# Google Ads Strategist's Context File for Claude

---

## Setup Instructions

**What this file is:** A knowledge file that loads into Claude a strategic framework for Google Ads built by Ryan Baker, a Google Ads strategist and trainer with 10+ years of experience. When you add this file to a Claude Project, Claude will draw on these principles when you ask Google Ads questions, instead of giving you generic platform advice.

**How to use it:**

1. In Claude.ai, create a new Project (or open an existing one).
2. Upload this file as a Project knowledge file.
3. Fill in Part 2 with your own account context. You can paste your answers directly in this file before uploading, or add a second document with your account details.
4. Start a conversation. You don't need to summarize the file, Claude will read it automatically when you ask questions.

**What to expect:** Claude will give you advice grounded in the framework in Part 1, applied to your specific account from Part 2. The more context you fill in, the more specific and useful the answers will be.

**Memory:** Part 3 tells Claude how to maintain a memory file across conversations so nothing gets lost when the conversation compacts.

---

## Part 1: Strategic Framework

### Core Philosophy

Google Ads is a direct response tool. Every dollar should be traceable to a conversion. The goal is revenue, not attribution perfection. Attribution is never perfect, and every platform takes more credit than it deserves. Don't let measurement debates slow down the work.

Start at the buying market (people actively searching for a solution) before moving up the funnel. Never build top-down. Earn your way into upper funnel by proving what works at the bottom first.

People drive performance more than algorithms do. Spikes and dips are often caused by external factors: seasonality, world events, competitor changes, economic shifts. Don't blame yourself for what you didn't cause, and don't take credit for what you didn't do. The "silo mentality" (the belief that account performance is entirely within the manager's control) is one of the most dangerous habits an ad manager can develop.

Reactive measures are not strategy. The goal is a framework that kicks in automatically when you look at any business, any account, any situation. Firefighting without a framework means the same fires come back.

Expect platform changes. Google Ads changes constantly. The underlying principles (market stages, intent, human psychology) do not. Build your practice on principles, not platform mechanics. Mechanics become obsolete. Principles compound.

---

### The Market Journey Path Framework

This is the single most important concept for structuring a Google Ads account.

**The five stages:**

1. **Contented:** The person doesn't know they have a problem. They're unaware of the pain you solve.
2. **Suffering:** They're aware of the problem but aren't actively looking for a solution yet. The perceived cost of solving it is higher than the perceived pain.
3. **Waiting:** They know solutions exist. They haven't moved because the perceived cost (money, time, hassle, risk) still outweighs the perceived pain.
4. **Buying:** They're ready to take action. They're searching, comparing, and evaluating options.
5. **Nurture:** They've already purchased. The goal is retention, upsell, and referral.

**Why this matters for Google Ads:** Most accounts only reach people in the Buying stage, which is roughly 3% of the total addressable market at any given moment. The other 97% are in earlier stages, and a typical Google Ads account does almost nothing to reach them.

**Campaign type by stage:**

| Stage | Best Campaign Types |
|---|---|
| Contented | Video (YouTube) |
| Suffering | Video, Display, Demand Gen |
| Waiting | Display, Demand Gen, Video (objection-handling), retargeting |
| Buying | Search |
| Nurture | Display retargeting, Demand Gen, Email (outside Google) |
| All stages (harder to control) | Performance Max |

**Budget allocation:**
- New account or limited budget: approximately 80% Search (buying market), approximately 20% non-search.
- At scale, the research-backed split is 60% brand building / 40% sales activation.
- If Search volume is limited in your niche, shift budget toward Meta and YouTube, not toward Google formats you've already maxed out.

**The psychology behind stage-based advertising:**

*Reticular Activation:* Awareness campaigns prime people so your brand is already familiar when they enter the Buying stage. They recognize you, which lowers resistance and raises trust.

*Mere Exposure Effect:* Repeated exposure increases preference, even without conscious attention. Showing up in earlier stages means you've already started earning trust before the sales conversation begins.

**Ad copy by stage:**
- Contented: Introduce the problem. Don't pitch a solution to someone who doesn't know they have a problem.
- Suffering: Educate on the problem and the category of solution. Build awareness that relief exists.
- Waiting: Handle objections directly. Address cost, friction, risk, and the "will this actually work for me?" question.
- Buying: Be the obvious, credible, trusted answer. Differentiate from competitors.
- Nurture: Reinforce the purchase decision. Deliver value. Build loyalty.

---

### How Google's Systems Actually Work

These are the mechanics most managers don't understand, and the gaps between what managers think is happening and what's actually happening cause most of the expensive mistakes.

**Keywords vs. search terms:** Keywords are what you target. Search terms are what users actually type. Google maps one to the other, and the mapping is loose, especially with broad and phrase match. Assume your keywords are triggering searches you didn't intend.

**Match type reality check:**
- "Exact match" is not exact. It allows close variants Google deems equivalent.
- "Phrase match" allows significant variation beyond the expected.
- "Broad match" is extremely broad. It can serve your ad for searches with very loose semantic relationships to your keyword.
- The labels imply precision the system doesn't deliver.

**The hidden search term problem:** Google withholds up to 60% of search term data, citing user privacy. Budget waste from search terms you can't see is structural, not just a management failure. You cannot fully solve it. You can minimize it through aggressive negative keyword management and smart campaign structure.

**Default settings serve Google's interests, not yours:**
- Automated assets: on by default, can generate and serve ad copy you never wrote
- PMax text customization: on by default
- AI Max for Search: opt-in, but pushes toward expanded targeting
- Auto-applied recommendations: on by default, can make changes to your account without explicit approval

Audit these settings in every new account and in every account you inherit.

**The three-layer stack:** When diagnosing any outcome (high CPA, low CTR, bad traffic), trace through all three layers:
1. Keyword: what you targeted
2. Search term: what the user actually typed
3. Ad copy: what was assembled and served (which Google's automation may have altered)

Do not skip layers. A bad outcome at the conversion stage often has its root cause at the search term or ad copy layer.

---

### Account Structure

Structure is strategy made visible. A well-structured account makes the right budget decisions obvious and makes diagnosis fast.

**Four-level hierarchy:** Account → Campaign → Ad Group → Keywords and Ads

**Primary organizing principle:** Segment by market journey stage first. Each stage gets its own campaign (or campaigns). Build out stages in budget priority order. Don't build what you can't fund.

**Separate campaigns for budget control.** If a few high-volume keywords consume all the budget, other valuable keywords become invisible. Break them into their own campaign with its own daily budget.

**Create a new ad group when:** keyword intent doesn't align tightly enough with existing ad groups. The standard is that the same ad copy should be highly relevant to every keyword in the ad group. When that breaks down, split the group.

**Always use "create a campaign without guidance"** instead of Google's wizard. More control, avoids defaults that aren't in the advertiser's interest.

**Always add "Impression Share Lost to Budget" as a column at campaign level.** This is the single most important diagnostic for understanding where budget constraints are limiting competition. If you're losing significant impression share to budget in a high-performing campaign, that's the first place to increase spend.

---

### Campaign Types

**Search:** Primary tool for the buying market. High intent. Start here. Once high-intent keyword impression share is strong, Search can also reach the suffering market through informational and symptom-based queries.

**Display:** Useful for cheap traffic, awareness, and retargeting. Strong for suffering, waiting, and retargeting stages. Always use placement exclusions (remove app, games, and junk placements from day one).

**Video (YouTube):** Best for suffering and contented stages. The hook must land in the first 5 seconds, before the skip option appears. Don't start with your brand name. "Ugly ads" (white background, black text, direct-to-camera or screen recording) are a fast, cheap way to validate messages before investing in production.

**Performance Max (PMax):** Spans Display, Video, Search, Gmail, and Shopping. More reach, less control. Run alongside other campaigns, not instead of them.

Critical PMax rule: Create a negative keyword list containing all target keywords from your dedicated Search campaigns, and apply it to PMax. This forces PMax to find incremental traffic instead of cannibalizing your existing Search campaigns.

If PMax performance degrades after adding negatives, hard reset: copy the campaign, pause the old one, and add all Search campaign target keywords as exact match negatives in the new PMax campaign.

**DSA (Dynamic Search Ads):** Google crawls your website and matches pages to searches you haven't explicitly targeted. Use as a keyword farming mechanism, not as a primary campaign. Apply the same negative keyword list used for PMax to prevent cannibalization of your Search campaigns.

**AI Max for Search:** Skeptical position. Google claims more conversions. In practice, most advertisers haven't found it consistently effective. It behaves similarly to adding a DSA layer with fully written ads and expanded targeting. Standard Search with strong negative keyword management outperforms it in most cases. Test cautiously, with conversion tracking in place before evaluating.

**Demand Gen:** Google's newest format. The only Google format that supports lookalike audiences. Appears on Discover and Gmail. Useful for retargeting and suffering/waiting markets. Still being evaluated by the industry. Worth testing if budget allows and you've exhausted Search volume.

**Shopping:** Google uses product feed data, no keyword targeting. E-commerce only. Not covered in depth here.

---

### Keywords and Match Types

**Broad match:** Designed to work with smart bidding and machine learning. Can outperform when conversion data is strong and negative keyword management is rigorous. Drives many conversions at lower CPCs in mature accounts. But: broad match with low conversion volume and high CPCs is uncontrollable and expensive. Don't use it until you have conversion volume to fuel the algorithm.

**Phrase match:** Matches searches that include the meaning of the keyword. Good in large accounts with extensive conversion data and strong negative keyword infrastructure.

**Exact match:** Highest CPC, highest intent. Use as sniper keywords: precise, high-intent targets in clean accounts where you need maximum control.

**The rule:** Broad match works best with smart bidding and high conversion volume. When conversions are low, exact and phrase match give more control and produce cleaner signals.

**The long-term shift:** As AI tools change how people search (longer, more conversational queries), broad match, DSA, and PMax become more relevant over time. The single keyword ad group with exact match only is increasingly dated. Adapt the structure as conversion volume grows.

**Keyword farming:** Sort the search terms report by conversions. High-performing search terms that aren't in your keyword list yet should be promoted to exact match keywords. DSA and PMax are effective farms for finding search terms you wouldn't have targeted directly.

**When to pause a keyword:**
- Spent 2-3x the target CPA with zero conversions
- Conversion rate under 1% and spent over $100
- Broad match keyword with low conversions and high CPC

**For high-spending keywords with some conversions:** Don't pause. Add negatives to filter out the bad traffic instead. Pausing removes the good traffic too.

---

### Negative Keywords and Algorithm Management

Negative keywords are not just a cost control tactic. They are how you train the algorithm. Every irrelevant click is a corrupted training signal. Better signals produce better performance over time.

The search terms report is the most underutilized tool in Google Ads. Review it weekly, looking at the previous 7 days.

**Match type rules for negatives:**
- Exact match negative: blocks only that precise search term. Safest option when uncertain.
- Phrase match negative: blocks any query containing the exact phrase in that order.
- Broad match negative: blocks queries containing all words in the negative keyword (in any order).
- When in doubt, use exact match negative.

**Block the smallest irrelevant unit.** Don't add a long irrelevant phrase as a negative when the irrelevant element is a single word. Find the smallest piece that's causing the problem and block that. Smaller negatives do more work.

**Apply at the right level:**
- Ad group level: when the term is irrelevant to that ad group but might be relevant elsewhere in the campaign
- Campaign level: when the term is irrelevant to the entire campaign
- Shared negative list: when the term is irrelevant across multiple campaigns

**Critical shared lists to build and maintain:**

1. **Branded terms list:** Apply to all non-brand campaigns. Keeps branded queries in your dedicated brand campaign where you can control the experience and the bid.
2. **Target keywords list (exact match):** All primary search campaign keywords, added as exact match negatives. Apply to DSA and PMax to prevent cannibalization.
3. **General negatives:** Competitor terms, irrelevant intent signals, anything that consistently wastes spend across the account.
4. **Jobs and careers list:** Keep this separate from general negatives. You may eventually run recruiting campaigns, and you don't want your general negative list blocking your own recruiting ads.

**High-leverage negative keyword work:**

Impressions with zero clicks are prime negative candidates. Filter the search terms report by impressions greater than 5-10, sort descending. High impressions with zero clicks hurts CTR, which hurts Quality Score, which raises your costs. This is often some of the highest-leverage work in an account.

**The Wastefinder diagnostic:** Build a custom report tracking keyword, city, campaign. Filter for zero conversions in the last 90 days. Sort by cost. This surfaces wasted spend at the geographic and keyword level simultaneously.

---

### Bidding Strategies

**Ad Rank = Bid + Quality Score.** Not bid alone. A better ad with lower bid can beat a worse ad with higher bid. Quality Score is made up of expected CTR, ad relevance, and landing page experience.

**Smart bidding works when the algorithm has clean signals.** Negative keyword management IS algorithm management. You're not just blocking bad traffic. You're cleaning the data the algorithm uses to decide where to spend.

**Avoid Maximize Clicks for lead gen.** It bids lower to get more volume, which drives lower-intent traffic. Potentially the worst bidding strategy for lead generation.

**Avoid Target Impression Share.** It leads to overbidding for a position on the page regardless of click quality or cost efficiency.

**Manual CPC:** Use when conversion data is too thin for smart bidding, or when cost per conversion is running 2x or more above target and you need to regain control before switching strategies. Three-step bulk bid setup:
1. Raise all bids to first page CPC
2. Raise all bids to top of page CPC
3. Decrease all bids by 20%, with a $1.00 minimum

**Maximize Conversions (no CPA target):** Early smart bidding phase. Can overspend per click if the daily budget isn't consistently hit. Good starting point before you have enough conversion data for Target CPA.

**Target CPA:** Use when you have sufficient conversion data and want cost control. Critical rule: set the target within 20% of the actual current cost per conversion. Going more than 20% above the actual CPA can destabilize the algorithm. If actual CPA is running 2x or more above target, don't inch the target upward. Switch strategies entirely.

**The conveyor belt mental model:** The algorithm is like a conveyor belt with cameras watching which customers convert. Every irrelevant click is noise in the data. Your job is to clean the signals so the algorithm gets better over time.

**Learning phases:** After significant changes (combining campaigns, switching bidding strategies, changing CPA targets), Google enters a learning phase. Performance is unreliable during this period. Do not make additional major changes while a learning phase is active. Wait for it to clear before evaluating.

**PMax and CPA target risk:** If PMax shows more engagement but fewer conversions, the algorithm is shifting to upper-funnel channels because it can't find cheap conversions at your CPA target. Fix: remove the CPA target from the PMax campaign entirely.

---

### Ad Copy

**Ad Strength is a trap.** It scores well when you follow Google's format recommendations. It does not predict conversion performance. Chasing "Excellent" Ad Strength by adding generic headlines can dilute relevance and hurt actual performance. Measure what matters: conversion rate, cost per conversion, CTR.

**Responsive Search Ads (RSAs):** Google selects combinations of your headlines and descriptions for each auction. You can provide up to 15 headlines and 4 descriptions. Use pinning sparingly, only for elements that are truly non-negotiable (brand name, primary CTA, a key differentiator that must appear).

**Testing:** Run 2-3 RSAs with meaningfully different angles in the same ad group. Let conversion data determine the winner. Don't test minor word variations. Test different strategic angles (pain-focused vs. outcome-focused, feature-focused vs. trust-focused).

**Audit automated assets in every account:**
- Account-level automated assets (on by default)
- PMax text customization (on by default)
- Auto-applied recommendations (on by default)

These can generate and serve ad copy the advertiser never wrote or approved. Find them, review them, and disable what isn't serving performance.

---

### Conversion Tracking

Without accurate conversion tracking, nothing else in this framework works. Set this up before optimizing anything else.

Don't mark everything as a primary conversion action. It confuses the algorithm. Keep primary conversions tied to actual business outcomes (form submissions, phone calls, purchases). Use secondary conversion actions for micro-conversions (page views, scroll depth, time on site).

If CTR is healthy but cost per conversion is high, the landing page is usually the bottleneck, not the campaigns. Don't keep adjusting bids and keywords when the real problem is what happens after the click.

---

### Optimization Cadence

**Weekly:**
- Search terms report (last 7 days, add negatives)
- Impression share lost to budget by campaign
- Cost per conversion by campaign
- Paused keywords check

**Monthly:**
- Cost per conversion by keyword, ad group, city, device, and demographic
- Bid adjustment audit
- Landing page performance review
- Ad copy performance review
- Budget vs. actual spend review

---

### Time Windows for Decisions

- 7 days or less: do NOT make major structural or bidding changes
- Bid adjustment decisions: 30 to 90 days of data minimum
- Keyword pruning: 90 to 180 days, require 2-3x target CPA with zero conversions
- Negating borderline search terms: 180 days, require 3-5x target CPA with minimal conversions
- Strategy evaluation (is this working?): at least 30 days before concluding something isn't working

---

### Troubleshooting Sequence When Performance Drops

Work through these in order before making changes:

1. Is a learning phase active?
2. Is there an external factor (seasonality, a news event, market shift)?
3. Has impression share changed (competition increase or competitor entering the market)?
4. Has the search term mix shifted (more irrelevant traffic coming through)?
5. Has the algorithm shifted channel mix (PMax moving budget toward Display or YouTube)?
6. Are bids competitive for the traffic you actually want?

Diagnose before you change. Most expensive mistakes come from making changes without understanding the root cause.

---

## Part 2: Your Account Context

Fill in each section below. The more specific your answers, the more useful Claude's advice will be. Leave any section blank if it doesn't apply.

---

### Business Overview

**Industry:**
[What industry are you in? Be specific, e.g., "commercial HVAC services in the Midwest" not just "HVAC"]

**Products or services:**
[What do you actually sell? What's the offer?]

**Target customer:**
[Who is the ideal buyer? Job title, company type, situation, or demographic]

**Geography:**
[Where are you running ads? National, regional, specific cities?]

**Revenue model:**
[How do you make money? One-time sale, subscription, service contract, high-ticket, e-commerce?]

---

### Goals and Targets

**Primary conversion goal:**
[What action counts as a conversion? Phone call, form fill, purchase, consultation booking?]

**Target cost per conversion:**
[What CPA are you trying to hit?]

**Monthly budget:**
[Total monthly ad spend budget]

**Current growth stage:**
[New account, growth phase, mature account, account in trouble?]

---

### Current Campaigns

**Campaign list:**
[List each campaign, its type (Search, Display, PMax, etc.), and its current bidding strategy]

**Known issues:**
[What are you currently concerned about or trying to fix?]

---

### Performance Baselines

**Monthly spend:**

**Average CPC:**

**Conversion rate:**

**Cost per conversion:**

**Impression share (Search):**

**Impression share lost to budget:**

**Impression share lost to rank:**

---

### Other Channels

**What else is running:**
[Are you running Meta, email, SEO, organic content, or anything else? How does Google Ads fit the overall strategy?]

**Attribution setup:**
[How are you attributing credit across channels? Last-click, data-driven, other?]

---

### Account History

**How long has the account been running:**

**Major recent changes:**
[Anything significant in the last 60-90 days: campaign restructures, bidding strategy changes, budget changes, landing page changes]

**What has worked:**

**What hasn't worked:**

---

### Open Questions and Concerns

[Use this section to flag anything you're uncertain about, anything you want Claude to watch for, or questions you want answered]

---

## Part 3: Memory Update Instructions

*The following section is addressed to Claude directly.*

---

**Instructions for Claude:**

At the end of any conversation that includes meaningful Google Ads decisions, recommendations, or new account information, you should offer to update a memory file that persists across conversations.

**Always ask the user at the end of any substantive session:**
"Should I update your Google Ads memory file with what we covered today?"

Do not update the file without asking first. The user may not want every conversation logged.

**If the user says yes, update or create the following file:**
`google-ads-memory.md` in the same directory as this context file (or wherever the user specifies).

**If the file doesn't exist yet, create it with an initial entry.**

**Format for each entry:**

```
## [YYYY-MM-DD]

**What we covered:**
[Brief summary of the conversation topic]

**Decisions made:**
[Any strategic decisions, campaign changes, structural decisions]

**Changes recommended or implemented:**
[Specific actions taken or planned]

**New things learned about the account:**
[Any new context about the business, campaigns, or performance that wasn't in the context file before]

**Performance observations:**
[Any notable metrics, trends, or patterns discussed]

**Open questions to revisit:**
[Anything unresolved, anything to check on, any hypotheses to test]
```

**Append each new entry. Never overwrite previous entries.** The memory file is a chronological log. Older entries stay intact.

**If the account context in Part 2 becomes outdated based on what was discussed**, note the discrepancy in the memory entry and recommend the user update Part 2 as well.

The memory file exists because Claude's context window has limits. When conversations are long or time passes between sessions, details get lost. This file is the mechanism for continuity. Keep it current.

---

*Context file built on the Google Ads strategic framework of Ryan Baker, Kingly Consulting.*