---
name: structure-a-google-ads-account
title: "Structure a Google Ads Account"
description: "ChatGPT prompt for Google Ads setup that builds a clean campaign and ad group structure with keywords, match types, and bidding tied to conversion economics."
cluster: paid-search
version: 1.1.0
---

# Structure a Google Ads Account

This prompt turns a blank or tangled Google Ads account into a correctly layered search program. It defines the business outcome, installs and verifies conversion tracking, groups keywords by shared ad promise, picks match types balanced for reach and control on conversion economics, and sets a bidding plan that lets the machine learn against the right goal. You paste your business inputs and get back a complete build plan you can execute setting by setting.

## When to use this

→ You are launching a brand new Google Ads account or a first search campaign and want the structure right before money starts flowing
→ You inherited an account with no conversion tracking, tangled ad groups, or broad match running blind, and you want to rebuild the foundation
→ You need a defensible plan for campaign structure, match types, negatives, and bidding that you can hand to a stakeholder or client for sign off

## The prompt

```text
You are a senior paid search operator who builds Google Ads accounts that automated
bidding can actually learn from. Your operating model: give automation strong goals
and guardrails, then let performance data, not tidy structure, decide what changes.
You optimize for business outcomes and conversion economics, never for click through
rate, keyword neatness, or surface metrics.

CONTEXT
Business and offer: {{BUSINESS_AND_OFFER}}
Primary business outcome to buy (purchase, qualified lead, call, booking): {{PRIMARY_OUTCOME}}
Website and key conversion page or event: {{SITE_AND_CONVERSION_EVENT}}
Average order value or deal value and gross margin: {{VALUE_AND_MARGIN}}
Lead to customer close rate if this is lead gen: {{CLOSE_RATE}}
Monthly budget and tolerance for a learning period: {{BUDGET_AND_LEARNING_TOLERANCE}}
Geography and languages: {{GEO_AND_LANGUAGES}}
Seed keyword themes or product categories: {{SEED_THEMES}}
Known terms to exclude and any brand terms: {{EXCLUSIONS_AND_BRAND}}
Analytics setup and whether it is linked: {{ANALYTICS_STATUS}}
Weekly hours available for query review and negatives: {{MAINTENANCE_HOURS}}

If any of PRIMARY_OUTCOME, VALUE_AND_MARGIN, SITE_AND_CONVERSION_EVENT, or
BUDGET_AND_LEARNING_TOLERANCE is missing or vague, ask me 2 to 4 clarifying
questions, then STOP and wait for my answers before building anything. Do not
guess economics. Proceed only once those inputs are solid.

METHOD
1. Define the outcome. State the single business action bidding must optimize toward.
   Mark it the primary conversion. List softer actions (soft form, download, view) as
   secondary conversions kept for observation, excluded from bidding.

2. Design conversion measurement before any spend.
   a. Choose a method: native Google Ads website tag when you want configurable
      attribution windows and clean bidding feedback; analytics import when reporting
      alignment with analytics is the priority.
   b. Specify the exact firing event or page, and deliberate click through and view
      through windows.
   c. Add a verification step: confirm the tag fires and records before launch. A
      conversion action created but never validated records nothing.
   d. If both methods track the same action, name them explicitly and pick ONE as the
      bidding source so the outcome is not double counted.

3. Set allowable acquisition economics.
   allowable CPA and target ROAS must account for gross margin, close rate, repeat
   purchase, refunds, and cash payback, not first order revenue alone.
   Compute: CPA = spend / conversions; ROAS = revenue / spend;
   customer acquisition cost = lead CPA / lead to customer rate.
   Never use lifetime value to justify an unaffordable payback period.

4. Group keywords by shared ad promise (ad copy first).
   a. List candidate keywords, ignore match type for now.
   b. For each, write the ad promise that answers its intent.
   c. Group keywords only when the SAME ad copy can credibly serve all of them.
   d. Split any theme needing a different promise into its own ad group.
   e. Name the landing page that pays off each ad group's promise, drawing on
      {{SITE_AND_CONVERSION_EVENT}}; flag any promise with no page that delivers it.
   Large loose ad groups force generic copy and break the query to ad to page path.

5. Choose match types on conversion economics, not relevance feel.
   Exact: tightest semantic match, lower volume, higher precision.
   Phrase: moderate, good for discovering customer language.
   Broad: widest discovery, most early waste, requires verified tracking plus a
   negative keyword process and budget tolerance.
   Decision rule: required conversion rate lift = expensive CPC / cheaper CPC. A costlier
   match earns its place only if its conversion rate or downstream value covers that ratio.
   Recommend a match type structure (separate by campaign, separate by ad group, or
   combined) sized to the maintenance hours available. Start simple; separate only when
   economics materially differ and automation allocates poorly.

6. Set campaign guardrails. Search only (disable Display expansion for a clean test),
   correct networks, geography, languages, budget, schedule, the verified primary
   conversion, and automation permissions. Disable settings that silently force broad
   match or auto create assets unless you are deliberately testing them.

7. Build negatives and prevent contamination.
   Account level negatives for categories unwanted everywhere. Campaign level for brand
   terms in non brand campaigns. Add brand terms as negatives where you measure non brand
   acquisition so easy brand demand does not inflate prospecting.

8. Choose a bidding sequence. New campaign with tolerance for learning: start Maximize
   Conversions to discover live CPC, conversion rate, and CPA. Move to Target CPA or
   Target ROAS once volume and value data are stable. Keep branded search on Manual CPC.
   For sparse volume or nuanced B2B, consider Manual CPC and manage directly. Never apply
   a strict target before the campaign has data; it starves delivery and learning.

9. Add audience segments in Observation mode (reach unchanged, performance segmented) to
   learn before excluding or targeting.

OUTPUT FORMAT
1. Clarifying questions (only if inputs are missing).
2. Conversion plan: primary and secondary actions, measurement method, firing event,
   attribution windows, verification checklist.
3. Allowable CPA / target ROAS with the math shown.
4. Account structure. For every ad group, fill this exact template:
   Campaign: <name> | Ad group: <name>
   Ad promise: <the one sentence promise every ad in this group must keep>
   Keywords: <keyword, match type, one line of conversion economics per choice>
   Landing page: <page that pays off the promise, or FLAG: no page delivers this yet>
5. Negative keyword lists by scope (account, campaign, ad group).
6. Bidding plan with the graduation trigger from Maximize Conversions to a target.
7. Pre launch settings checklist and first two weeks optimization cadence (weekly query
   review, negatives, promote or reroute winners; twice weekly pacing; monthly audience
   and Auction Insights review).
8. Verify in the live interface: a short list of every platform setting name, default
   behavior, feature, and attribution option the plan relies on, so I can confirm the
   current names and behavior in the account before launch; these change over time.

SELF CHECK before finishing
→ Is tracking verified end to end, deduplicated, and is the primary conversion a real
  business outcome, not a soft action?
→ Does every ad group hold keywords that one ad promise can serve seamlessly, and does
  each ad group name a landing page that pays off that promise?
→ Is each match type choice justified by conversion economics, not click through rate?
→ Are brand and non brand separated so brand demand cannot inflate prospecting?
→ Does the bidding plan avoid a strict target before the campaign has data?
→ Flag any number you cannot tie to the inputs I gave, and state assumptions plainly.
```

## Prompt versions

The standard prompt above works on any model. Use these variants when you want a different tradeoff.

### Frontier model version

Built for the most capable models (Claude Opus and beyond). States the goal, constraints, and quality bar up front, then trusts the model to choose its path.

```text
You are a senior paid search operator. Build me a complete Google Ads account structure
that automated bidding can learn from, optimized for business outcomes and conversion
economics, never for click through rate or tidy keyword grouping.

DELIVERABLE
Return a build plan I can execute setting by setting: conversion plan, allowable
economics with the math shown, full campaign and ad group structure, scoped negative
lists, a bidding sequence, and a pre launch settings checklist. Lead with the single
business outcome bidding will optimize toward, then everything that supports it.

CONTEXT
Business and offer: {{BUSINESS_AND_OFFER}}
Primary outcome to buy: {{PRIMARY_OUTCOME}}
Site and key conversion page or event: {{SITE_AND_CONVERSION_EVENT}}
Value and gross margin: {{VALUE_AND_MARGIN}}
Lead to customer close rate if lead gen: {{CLOSE_RATE}}
Budget and learning tolerance: {{BUDGET_AND_LEARNING_TOLERANCE}}
Geography and languages: {{GEO_AND_LANGUAGES}}
Seed keyword themes: {{SEED_THEMES}}
Terms to exclude and brand terms: {{EXCLUSIONS_AND_BRAND}}
Analytics status: {{ANALYTICS_STATUS}}
Weekly hours for query review: {{MAINTENANCE_HOURS}}

PRINCIPLES (non negotiable)
→ Specify conversion measurement before any spend: one firing event, deliberate
  attribution windows, an explicit verify before launch step, and exactly one source
  feeding bidding so outcomes are never double counted.
→ Derive allowable CPA and target ROAS from margin, close rate, and cash payback, not
  first order revenue; show the arithmetic. Never justify an unaffordable payback with
  lifetime value.
→ Group keywords by shared ad promise: keywords share an ad group only when one ad copy
  can credibly serve them all, and each ad group names a landing page that pays off its
  promise.
→ Choose match types on conversion economics, not relevance feel. A costlier match earns
  its place only when its conversion rate or downstream value covers expensive CPC over
  cheaper CPC. Broad requires verified tracking, a negatives process, and budget
  tolerance. Size structural separation to the maintenance hours available; start simple.
→ Separate brand from non brand so brand demand cannot inflate prospecting. Scope
  negatives at account, campaign, and ad group levels.
→ Start bidding in discovery (Maximize Conversions) and name the exact trigger to
  graduate to Target CPA or ROAS; never apply a strict target before the campaign has data.

QUALITY BAR
Excellent output ties every number to my inputs, states assumptions plainly, and closes
with a list of the live platform setting names, defaults, and attribution options the
plan relies on so I can confirm them in the account before launch.

BOUNDARIES
Do not invent data, statistics, or economics. Do not pad with generic best practice. If
PRIMARY_OUTCOME, VALUE_AND_MARGIN, SITE_AND_CONVERSION_EVENT, or
BUDGET_AND_LEARNING_TOLERANCE is missing or vague, ask one focused question and stop
rather than guessing.
```

### Quick version

Five lines or fewer, for when speed matters more than rigor.

```text
Act as a paid search operator. Build a Google Ads structure for {{BUSINESS_AND_OFFER}}
optimizing toward {{PRIMARY_OUTCOME}}, given {{SEED_THEMES}}, {{VALUE_AND_MARGIN}}, and
{{BUDGET_AND_LEARNING_TOLERANCE}}. Give me campaigns, ad groups (each with one ad promise
and a landing page), keywords with match types, scoped negatives, and a bidding plan.
Justify every match type by conversion economics, and tie all numbers to my inputs.
```

## How to customize

→ `{{PRIMARY_OUTCOME}}`: the one action worth buying, for example "completed checkout" or "qualified demo request", not "newsletter signup"
→ `{{VALUE_AND_MARGIN}}`: for example "average order 180 dollars at 55 percent gross margin" so the allowable CPA math is real
→ `{{CLOSE_RATE}}`: for lead gen, for example "12 percent of demo requests become customers"; leave blank for ecommerce
→ `{{SEED_THEMES}}`: your product categories or the intents customers search, for example "commercial lawn care, weekly grass cutting, seasonal cleanup"
→ `{{EXCLUSIONS_AND_BRAND}}`: categories you never want (for example "free", "jobs", "DIY") plus your brand name so it can be split out
→ `{{MAINTENANCE_HOURS}}`: honest weekly time for query review; this sizes how much structural separation the plan should recommend

## What good output looks like

→ Conversion tracking is specified down to the firing event and attribution windows, with an explicit verify before launch step, and only one source feeds bidding
→ Allowable CPA or target ROAS is calculated from margin and close rate with the arithmetic shown, not asserted
→ Every ad group maps to a single ad promise and a named landing page that pays it off, and each match type choice cites conversion economics, not relevance or click through rate
→ Negatives are scoped correctly and brand terms are separated from non brand measurement
→ The bidding plan starts in discovery and names the exact trigger to graduate to a target, rather than applying a strict target on day one
→ The plan closes with a verify in the live interface list, so setting names, defaults, and attribution options get confirmed in the account instead of trusted from memory

## Related prompts

→ [Audit Google Ads for Wasted Spend](./audit-google-ads-wasted-spend.md): once the account is structured, hunt down the queries, match types, and settings quietly burning budget
→ [Defend Brand Terms From Competitors](./defend-brand-terms.md): take the brand versus non brand separation here into a full plan for protecting your own name
→ [Fix a Low Quality Score](./fix-a-low-quality-score.md): tighten the query to ad to page relevance that this structure sets up so your costs fall