Module 7: Reading numbers and acting on them

Diagnosing drop-offs

When conversions dip, diagnose in order: (1) Is traffic down or is conversion rate down? Check sessions first — a traffic drop is a visibility problem (rankings, ads paused, seasonality), while steady traffic with fewer conversions is a website problem (broken form, pricing change, new friction). (2) Which channel moved? A single-channel drop points at that channel; an all-channel drop points at the site. (3) When did it start? Match the date to changes you made — new site design, paused campaigns, holiday weeks.

For website problems, check the boring stuff first: submit your own form (does it still work?), load the site on your phone (does the CTA show?), check for expired SSL or payment errors. Half of all "marketing stopped working" mysteries are broken plumbing, not strategy failures.

Document every diagnosis in the Decisions tab, including dead ends. "Checked — form works fine; drop was seasonal (July 4th week)" prevents next year's panic when the same dip recurs.

Build a personal checklist of the usual suspects and run it top-down every time: form functionality, mobile CTA visibility, SSL/payment errors, recent site changes, ad account status, seasonality. The same five culprits cause 90% of dips. A written checklist turns a stressful mystery into a 20-minute routine — and it means anyone on your team can run the diagnosis, not just you.

Common traps: vanity, correlation, and small samples

Three traps catch everyone. Vanity metrics: traffic is up 40%! — but conversions flat. Traffic that doesn't convert is a cost, not a win. Always pair volume metrics with conversion metrics. Correlation vs. causation: sales rose the same month you started blogging — but it's also holiday season. Before crediting a channel, ask what else changed. Small samples: 3 conversions from 10 visitors is a 30% rate that means nothing; wait for 100+ conversions before declaring winners.

The antidote to all three is the same: longer timeframes, bigger samples, and always asking "compared to what?" A number without a comparison (last month, last year, the other channel) is trivia. A number with a comparison is insight.

A practical minimum: 100 conversions per variation before you trust a comparison. Below that, randomness dominates — a "winning" headline with 12 conversions might just be luck. If 100 conversions takes three months at your volume, that's fine; the decision waits. Patience with small samples is one of the highest-ROI analytics skills there is.

Knowing when to get help

This course takes you far — but know the edges. Call in expertise when: you're spending real ad money ($2,000+/mo) without clean attribution, you need server-side tracking or complex e-commerce funnels, you're facing a sudden traffic collapse you can't diagnose, or legal privacy requirements (health, finance, EU visitors) complicate your setup.

When hiring help, your new fluency is the asset: you can now brief precisely ("our GA4 shows checkout abandonment spiking on mobile since March; we need the funnel diagnosed"), evaluate proposals critically, and verify the work afterward. The most expensive analytics mistake is hiring blind — you're no longer blind.

And remember what this course was really about: not tools, but decisions. The spreadsheet, the monthly ritual, the five questions — they're all scaffolding for one habit: letting evidence, not instinct, allocate your next dollar and hour. You now have that habit. Everything else is refinement.

If you take one thing from this course, make it the Decisions tab. Six months from now it will be the most valuable document in your business — a record of what you tried, what worked, and what you learned. Most businesses repeat their mistakes annually; yours won't, because you wrote them down.