Google Ads SOP | MM Internal

1

Setup

Step 1 — Keyword research

Use Google Ads Keyword Planner (Tools & Settings → Planning → Keyword Planner) or the Keyword Research GPT ↗. Analyse by monthly searches, competition, and suggested bids. Export and organise into themed groups.

Step 2 — Match types

Match type	Format	When to use
Phrase match	`"running shoes"`	Default starting point — shows for phrases containing the keyword
Exact match	`[running shoes]`	High-intent terms where you want tight control
Broad match	`running shoes`	Limit to specific use cases or audience discovery only

🚨

Never add broad match keywords. The MM Ads Autopilot enforces this at code level — it will never propose or add broad match keywords. Always use phrase or exact match.

Step 3 — Campaign naming

Follow this convention: MM - [Objective] - [Service] - [Geo] - [Network]

Example: MM - LeadGen - Logbook Servicing - NSW - Search

Step 4 — Write ad copy

Use the client's dedicated GPT if one exists (see GPT Reference ↗). Otherwise use:

Search ads: MM GAds Copy Creator ↗
Performance Max: MM PMax Copy Creator ↗

Step 5 — Extensions

Navigate to Ads & Extensions → Extensions. Add: Sitelink, Callout, Structured Snippet. Keep copy short and focused on selling points or CTAs.

Step 6 — Buddy review before launch

Complete setup and pause the campaigns. Assign task to "Needs Review" in ClickUp. Do not launch until reviewed.

Step 7 — Budget estimation

Use Keyword Planner to view estimated costs and clicks for target keywords. Multiply average CPC by desired clicks to estimate daily/monthly budget. Add 10–20% buffer for initial testing.

Step 8 — Turn off automatic broad match

Google Ads defaults new campaigns to broad match. Add quotation marks or brackets to all keywords immediately after creation. Check match type settings under Keywords → Match Types regularly.

2

Optimisation

ℹ️

Autopilot handles weekly optimisation. The MM Ads Autopilot runs every Monday and proposes keyword pauses, negative keywords, new keywords, ad copy rewrites, and more. Review and approve in Slack. Manual optimisation below is for tasks outside the Autopilot's scope or for on-demand analysis.

Find and replace winning ad combos

Go to Ads → check CTR, conversions, ROI. Identify top-performing headlines or descriptions. Copy these to lower-performing ad groups. Video walkthrough ↗

Exclude age and device

Go to campaign Demographics (for age) or Devices section. Exclude low-performing segments. Adjust bids down (-100%) to exclude poor-performing devices. Video ↗

Account-level negative keywords

Navigate to Tools & Settings → Shared Library → Negative Keyword Lists. Add common irrelevant terms that apply across multiple campaigns. Assign to all relevant campaigns. Video ↗

Create experiments

Click Drafts & Experiments → + New Experiment. Select original campaign, set your split (e.g. 50% traffic) and duration. Use for testing bidding strategies, creatives, or targeting changes. Video ↗

Exclude URLs from campaigns

For Display or PMax: navigate to Content → Exclusions. Add specific URLs or placements to exclude. Video ↗

3

Improve lead quality

If leads are coming through but quality is poor (high non-quotable rate in WhatConverts), tighten spend using manual CPC bidding:

1

Go to campaign Settings → Bidding → Change bid strategy → select Manual CPC

2

Review keywords tab — identify keywords producing low-quality leads or high-cost clicks

3

Lower bids for poor-quality keywords. Increase slightly for high-quality converters.

4

Check Placements, Devices, and Locations reports — adjust bids down for poor performers

5

Add irrelevant search terms as negative keywords regularly

6

Review and adjust bids weekly, especially in the first month of manual bidding

⚠️

Always check WhatConverts lead quality before blaming campaign performance. High non-quotable rate may mean the client isn't answering calls or is marking real leads as spam. Confirm with the client before making significant bid changes.

4

Autopilot workflow

Once a client's account has 14+ days of data, MM Ads Autopilot begins monitoring and proposing optimisations every Monday. This replaces the bulk of manual weekly optimisation.

Your Monday routine

Open #mm-ads-autopilot in Slack
Read the portfolio digest — check which clients are red or yellow
Review proposed actions for each client — Approve, Reject, or Review individually
If rejecting, explain why — this trains the system for future weeks

On-demand analysis

Ask the Autopilot bot in Slack at any time:

"How is [client] performing this week?"
"What keywords are wasting budget for [client]?"
"Run analysis for [client]" — triggers a full pipeline run (2–3 minutes)

What the Autopilot proposes vs what needs human decision

Autopilot handles	Always escalate to Josh
Keyword pauses (high CPA / zero conversions)	Budget increases or decreases
Negative keyword additions	Campaign structure changes
New keyword suggestions from GSC data	Bidding strategy changes
Ad copy rewrites with preview	Ad account suspended
Paid/organic overlap flags	Policy violations
Anomaly alerts (CPA spike, zero conversions)	Campaign pauses/enables

For full Autopilot documentation see the Autopilot User Guide ↗.