AI Keyword Tool: Adapt Your Google Ads to Changing Search Behavior with This Prompt
As AI-driven tools change the way people search for information, conversational queries are becoming increasingly relevant to the PPC keyword strategy. Using data from our accounts, we find that while longer keywords continue to account for less overall search volume, they convert better and at an increasingly higher rate, and should therefore be an important part of your keyword approach in a rapidly changing, AI-driven PPC landscape.
In this article, we discuss the trends and impacts using data from our accounts, and share a prompt we’ve developed to find relevant keyword ideas to adapt to the AI-driven search shifts.
The Basics: Short-Tail vs. Long-Tail Google Ads Keywords
Keywords are the foundation of a paid search campaign setup, signaling to the platforms which searches particular ads should be eligible for. Together with match types, they control how closely a search has to resemble the keyword before an ad can appear. Keywords are built from the products, services and locations on offer, then refined over time using the search term report and negative keyword placements.
Keyword Type | Word Count | Example | Characteristics/Intent |
Short-tail | 1–2 words | home care | High volume and broad intent. Searchers could be researching, comparing, or ready to buy. |
Mid-tail | 3–4 words | in home care services | Adds specific details, usually indicating clearer intent. |
Long-tail | 5+ words | affordable in home care services for seniors | Low volume per term, but specific intent and thinner competition; often full questions or requests with constraints. |
What Our Client Data Shows
To see how search behavior is showing up in our client accounts, we reviewed search term data across client campaigns. We grouped searches by word count to compare how shorter and longer queries contributed to clicks and conversions. Looking at data from October 2025 to October 2026, we find a number of insights. First, longer queries represent a substantial share of reported search terms.
Figure 1: Distribution of Reported Search Terms Across Client Accounts, October 2025-2026
Second, longer queries also show stronger conversion rates.
Figure 2: Conversion Rate, by Query Length, October 2025-October 2026
Finally, cost by conversion is lower for longer queries.
Figure 3: Cost By Conversion (USD), By Query Length, October 2025-2026
Comparing the 2025-2026 period to a year before, we find that while shorter searches still account for the largest share of total searches, the share of conversion attributable to long queries (5 words or more) is increasing, indicating that they are becoming more valuable to conversion performance. In 2025-2026, longer queries generated 12.3% of clicks but 30.8% of conversions. Also, their share of conversions increased from 25.2% to 30.8% year-on-year.
Metric | 2024–2025 | 2025–2026 |
Longer queries’ share of clicks | 12.0% | 12.3% |
Longer queries’ share of conversions | 25.2% | 30.8% |
Keywords are getting longer as people use AI to type or speak their questions into and your older campaign setup may no longer suffice. While broad, phrase match and AI Max may capture some of those longer tail queries, the only way to guarantee that you will show up on these queries is by adding them to your account. Another bonus of this strategy is that you can also observe performance and adjust bids and ad copy accordingly.
Our AI-Style Keyword Mining Tool
To respond to this changing landscape, we developed a new prompt that helps add longer tail keywords based on search term reports, and more “question” or “i’ types of searches. Our AI keyword tool can be customized for AI tools and works in two parts:
Part A mines actual searches. It looks for questions, first-person context, comparison language and searches containing several constraints. It filters against your business description, excludes already-added terms, ranks candidates by conversions, clicks and impressions, and separates uncertain offers into a “confirm first” list.
Part B develops new variations. Using your leading non-brand keywords, it suggests conversational versions built around real products, services, locations and customer needs. These are ideas to test, not searches with proven demand.
How to Use the Prompt
Export your search term report. In Google Ads, go to Insights and reports, then Search terms. Use the last 90 days (or longer for low-volume accounts) and include the Added/Excluded column. Download as CSV or Excel.
Optional: export your keyword list. A keyword report with impressions helps your AI chatbot skip what you already have and pick the right seeds for Part B.
Open your AI chatbot. Use a plan that can read files and run code so it can handle the export and build the CSV and spreadsheet.
Copy the prompt below and fill in the “About my business” section. This is what keeps the keywords on-brand, so be specific about what you don't sell.
Attach your files and send. Review the results, especially the confirm-first list, before uploading
The prompt (copy everything in the grey box)
I'm attaching a Google Ads (or Microsoft Ads) search term report. Please help me find long-tail, conversational "AI-style" search queries and turn them into keywords. Do two parts. ABOUT MY BUSINESS (FILL IN): - What we sell / services we offer: - What we do NOT sell (exclusions): - Locations we serve and language(s): - Our brand name and main competitors: - Goal: sales, leads, or bookings? BEFORE YOU START - Load the report with code. Google Ads exports are often UTF-16, tab-separated, with title/date rows above the headers and total rows at the bottom. Skip those. - If I attach my current keyword list, exclude anything already in the account. Compare while ignoring case, punctuation, word order, plurals and close variants. PART A: MINE AI-STYLE QUERIES FROM THE REPORT Flag search terms that read like someone typing to a chatbot: - Full sentences or questions ("what is the best ___ for ___") - Several constraints in one query (use case + budget + size + location) - First-person context ("for my small business", "I have a ___ that…") - Comparison or recommendation language ("___ vs ___", "which ___ should I buy") - Natural-language modifiers ("that actually works", "easy to install", "for beginners") These are usually 6+ words, but judge by phrasing, not length. Then filter: - Keep only terms relevant to what we sell, with buying or lead intent. - Exclude terms already added as keywords, competitor brands, job seekers, DIY/free intent, off-topic queries and anything on my exclusion list. - Rank by conversions, then clicks, then impressions. Show impressions, clicks, conversions and cost/conv. for each. - Put anything that depends on a policy you're unsure we offer in a separate "confirm first" list. PART B: WRITE AI-STYLE VARIATIONS OF MY TOP KEYWORDS 1. Pick my top 15–20 non-brand keywords by impressions, and merge near-duplicates into one theme each. 2. Write 4–6 conversational variations per theme, using: - First person: "i need to ___", "i am looking for ___", "we need ___ for ___" - Questions: "how do i ___", "how much does it cost to ___", "where can i ___", "what is the best ___ for ___" - Requests with a real constraint: use case, size, budget, timing, location, audience 3. Use only our real products, services and locations. Keep buying intent and drop purely informational questions. Don't include our brand or competitor names. 4. Drop anything already in the account, and remove duplicates. 5. Default to phrase match. Suggest adding these to the seed's ad group or to a test ad group like "AI-Style Queries". Some may show "low search volume" at first; pause any that never serve after 30–60 days. KEYWORD RULES (STRICT, BOTH PARTS) - Max 10 words and 80 characters per keyword, including spaces. - If a term exceeds a limit, shorten it while keeping its intent, and show the original next to it. - Remove characters Google Ads rejects (? ! , @ % * ( ) = ; < > | ^ ~ { } \) and apostrophes. Lowercase everything. Keep accents. - Recommend phrase or exact match for each keyword, with a one-line reason. - Before showing results, run a code check that every keyword passes these rules, and fix or drop failures. OUTPUT Part A: 1. A table: Keyword (and original if shortened) | Campaign / ad group | Match type | Words / chars | Impr. | Clicks | Conv. | Cost/conv. | Why it qualifies 2. Keyword themes, and whether each fits an existing ad group or needs a new one 3. A Google Ads Editor bulk upload CSV (Campaign, Ad Group, Keyword, Match Type), plus a separate CSV for confirm-first terms 4. Recurring questions or needs that could inform ad copy or landing pages Part B: an .xlsx with: - A "Variations" sheet: Seed keyword | Seed impressions | Language | Pattern | New keyword | Words | Characters | Match type | Campaign | Ad group | Notes. Words and Characters should be formulas, with a check column that flags anything over the limits. - An "Editor upload" sheet: Campaign, Ad Group, Keyword, Match Type In the chat, show just 5–8 of the best variations and the total count. |
Running this analysis for a sample client in the hospitality industry, you can see below the examples of keywords that the tool may suggest after running the analysis.
Original keyword | New conversational keyword |
restaurant near me | where can i get dinner and cocktails near me |
bar near me | we need a bar with food near me |
cocktail bar near me | i am looking for a cocktail bar near me |
drinks near me | where can i get drinks after dinner near me |
burgers near me | where can we eat burgers and drink near me |
Prefer To Have It Done For You?
While we do not recommend rebuilding your entire keyword strategy around longer conversational searches, you should proactively revisit and adapt your keyword strategies to the ongoing AI-driven changes in search behavior.
The prompts we shared can help you in this effort. But if you prefer for us to professionally run this analysis on your account and handle the rollout, get in touch. We’d be happy to chat!
