Start CRO Without a Program

Leave with a leak, not a backlog

If you don't have a CRO team, don't build a fake one. Don't buy a testing app so you feel busy. Take one hour. Leave with a named leak, one fix, an owner, and a day-7 check.

The companion Article defines operator CRO. This Guide is the runbook: clicks, worksheet, fix menus, and handoff language I use when a brand has ads and a store but no conversion program.

Expected effect. You can run this alone and hand a shippable brief to whoever edits the theme.

Scope lock. Judgment and order only. No proprietary audit scores, ICE matrices, or full playbook dumps.

Before you start

Plain-English glossary

Copy-paste worksheet (use this)

Hit Copy, then paste into Notes / Docs / Notion. Fill it during the hour. That's the deliverable.

CRO FIRST HOUR - {store name} - {date}

Primary money question:
[ ] Purchases  [ ] Conversion value  [ ] Leads
Exact: ________________________________

Journey (one path only):
Entry: ________________________________
Next:  ________________________________
Buy:   ________________________________

Window: last 28 days (or _______)

FUNNEL (fill what Shopify shows you)
Sessions / visitors on entry: ______
Product views (or PDP views): ______
Add to carts: ______
Reached checkout: ______
Purchases: ______

Biggest drop (circle): entry→PDP | PDP→ATC | ATC→checkout | checkout→purchase
Drop in plain English: ________________________________

Leak type (circle one):
Friction | Anxiety | Mismatch | Choice overload | Latency

One-sentence diagnosis:
Shoppers leak at ________ because ________.

Fix (one only):
________________________________

Owner: ________  Ship by: ________
Primary metric: ________  Guardrail: ________
Mode: [ ] Ship & watch  [ ] Formal A/B  [ ] Wait
Day-7 review date: ________
What "good" looks like: ________

Shopify click map (menus move; job doesn't)

Judgment, not a score. You're hunting the steepest cliff on one journey. Don't average the whole store into one vanity conversion rate and call it done.

Leak type → fix menu (pick one)

Free and complete for hour one. Pick a single fix from the matching row.

Mobile walk checklist

• Open the entry URL on your phone (or the device mix that dominates traffic).
• Don't use staff discounts or logged-in wholesale pricing.
• Time-to-first useful content: can you tell what this is in 3 seconds?
• Can you filter / find a product without guessing?
• On PDP: price, shipping, returns, and proof visible without a treasure hunt?
• ATC → cart → checkout: any surprise fee or dead end?
• Write the first moment you felt friction, anxiety, mismatch, overload, or lag.

Step 1 · Lock primary + journey (5 min)

What. Two lines that keep the hour honest.

Why. Without them, you'll polish the homepage forever.

Who. Founder or marketing lead.

• Paste the worksheet into Notes / Docs / Notion.
• Check one primary money question.
• Write one journey with Entry / Next / Buy.
• Prefer paid landing or highest-volume collection/PDP path.
• Homepage only if that's literally where the traffic lands.

Done when. Worksheet has Primary + Journey filled. Nothing else yet.

Get help when. If leadership argues about the KPI, stop. Align money before Analytics.

Step 2 · Pull funnel ratios (15 min)

What. Find the steepest step-to-step drop on that journey.

Why. The drop tells you where. Tactics lists don't.

Who. Anyone with Analytics access.

• Use the Shopify click map above (Reports search is fine if menus moved).
• Fill sessions, product views, ATC, checkout, purchases for 28 days.
• Compute rough step rates (or eyeball which cliff is tallest).
• Circle biggest drop on the worksheet.
• If purchase count is nonsense vs Orders, stop. Measurement first.

Done when. Biggest drop named in plain English on the worksheet.

Get help when. Ads / GA4 / Shopify can't agree on orders. Use GA4 on Shopify: Why Your Numbers Never Match and the Shopify GA4 Tracking Sanity Check.

Step 3 · Mobile shopper walk (10 min)

What. Feel the leak. Assign a leak type.

Why. Ratios show where. The walk shows why.

Who. Same owner. Phone required.

• Run the mobile walk checklist above, start to buy.
• Optional: 3–5 session recordings on that URL if you already have them.
• Circle one leak type (two max if they're the same root, e.g. overload + friction).
• Write: Shoppers leak at ____ because ____.

Done when. Diagnosis sentence is on the worksheet.

Get help when. Checkout itself is broken (errors, payment fails). Involve ops before CRO polish.

Step 4 · Pick one fix from the menu (10 min)

What. Smallest change that attacks the named leak.

Why. Twenty ideas recreate a fake program.

Who. Owner + whoever edits theme / merch / copy this week.

• Open the leak → fix menu. Pick one row item, not five.
• Write owner + ship-by date this week.
• Write primary metric + one guardrail (usually AOV or refund rate).
• If it needs a redesign or new app stack, park it. Out of scope for hour one.

Done when. One fix, owner, metric, guardrail on the worksheet.

Get help when. Use the handoff brief below if you aren't the person who can edit the theme.

Step 5 · Choose mode + day-7 review (5–10 min)

What. Ship-and-watch, formal A/B, or wait.

Why. Fake science calendars waste weeks.

Who. Owner decides.

• Default to ship-and-watch for obvious reversible fixes.
• Formal A/B only with two honest options + enough traffic to finish.
• Wait if tracking is broken, journey is broken, or diagnosis is still mushy.
• Put day-7 review on the calendar with “good looks like ___”.
• Send handoff brief if someone else ships.

Done when. Mode + review date filled. Calendar invite exists.

Get help when. If nobody can ship in 7 days, the bottleneck is ownership, not more research.

Filled example (steal the shape)

Copyable handoff brief

Send this when you're not the person who can edit the theme.

Subject: CRO first-hour fix - ship this week

Journey: {entry → … → buy}
Diagnosis: Shoppers leak at {step} because {reason}.
Leak type: {friction|anxiety|mismatch|overload|latency}

Please ship ONE change:
{exact change in plain English}

Out of scope this week:
{what not to redesign}

Primary metric: {metric}
Guardrail: {metric must not collapse}
Mode: ship and watch (no A/B required)
Review date: {day+7}
Proof I'll check: Shopify {report} for the journey URLs

Owner accepting done: {name}
Implementer: {name}

Day-7 verification

• Same window length as baseline (usually 7 days post-ship vs prior 7), same journey URLs.
• Primary moved the right way? By how much (directional is enough at low traffic)?
• Guardrail held?
• Any new complaints (CX, returns, “where's the filter”)?
• Keep, iterate once, or roll back. Write one sentence.
• Only then consider a second leak. Don't open a 20-item backlog.

Common mistakes