Best ChatGPT Prompts for Google Ads

Two things about Google Ads in 2026 change what AI is actually useful for.

First, the manual levers keep disappearing. AI Max for Search is out of beta and expanding into Shopping and travel formats, Dynamic Search Ads are being auto-upgraded into it ahead of a 2027 sunset, and Performance Max keeps absorbing more of the buying decision. You no longer win by adjusting bids. You win on the quality of the inputs you feed the machine — assets, conversion data, negatives, audience signals, feeds.

Second, and more practically: models cannot count characters. Ask for fifteen responsive search ad headlines and you will get several over the 30-character limit, confidently presented as compliant. That single failure wastes more time than anything else in AI-assisted PPC, and every copy prompt below forces a self-audit against the real limits.

These 30 prompts assume you paste real account data. A model has no access to your CPCs, your conversion rates or your search terms, and one asked to work without them will invent all three.

What AI Can and Cannot Do Here

Genuinely good at: generating asset variety at volume, clustering search terms into themes, drafting negatives from a term report, spotting patterns in pasted data, writing landing page copy that matches ad copy, and structuring a test so the result means something.

Bad at: counting characters, knowing your account, estimating CPCs or search volumes, and judging what Google's policy team will approve.

The rule that follows: use it for generation and analysis of data you supply, then verify everything numeric yourself. Also note that Google introduced AI-generated content label requirements in 2026 — check the current disclosure rules for your ad formats and markets before shipping generated creative at scale.

Eight Rules That Make These Prompts Work

1. State the character limits in the prompt and demand a count. Headlines are 30 characters, descriptions 90, path fields 15. Ask for a character count beside every line and reject the ones that fail.
2. Paste real data or expect fiction. Search term reports, conversion figures, CPCs. Without them the model invents numbers that look plausible.
3. Never ask for keyword volumes or difficulty. Generate and cluster candidates with AI; get every metric from Keyword Planner or a real tool.
4. Ad Strength is not a performance metric. It measures asset variety, not results. Do not optimise toward "Excellent" at the cost of message clarity.
5. Conversion tracking quality is upstream of everything. Smart Bidding optimises toward whatever you told it to value. If that signal is wrong, better ad copy cannot save the account.
6. Give it the offer, not just the product. Price, guarantee, differentiator, delivery time. Ads with a concrete offer beat ads with adjectives.
7. Ask what to exclude. Negatives, audience exclusions and brand exclusions do more for wasted spend than any headline.
8. Every claim must be substantiable. Superlatives, comparative claims and pricing statements all carry policy and legal exposure.

Account Strategy and Structure (Prompts 1–4)

1. The Account Structure Plan

Act as a senior paid search strategist. Design the account structure for this business.

BUSINESS: [WHAT YOU SELL AND TO WHOM]
PRODUCTS OR SERVICES: [THE FULL LIST, GROUPED AS YOU THINK OF THEM]
MONTHLY BUDGET: [FIGURE AND CURRENCY]
GEOGRAPHY: [MARKETS AND LANGUAGES]
CONVERSION GOAL: [WHAT COUNTS AS SUCCESS AND ITS VALUE TO YOU]
CURRENT STRUCTURE: [WHAT EXISTS NOW, OR "NEW ACCOUNT"]

Produce:
1. The campaign structure, with the reasoning for each split — by product line, by intent, by margin, or by geography
2. Which campaign types to run: Search, Performance Max, Shopping, Demand Gen, and why each earns a place at my budget
3. Ad group structure within each campaign, and how tightly to theme them given modern broad match and Smart Bidding
4. What to keep separate for budget control versus what to consolidate so Smart Bidding has enough data
5. Brand versus non-brand separation, and how to measure brand honestly

Then tell me:
- The minimum viable structure if my budget is too thin for the ideal one
- How many conversions per campaign per month I need before Smart Bidding works properly, and what to do below that threshold
- What I am likely over-segmenting

Do not estimate CPCs, volumes or expected performance. Structure and reasoning only.

2. The Campaign Type Decision

Help me decide which Google Ads campaign types to run.

BUSINESS MODEL: [ECOMMERCE / LEAD GEN / LOCAL / SAAS / MARKETPLACE]
BUDGET: [MONTHLY FIGURE]
PRODUCT FEED: [DO YOU HAVE ONE, AND ITS QUALITY]
CREATIVE ASSETS AVAILABLE: [IMAGES, VIDEO, COPY — BE HONEST ABOUT WHAT EXISTS]
CONVERSION VOLUME: [CURRENT MONTHLY CONVERSIONS]
EXISTING PERFORMANCE: [WHAT IS WORKING AND WHAT IS NOT]

For each campaign type, tell me:
1. Whether it fits my situation, with a clear yes or no
2. The minimum budget and conversion volume it realistically needs to work
3. What it will cannibalise from my existing campaigns
4. The specific risk of running it at my scale

Cover Search, Performance Max, Standard Shopping, Demand Gen, Video and App where relevant.

Then be direct about the trade-offs:
- Where Performance Max will absorb branded traffic and flatter its own reporting, and how to prevent that
- Whether my creative assets are strong enough for PMax to work, or whether it will lean on weak assets
- What I should NOT run yet, and what has to be true before I do

Prioritise by expected contribution to profit rather than by volume of clicks.

3. The Competitor Ad Analysis

Analyse the competitor ads I am seeing and find my angle.

MY OFFER: [WHAT YOU SELL, PRICE, DIFFERENTIATOR]
COMPETITOR ADS: [PASTE THE HEADLINES AND DESCRIPTIONS YOU OBSERVE ON YOUR TARGET TERMS]
MY CURRENT ADS: [PASTE]
TARGET TERMS: [THE QUERIES]

Analyse:
1. The messaging conventions everyone in this auction uses
2. Which claims are genuinely differentiating versus which are category table stakes
3. What nobody is saying that my offer could credibly say
4. Where my current ads blend into the set

Then produce:
- Three positioning angles that would stand out, ranked by how defensible they are
- For each: the proof I would need to support it
- The angle to avoid because a competitor owns it credibly and I would look derivative

Rules: base this only on the ads I pasted. Do not assert what any competitor's performance, budget or landing page contains — you cannot see those. Do not suggest claims about competitors by name, since comparative advertising carries policy and legal exposure.

4. The Budget Allocation Model

Allocate my Google Ads budget across campaigns.

TOTAL MONTHLY BUDGET: [FIGURE]
CAMPAIGNS: [LIST EACH WITH ITS CURRENT SPEND, CONVERSIONS, CONVERSION VALUE AND CPA OR ROAS]
BUSINESS PRIORITIES: [GROWTH, EFFICIENCY, OR A SPECIFIC TARGET]
CONSTRAINTS: [MINIMUM VOLUMES, SEASONAL FACTORS, INVENTORY LIMITS]

Produce:
1. A recommended allocation, with the reasoning per campaign
2. Which campaigns are budget-constrained and losing available profitable volume
3. Which are spending into diminishing returns
4. The order to shift budget in, and how fast — including why moving too much at once resets the learning phase
5. What to protect regardless of efficiency, such as brand defence

Then tell me:
- The single reallocation with the biggest expected impact
- What to measure over the next 30 days to confirm it worked
- The signal that would mean reverse it

Work only from the figures I pasted. Where my data is too thin to support a conclusion for a given campaign, say so rather than reallocating on a guess.

Keywords and Search Terms (Prompts 5–9)

5. The Keyword Ideation Pass

Generate and cluster keyword candidates for my campaigns.

BUSINESS: [WHAT YOU SELL]
PRODUCTS OR SERVICES: [SPECIFIC LIST]
CUSTOMER LANGUAGE: [HOW CUSTOMERS ACTUALLY DESCRIBE THE PROBLEM — from sales calls, support, reviews]
GEOGRAPHY: [MARKETS]
WHAT YOU DO NOT SELL: [IMPORTANT — this drives negatives]

Produce:
1. Keyword candidates grouped by intent: transactional, commercial investigation, and informational
2. Within transactional, sub-grouped by theme so they map to ad groups
3. Modifier variations — brand, location, urgency, price, comparison, problem-based
4. A separate list of terms that look relevant but signal the wrong intent for a paid click
5. An initial negative keyword list drawn from what I said I do not sell

Rules:
- Do NOT estimate search volume, competition or CPC. Generate and classify only; I will pull metrics from Keyword Planner
- Use the customer language I supplied rather than industry jargon where they differ
- Flag any term that is ambiguous across industries and would need heavy negatives

Output as a table: keyword, intent, suggested ad group, match type recommendation, notes.

6. The Search Term Mining Prompt

Analyse this search term report and tell me what to do.

SEARCH TERM REPORT: [PASTE — term, impressions, clicks, cost, conversions, conversion value]
CAMPAIGN CONTEXT: [WHAT THE CAMPAIGN SELLS AND ITS TARGET]
CURRENT NEGATIVES: [PASTE YOUR EXISTING LIST]

Produce four lists:
1. ADD AS NEGATIVE — terms wasting spend, with the match type for each negative and the level to add it at
2. ADD AS KEYWORD — converting terms worth their own keyword and ad group
3. WATCH — terms with signal but insufficient data, and the threshold at which to act
4. INTENT MISMATCH — terms getting clicks where the landing page cannot satisfy them

For the negatives, group them into themes rather than listing one-offs, and give me the pattern each theme represents so I can pre-empt it.

Then tell me:
- The proportion of spend going to terms that will never convert
- Whether the waste pattern points at a keyword problem, a match type problem, or a landing page problem
- The single negative that would save the most money

Be careful with negatives that are too broad. Flag any negative you propose that risks blocking valuable traffic, and say what it would block.

7. The Negative Keyword Architecture

Design a negative keyword structure for my account.

ACCOUNT: [CAMPAIGNS AND WHAT EACH SELLS]
KNOWN WASTE THEMES: [WHAT YOU ALREADY BLOCK]
BUSINESS EXCLUSIONS: [WHAT YOU DO NOT SELL, DO NOT SERVE, OR CANNOT FULFIL]
BRAND TERMS: [YOURS AND ANY YOU MUST AVOID]

Produce:
1. Account-level negative lists, grouped by theme — job seekers, free-seekers, DIY intent, competitors, irrelevant industries, adult or unsafe terms
2. Campaign-level negatives that separate my campaigns from each other
3. Ad group level negatives where tight theming matters
4. The match type for each negative and why

Then cover:
- The cross-campaign negative strategy that stops campaigns bidding against each other
- How to keep brand traffic out of non-brand campaigns and vice versa
- Which negatives to review monthly versus which are permanent
- The maintenance routine, and who should own it

Flag any negative that could block a legitimate long-tail variation, since over-negating is a slower but real way to lose profitable volume.

8. The Match Type Strategy

Recommend match types for my keyword set.

KEYWORDS: [PASTE, WITH CURRENT MATCH TYPES IF ANY]
BIDDING STRATEGY: [WHAT YOU USE]
CONVERSION VOLUME: [MONTHLY, PER CAMPAIGN]
NEGATIVE COVERAGE: [HOW MATURE YOUR NEGATIVE LISTS ARE]
BUDGET: [MONTHLY]

For each keyword or cluster, recommend a match type with reasoning, considering:
1. How much control I need versus how much reach
2. Whether my conversion volume supports broad match with Smart Bidding, or whether it will spend into noise
3. Whether my negative lists are mature enough for broad match yet
4. Where phrase match is the right middle ground

Then explain:
- Why broad match plus Smart Bidding behaves differently from broad match alone, and what has to be true for it to work
- The risk of going broad too early with thin conversion data
- How to test a match type change without destabilising a working campaign
- What to monitor in the search term report in the first two weeks after any change

Be conservative where my conversion volume is low. Broad match with insufficient conversion signal is one of the fastest ways to waste a budget.

9. The Query Intent Audit

Audit whether my keywords match what my landing pages actually deliver.

KEYWORDS AND THEIR LANDING PAGES: [PASTE PAIRS]
LANDING PAGE CONTENT: [SUMMARISE WHAT EACH PAGE OFFERS AND ITS PRIMARY ACTION]
PERFORMANCE: [CLICKS, CONVERSION RATE AND BOUNCE PER PAIR IF AVAILABLE]

For each pair, assess:
1. What a searcher using this term expects to find
2. Whether the landing page delivers that in the first screen
3. Where the gap is — wrong page, right page but wrong message, or right message buried
4. Whether the mismatch explains the performance I am seeing

Then produce:
- The pairs to fix first, ranked by wasted spend
- For each: change the keyword, change the page, or change the ad
- Which keywords have no suitable page at all and need one built

Do not assume a low conversion rate is always a landing page problem. Where the data suggests the keyword itself is the wrong intent to buy, say so — the cheaper fix is often to stop bidding rather than to rebuild a page.

Responsive Search Ad Copy (Prompts 10–14)

10. The RSA Asset Set

The character-count discipline is the whole point here.

Write a complete responsive search ad asset set.

PRODUCT OR SERVICE: [WHAT IT IS]
TARGET KEYWORDS: [THE AD GROUP THEME]
OFFER: [PRICE, GUARANTEE, DELIVERY, TRIAL, OR OTHER CONCRETE HOOK]
DIFFERENTIATORS: [WHAT IS ACTUALLY TRUE AND PROVABLE]
AUDIENCE: [WHO IS SEARCHING AND WHAT THEY WORRY ABOUT]
LANDING PAGE: [WHAT IT SAYS AND ITS PRIMARY ACTION]
BRAND VOICE: [SAMPLE]

Produce:
- 15 headlines, each 30 CHARACTERS OR FEWER
- 4 descriptions, each 90 CHARACTERS OR FEWER
- 2 path fields, each 15 CHARACTERS OR FEWER

MANDATORY FORMAT: print the character count in brackets after every single line, like this — Free Delivery Over 50 [21]

Then self-audit before finishing:
1. Re-count every line and list any that exceed the limit
2. Replace any that failed, and re-count the replacement
3. Confirm at the end that all assets are within limits

Coverage requirements across the 15 headlines:
- At least 3 containing the primary keyword naturally
- At least 3 leading with the offer
- At least 3 addressing an objection
- At least 2 with a call to action
- At least 2 with proof — a number, a guarantee, a credential I supplied
- Enough variety that Google can assemble different combinations without repeating the same message

Rules: every claim must come from what I supplied. No unsupported superlatives such as "best" or "number one". No comparative claims about named competitors. No exclamation marks in more than one asset. No ALL CAPS words.

11. The Headline Angle Expansion

Expand my best-performing headline into new angles to test.

CURRENT WINNER: [PASTE THE HEADLINE AND ITS PERFORMANCE]
WHAT I THINK MAKES IT WORK: [YOUR HYPOTHESIS]
PRODUCT AND OFFER: [DETAIL]
UNDERPERFORMING HEADLINES: [PASTE, SO WE DO NOT REPEAT THEM]

Produce 12 new headlines, each 30 CHARACTERS OR FEWER with the count shown in brackets, spread across these angles:
- The same psychological trigger, different wording
- The opposite framing — loss instead of gain, or vice versa
- A more specific version, with a number or detail
- A question rather than a statement
- The objection the winner implicitly answers, made explicit
- A different stage of buyer awareness

For each, state which angle it represents and what it tests that the current winner does not.

Then:
- Rank them by how different they are from the current winner, since near-duplicates teach you nothing
- Flag any that would cannibalise an existing headline's role in the asset mix
- Self-audit the character counts and correct any that exceed 30

Remember Ad Strength measures asset variety, not performance. Do not pad the set with weak variations purely to raise it.

12. The Ad Policy Pre-Check

Review this ad copy for Google Ads policy risk before I upload it.

AD COPY: [PASTE ALL HEADLINES, DESCRIPTIONS AND PATHS]
INDUSTRY: [YOUR CATEGORY]
CLAIMS MADE: [WHAT THE COPY ASSERTS]
SUBSTANTIATION AVAILABLE: [WHAT YOU CAN ACTUALLY PROVE, AND WHERE]
MARKETS: [COUNTRIES YOU ADVERTISE IN]

Check for:
1. Unsupported superlatives and absolute claims
2. Comparative claims referencing competitors
3. Trademark usage in ad text
4. Claims requiring substantiation — performance, results, health, financial, environmental
5. Restricted or sensitive category issues for my industry
6. Formatting violations — excessive capitalisation, repeated punctuation, symbol misuse, gimmicky spacing
7. Clickbait or misleading urgency
8. Anything requiring a disclaimer, including AI-generated content disclosure where my formats or markets require it
9. Landing page consistency — claims in the ad that the page does not support

For each issue: the exact text, the problem, the risk level, and a compliant rewrite that preserves the intent.

State clearly that this is a pre-check and not a compliance guarantee — policy enforcement varies by market and reviewer, and anything in a regulated category should go past whoever handles your legal review. Flag which items specifically warrant that.

13. The Ad Extension Set

Write the asset extensions for this campaign.

CAMPAIGN: [WHAT IT SELLS]
LANDING PAGES AVAILABLE: [LIST WITH THEIR PURPOSE]
OFFERS AND USPS: [THE REAL ONES]
BUSINESS DETAILS: [LOCATIONS, HOURS, PHONE, PRICES IF PUBLIC]

Produce, with character counts shown for every line:
1. Sitelinks — 6, each with headline (25 chars) and two description lines (35 chars each), pointing at genuinely different pages
2. Callouts — 8, each 25 characters or fewer, no duplication of headline messaging
3. Structured snippets — the right header type for my business, with values
4. Price assets where relevant
5. Promotion assets if I have a real offer
6. Call and location assets where applicable

Then:
- Self-audit every character count and fix any that exceed the limit
- Explain which extensions matter most for my campaign type and why
- Flag any that would be better left off than filled in weakly

Rules: sitelinks must point at meaningfully different destinations, not the same page with different labels. Callouts must add information the headlines do not already carry. Do not invent business details I did not supply.

14. The Ad Copy Post-Mortem

Analyse why some of my ad assets are underperforming.

ASSET PERFORMANCE: [PASTE HEADLINES AND DESCRIPTIONS WITH THEIR IMPRESSIONS AND PERFORMANCE RATING]
CAMPAIGN CONTEXT: [WHAT IT SELLS, TARGET, AND THE KEYWORDS]
LANDING PAGE: [WHAT IT SAYS]

Analyse:
1. What the strong assets have in common — angle, specificity, structure, offer presence
2. What the weak ones share
3. Whether low-impression assets are genuinely weak or simply never served enough to judge
4. Whether the mix is unbalanced, with too many assets making the same point

Then:
- Which assets to remove, and which to leave despite weak ratings because they cover a role nothing else does
- Replacements for the removed ones, with character counts, following the pattern of what works
- What this tells me about my audience that I can apply to the landing page too

Be careful about the difference between a poor asset and an under-served one. Google's asset ratings are noisy at low volume, and pruning aggressively on thin data makes the set worse.

Performance Max and AI Max (Prompts 15–19)

15. The PMax Asset Group Brief

Design the asset groups for a Performance Max campaign.

PRODUCTS OR SERVICES: [THE RANGE]
AUDIENCE SEGMENTS: [WHO BUYS, AND HOW THEY DIFFER]
CREATIVE ASSETS AVAILABLE: [IMAGES, VIDEO, LOGOS — HONESTLY]
FEED: [IF ECOMMERCE, DESCRIBE ITS QUALITY AND COVERAGE]
CONVERSION GOAL AND VALUE: [DETAIL]

Produce:
1. How many asset groups to create, and the logic for splitting them — by product line, audience, or margin
2. Per asset group: the theme, the products or services it covers, and the audience signal
3. The full asset requirements per group with character limits stated: short headlines (30), long headlines (90), descriptions (90), business name (25)
4. The image and video requirements, and what to do if I lack video
5. Search themes to supply, and how they differ from keywords
6. Brand exclusions to apply

Then flag:
- Where my available creative is too thin for an asset group to perform, and what to shoot or make first
- How PMax will report on this, given asset-group-level and channel-level reporting, and what will still be opaque
- How to stop PMax absorbing branded search that would have converted anyway

Be direct if my creative inventory cannot support the number of asset groups I want.

16. The PMax Diagnostic

Diagnose my Performance Max campaign.

PERFORMANCE: [SPEND, CONVERSIONS, CONVERSION VALUE, CPA OR ROAS, BY ASSET GROUP IF AVAILABLE]
CHANNEL BREAKDOWN: [IF YOU HAVE CHANNEL-LEVEL REPORTING, PASTE IT]
SEARCH TERMS OR CATEGORIES: [WHAT YOU CAN SEE]
ASSET PERFORMANCE: [RATINGS BY ASSET GROUP]
COMPARISON: [HOW YOUR SEARCH CAMPAIGNS PERFORM ON SIMILAR PRODUCTS]

Work through:
1. Whether PMax is genuinely incremental or is claiming conversions that branded search would have won anyway
2. Which asset groups carry the performance and which are dragging
3. Whether spend is concentrated in a channel that is not the one I want
4. Whether weak creative in one group is pulling the whole campaign down
5. Whether the feed, where applicable, is limiting what PMax can show

For each: what the data indicates and what additional data would confirm it.

Then give me:
- The most likely problem and its fix
- The test that would prove incrementality rather than assuming it
- What to change first, and what not to touch while it relearns

Be honest about the limits of PMax reporting. Where my data genuinely cannot answer a question, say so instead of inferring a cause.

17. The AI Max Readiness Check

Assess whether my Search campaigns are ready for AI Max.

CAMPAIGNS: [STRUCTURE, KEYWORDS, MATCH TYPES]
PERFORMANCE: [CONVERSIONS, CPA OR ROAS, BY CAMPAIGN]
NEGATIVE LISTS: [MATURITY AND COVERAGE]
LANDING PAGES: [HOW MANY, AND HOW WELL THEY COVER THE RANGE]
BRAND CONTROLS: [WHAT YOU HAVE IN PLACE]
CURRENT USE OF DSA: [IF ANY]

Assess readiness against:
1. Conversion volume — is there enough signal for expanded matching to learn from
2. Negative keyword coverage — expanded matching without mature negatives widens waste
3. Landing page coverage — final URL expansion needs pages worth expanding to
4. Brand protection — what stops it bidding on terms I want excluded
5. Reporting expectations — what I will and will not be able to see

Then cover the three AI Max features individually — search term matching, text customisation, and final URL expansion — and tell me which to enable, which to hold, and why for each.

Note that Dynamic Search Ads are being auto-upgraded into AI Max ahead of a 2027 sunset, with campaign-level broad match and automatically created assets upgrading earlier. If I run DSA, tell me what to do before that happens to me rather than after.

Give me the rollout order, what to watch in the first two weeks, and the signal that means turn a feature back off.

18. The Shopping Feed Optimisation

Optimise my product feed for Shopping and Performance Max.

FEED SAMPLE: [PASTE SEVERAL ROWS INCLUDING TITLE, DESCRIPTION, PRODUCT TYPE, GTIN, BRAND, ATTRIBUTES]
CATEGORY: [WHAT YOU SELL]
DISAPPROVALS OR WARNINGS: [ANY MERCHANT CENTER ISSUES]
PERFORMANCE: [WHICH PRODUCTS GET IMPRESSIONS AND WHICH DO NOT]

Produce:
1. A product title formula for my category, with the attribute order that matters most for matching, and rendered examples from my rows
2. What belongs in the description versus the title
3. Required and recommended attributes I am missing that would improve matching
4. Which of my sample rows are weakest and why
5. Custom labels worth adding for campaign segmentation

Then:
- Identify feed problems likely causing products to get no impressions
- Flag anything that risks a Merchant Center disapproval
- Tell me the single feed change that would most improve coverage

Rules: titles must remain accurate. Do not recommend keyword stuffing a title with terms the product does not genuinely match — it wins impressions on queries that will not convert and it risks disapproval.

19. The Automation Guardrails

Design the guardrails for my automated campaigns.

CAMPAIGNS ON AUTOMATION: [WHICH, AND WHICH FEATURES]
BUDGET AT RISK: [MONTHLY]
BUSINESS CONSTRAINTS: [WHAT MUST NEVER HAPPEN — categories you cannot serve, regions, margins, stock limits]
BRAND TERMS: [YOURS]

Specify:
1. Negative keyword lists that apply account-wide regardless of automation
2. Brand exclusions and where to apply them
3. Audience exclusions
4. Content and placement exclusions
5. Budget caps and pacing controls
6. Location and language settings that prevent drift
7. Conversion goal configuration, and which goals each campaign should optimise toward

Then:
- The monitoring routine — what to check daily, weekly and monthly, and roughly how long each takes
- The specific metrics that indicate automation is drifting somewhere unprofitable
- The kill criteria: what result means pause it rather than tune it

The point of guardrails is that automation optimises for the goal you gave it, not the goal you meant. Tell me where my stated setup leaves room for it to do something profitable-looking but commercially wrong.

Landing Pages and Quality Score (Prompts 20–22)

20. The Ad-to-Page Message Match

Align my landing page with the ads driving traffic to it.

ADS: [PASTE HEADLINES AND DESCRIPTIONS]
KEYWORDS: [WHAT TRIGGERS THEM]
LANDING PAGE: [PASTE THE ABOVE-THE-FOLD COPY AND DESCRIBE THE PAGE]
CONVERSION RATE: [CURRENT, IF KNOWN]

Assess:
1. Whether the page confirms the promise the ad made, in the first screen and in the same language
2. Where the ad's specific offer disappears on the page
3. Whether the primary action is obvious and matches the ad's implied next step
4. Friction between arrival and conversion — form length, required fields, distractions, unnecessary steps
5. Whether the page answers the objection the searcher arrived with

Then produce:
- The rewritten above-the-fold copy, echoing the ad's language
- The three highest-impact page changes, ranked
- What to test first and how to know it worked

Rules: do not add claims to the page that the product cannot support just to match an ad. If the ad is overpromising, the correct fix is to change the ad — say so plainly if that is what the evidence shows.

21. The Quality Score Diagnosis

Diagnose my Quality Score components and tell me what is actually fixable.

KEYWORDS WITH SCORES: [PASTE KEYWORD, QUALITY SCORE, EXPECTED CTR, AD RELEVANCE, LANDING PAGE EXPERIENCE]
ADS: [PASTE THE COPY]
LANDING PAGES: [DESCRIBE]
PERFORMANCE: [CTR, CONVERSION RATE, AVERAGE CPC]

For each low-scoring keyword, identify:
1. Which component is dragging it, and what that component actually measures
2. The specific cause — ad relevance, page experience, or historical CTR
3. The fix, and how long it realistically takes to show

Then:
- Group keywords by shared root cause so I fix patterns rather than individual keywords
- Tell me which low scores are not worth fixing because the keyword should be paused instead
- Explain what Quality Score does and does not influence, so I calibrate how much effort it deserves

Be measured about this. Quality Score is a diagnostic, not a target, and chasing it on low-volume keywords is usually a poor use of time compared with fixing conversion tracking or landing page friction.

22. The Conversion Tracking Audit

The most important prompt in this set — everything Smart Bidding does depends on this being right.

Audit my conversion tracking setup.

CONVERSION ACTIONS: [LIST EACH — name, category, count setting, value, attribution model, conversion window, whether it is in "Conversions" for bidding]
HOW THEY FIRE: [TAG, IMPORT, OFFLINE, ENHANCED CONVERSIONS]
BUSINESS MODEL: [WHAT ACTUALLY MAKES YOU MONEY]
KNOWN ISSUES: [DISCREPANCIES YOU HAVE NOTICED]

Audit for:
1. Actions counted as conversions that are not real business outcomes
2. Double counting — the same outcome tracked twice through different paths
3. Wrong count setting: every versus one, and which suits each action
4. Missing or inaccurate conversion values, and where a flat proxy value would beat none
5. Conversion windows inappropriate for my sales cycle
6. Which actions should be primary for bidding and which should be secondary observation only
7. Whether offline conversions or enhanced conversions would materially improve the signal

Then tell me:
- The single tracking problem most likely to be misdirecting Smart Bidding right now
- What my bidding is currently optimising toward versus what I actually want
- The fix order, safest first, and what will temporarily disrupt bidding when I change it

Be explicit: Smart Bidding optimises toward whatever I told it to value. If the signal is wrong, no amount of ad copy or bid tuning fixes the account.

Bidding, Budget and Measurement (Prompts 23–27)

23. The Bidding Strategy Selection

Recommend a bidding strategy for each of my campaigns.

CAMPAIGNS: [FOR EACH — type, monthly spend, conversions per month, conversion value, current strategy and its target, current performance]
BUSINESS GOAL: [VOLUME, EFFICIENCY, PROFIT, OR A SPECIFIC TARGET]
MARGIN DATA: [IF AVAILABLE]

For each campaign, recommend a strategy and explain:
1. Why it suits this campaign's data volume and goal
2. The minimum conversion volume it needs, and whether I meet it
3. What target to set initially, and how to derive it from my current performance rather than guessing
4. How long the learning period will take and what to expect during it
5. What to do if I do not have enough conversions for the ideal strategy

Then:
- Flag any campaign where the current strategy is fundamentally mismatched to its data
- Explain when Maximise Conversions with no target beats a tCPA, and vice versa
- Warn me about the changes that reset learning, and how to sequence changes to minimise disruption

Do not recommend a value-based strategy where I have not confirmed that my conversion values are accurate. Tell me to fix the values first if that is the situation.

24. The Bid Target Calculation

Help me set the right bidding target from my actual economics.

REVENUE PER CONVERSION: [FIGURE]
GROSS MARGIN: [PERCENTAGE OR FIGURE]
OTHER VARIABLE COSTS PER SALE: [SHIPPING, PAYMENT FEES, RETURNS RATE]
LEAD TO SALE RATE: [IF LEAD GEN]
CUSTOMER LIFETIME VALUE: [IF YOU HAVE IT, AND HOW CONFIDENT YOU ARE IN IT]
CURRENT PERFORMANCE: [CPA OR ROAS, AND SPEND]
GROWTH VERSUS PROFIT PRIORITY: [WHICH MATTERS MORE RIGHT NOW]

Calculate, showing your working:
1. Break-even CPA or ROAS on first purchase alone
2. Target allowing my required profit margin
3. Target if I am willing to invest against lifetime value, and the risk that carries
4. The maximum I could pay and still be profitable

Then:
- Recommend the target to set, and why
- State how sensitive the answer is to my lifetime value figure, since that is usually the least reliable input
- Tell me what happens to volume if I tighten the target, and what happens to profit if I loosen it

If my lifetime value number is an estimate rather than measured, say plainly how much weight it should carry in this decision.

25. The Test Design

Design a valid experiment for this change.

CHANGE I WANT TO TEST: [DESCRIBE]
CAMPAIGN: [WHICH ONE, AND ITS MONTHLY CONVERSIONS AND SPEND]
CURRENT PERFORMANCE: [BASELINE METRICS]
HYPOTHESIS: [WHAT YOU EXPECT AND WHY]

Produce:
1. The test structure — experiment, split, or before-and-after — and why that one suits this change
2. The split percentage and the reasoning
3. How long it needs to run, based on my conversion volume, to detect a meaningful difference
4. The primary metric, and the secondary metrics that would reveal an unintended consequence
5. What result would count as a win, stated numerically before the test starts
6. What could confound this test — seasonality, other changes, learning periods

Then tell me honestly:
- Whether my conversion volume is sufficient to detect the effect size I care about, or whether this test will produce noise
- If it is not, what to do instead
- The most common way tests like this get misread

Do not design a test my account cannot support. Saying "you do not have the volume for this" is a more useful answer than a test that will produce a meaningless result.

26. The Performance Report

Turn this account data into a report for a non-technical stakeholder.

DATA: [PASTE PERFORMANCE BY CAMPAIGN — spend, conversions, value, CPA or ROAS, period over period]
BUSINESS CONTEXT: [WHAT THEY CARE ABOUT]
PERIOD: [DATE RANGE AND COMPARISON]
KNOWN EVENTS: [ANYTHING THAT AFFECTED THE PERIOD]

Write a report that:
1. Opens with the commercial outcome, not the activity
2. States what changed and why, separating what I did from what the market did
3. Explains any decline honestly and early, with the plan attached
4. Covers what is being done next, with dates
5. Ends with anything needed from them

Rules:
- No vanity metrics. Impressions and clicks only appear if they explain a change in the outcome
- Every number from the data I pasted
- Under 500 words
- Explain jargon inline where it is unavoidable, or cut it
- If the period was bad, the report should read as a bad period handled competently rather than a good period spun

Then flag any claim in my data that I should be able to substantiate if challenged.

27. The Incrementality Question

Assess how much of my paid search performance is genuinely incremental.

CAMPAIGNS: [BRAND AND NON-BRAND SPEND, CONVERSIONS AND VALUE]
ORGANIC PERFORMANCE: [FOR THE SAME TERMS, IF AVAILABLE]
BRAND SEARCH VOLUME: [TREND]
ATTRIBUTION MODEL: [WHAT YOU USE]
OTHER ACTIVE CHANNELS: [WHAT ELSE IS DRIVING DEMAND]

Assess:
1. How much branded paid conversion volume would likely have arrived through organic anyway
2. Whether Performance Max is claiming credit for demand created elsewhere
3. Where the attribution model is likely over-crediting paid search
4. Which campaigns are most likely genuinely incremental

Then:
- Design a practical incrementality test I could actually run, given my budget and risk tolerance
- State what it would cost to run and what it would tell me
- Give the honest caveats of that test design

Be candid about uncertainty. Incrementality is genuinely hard to measure, and a confident number here would be less useful than a clear description of what I do and do not know.

Diagnostics and Scaling (Prompts 28–30)

28. The Performance Drop Diagnosis

Diagnose a drop in Google Ads performance.

THE DROP: [WHEN IT STARTED, WHICH METRIC, HOW STEEP]
SCOPE: [WHICH CAMPAIGNS, AD GROUPS OR PRODUCTS]
DATA: [PASTE BEFORE AND AFTER — impressions, clicks, CTR, CPC, conversion rate, CPA, impression share]
WHAT CHANGED MY SIDE: [EDITS, BUDGET CHANGES, LANDING PAGE DEPLOYS, TRACKING CHANGES, STOCK OR PRICE CHANGES]
EXTERNAL FACTORS: [SEASONALITY, COMPETITORS, ANYTHING KNOWN]

Work through in this order and tell me what my data indicates for each:
1. Tracking — is this a measurement problem rather than a performance problem
2. Volume — did impressions fall, indicating auction or budget changes, or did conversion rate fall, indicating a page or offer problem
3. Auction pressure — impression share lost to rank or budget, and CPC movement
4. My own changes — the timeline against my edits, including anything that reset learning
5. Landing page or site availability
6. Seasonality and demand
7. Policy — any disapprovals or limited status

For each: the evidence that would confirm or rule it out.

Then give me the most likely cause, your confidence, the diagnostic to run next, and the fix. If impressions held but conversion rate fell, say clearly that this points at the page or the offer rather than the campaign.

29. The Scaling Plan

Build a plan to scale this campaign without breaking its efficiency.

CAMPAIGN: [WHAT IT IS, CURRENT SPEND, CONVERSIONS, CPA OR ROAS]
IMPRESSION SHARE: [CURRENT, AND SHARE LOST TO BUDGET AND TO RANK]
TARGET: [DESIRED SPEND OR VOLUME]
CONSTRAINTS: [EFFICIENCY THRESHOLD, INVENTORY, FULFILMENT CAPACITY]

Produce:
1. How much headroom exists in the current campaign before I need new sources of volume — use my impression share figures to reason about this
2. The order to unlock volume in: budget, targets, match types, geography, new keywords, new campaign types
3. The rate to scale at, and why moving too fast destabilises Smart Bidding
4. Expected efficiency impact at each step, directionally rather than numerically
5. Where the ceiling likely is for this campaign

Then:
- The single highest-confidence next step
- The leading indicator that scaling is working before CPA data matures
- The point at which I should stop and consolidate rather than push further

Be realistic. Most campaigns cannot double spend at the same efficiency, and telling me the trade-off honestly is more useful than an optimistic plan.

30. The Account Audit

Audit my Google Ads account as an external specialist would. Be direct.

ACCOUNT OVERVIEW: [CAMPAIGNS, SPEND, STRUCTURE]
PERFORMANCE: [BY CAMPAIGN]
SETTINGS: [BIDDING, LOCATIONS, NETWORKS, AD SCHEDULE, DEVICE ADJUSTMENTS]
CONVERSION SETUP: [ACTIONS AND CONFIGURATION]
NEGATIVES: [COVERAGE]
ASSETS: [AD COPY AND EXTENSIONS]

Audit and rank findings by money at stake:
1. Conversion tracking correctness — everything depends on this
2. Wasted spend: search terms, placements, geography, devices, schedule
3. Structural problems limiting Smart Bidding
4. Missing campaign types or coverage gaps
5. Ad copy and asset quality
6. Landing page and message-match issues
7. Settings that are silently costing money — search partners, display expansion, broad location targeting
8. Anything at policy risk

For each finding: the issue, the money at stake, the fix, and the effort required.

Then give me:
- The three fixes to do this week
- The one structural change with the largest long-term impact
- What is fine as it is, so I do not churn the account unnecessarily

Do not list strengths unless I ask. I want the problems, ordered by what they cost me.

Running These in Order

1. Fix measurement first — prompt 22. Everything downstream optimises toward whatever this says.
2. Get the structure right — prompts 1 to 4.
3. Build the keyword and negative layer — prompts 5 to 9, then 6 and 7 on a recurring cycle.
4. Write the assets — prompts 10 to 14, with the character self-audit every time.
5. Set up automation deliberately — prompts 15 to 19, with guardrails before scale.
6. Align the page — prompts 20 and 21.
7. Set targets from economics — prompts 23 and 24.
8. Then test, report, diagnose and scale — 25 to 30 as the ongoing loop.

Mistakes That Cost Money

1. Trusting the character counts. Models cannot count reliably. Demand a count per line and verify before uploading.
2. Letting it invent CPCs or search volumes. Every number it produces without your data is fabricated.
3. Optimising for Ad Strength. It measures variety, not performance. Excellent copy can score Good.
4. Broad match before mature negatives. Fast way to spend a budget on nothing.
5. Wrong conversion actions marked primary. Smart Bidding will faithfully buy you the wrong outcome.
6. Judging PMax without asking about incrementality. It reports well partly because it absorbs demand you already had.
7. Changing several things at once. Learning resets and you cannot attribute the result.
8. Ignoring the DSA-to-AI-Max upgrade timeline. Auto-upgrades are coming; decide before they decide for you.
9. Shipping copy without a policy pre-check. Disapprovals at scale stall a launch.
10. Fixing ads when the landing page is the problem. Diagnose before you rewrite.

Where to Run These

The analysis prompts reward a frontier reasoning model that will show its working and admit when your data is insufficient. For anything needing live SERP or competitor observation, an agentic tool that can genuinely browse beats a model reasoning from training data.

For the creative side of paid media, ChatGPT prompts for AI ads and commercials covers video and display assets, and 42+ AI UGC video prompts covers creator-style ad formats. On the organic side, eCommerce SEO prompts use the same discipline applied to search you do not pay for.

Keep Reading

More on paid and organic acquisition: ChatGPT prompts for AI ads, eCommerce SEO prompts, 42+ AI UGC video prompts, cold email outreach prompts, 23+ AI prompts for entrepreneurs, and AI prompts for YouTube thumbnails. Or browse all guides and prompts on PromptsRush.