (4th occurrence this year)
  Row 5: Support tickets use the phrase 'too many steps to order' -> LANGUAGE (2nd occurrence)
  Row 7: Users who reorder within 48 hrs retain 2x -> BEHAVIORAL (1st occurrence)
  Row 9: Competitor X's city-launch playbook mirrors our own from 2023 -> STRUCTURAL (1st occurrence)

COUNT BY TYPE
  Temporal: 4 occurrences -> test-ready
  Language: 2 occurrences -> not yet
  Behavioral: 1 occurrence -> not yet
  Structural: 1 occurrence -> not yet

Healthy

Only the week-3 CTR dip (4 occurrences, temporal) moves to a test brief this sprint; everything else stays in the journal.

Unhealthy

Building a retention campaign around the 'reorder within 48 hrs' behavioral signal off a single occurrence.

What this means

Pattern type tells you what kind of signal you're looking at; occurrence count tells you whether you're allowed to act on it yet.

So what do I do about it?

Symptom	Action	Effort
A single strong data point gets treated as a proven pattern	Require 3+ independent occurrences of the same pattern type before it earns a test budget	5 min
The pattern journal has entries but no type tags	Add a Pattern Type column and re-tag the backlog before the next monthly review	30 min

YouYou can do this yourself, no engineering access required.

Final deliverable

A tagged signal log with occurrence counts per pattern type and a one-line recommendation on which single pattern is safe to test this sprint.

See a reference example

Sample output

Squarespace, Q2 signal log (excerpt)

TEMPORAL (3 occurrences) -> test-ready
  Free-trial signups dip every week the in-app tour is skipped

STRUCTURAL (1 occurrence) -> not yet
  Competitor's template-marketplace launch mirrors Squarespace's 2019 playbook

Success criteria

You're done when you can:

Correctly tags all 10 signals by pattern type
Correctly flags which type has 3+ occurrences and which does not
Recommends only the test-ready pattern for action