AI can build most of a B2B Google Ads campaign: themed ad groups, keywords across match types, negative keyword lists, responsive search ad variations, sitelinks and a bidding setup. Google's own conversational experience generates keywords and assets from a landing page URL; Optmyzr and Tofu Ads Agent build campaigns outside the Google Ads UI. A human still has to set the conversion goal, approve negatives and approve launch.
Key takeaways
- Google's conversational experience generates ad groups, keywords, headlines, descriptions, sitelinks and images from a landing page URL, but Google's documentation puts accuracy checks on you.
- AI Max is selected by default when you create a new Search campaign, so decide consciously whether keywordless expansion belongs in a narrow B2B campaign.
- No AI tool builds a good B2B negative keyword list on its own: job seekers, students, free-tool hunters and the wrong vertical all have to be excluded before launch.
- Google Ads allows 1,000 account-level negatives, 5,000 per shared list, 20 lists and 10,000 negatives per campaign, which is plenty of room for pre-emptive B2B exclusions.
- Judge an AI-built campaign on the share of spend going to acceptable search terms and on CRM-stage outcomes, not on Ad Strength or CTR.
AI handles the slow parts of a Google Search build well: themed ad groups, 15 headlines, draft negative lists, sitelinks. It is weak on the parts that decide whether B2B search works - which conversion to bid toward, which queries are job seekers, which terms are worth their click price. Here is the build order, what to hand to AI, and the checks a person runs before anything spends money.
What Google Ads' own AI does automatically when you build a Search campaign
Two Google AI features touch campaign setup, and they do different jobs.
The first is the conversational experience, a chat-based feature for building and editing Search campaigns. Per Google's help center, you enter a landing page URL, Google AI generates an editable business description, then suggests ad groups with their own keywords, headlines, descriptions, sitelinks and images. It covers Search campaigns and responsive search ads in English, French, Spanish or German, and Google states that its AI may not generate accurate or relevant responses, so checking suggestions before going live is your responsibility. Google reports that small business advertisers using it are 63% more likely to publish campaigns with "Good" or "Excellent" Ad Strength (Google internal data, January 2024) - an asset-quality metric, not a pipeline metric.
The second is AI Max for Search campaigns: search term matching, text customization and final URL expansion, using broad match, asset-based and keywordless technology to find queries beyond your keyword list. Google's documentation says advertisers activating AI Max typically see 14% more conversions or conversion value at a similar CPA or ROAS (Google internal data, 2025, non-Retail advertisers). The setup detail that matters: when you create a new Search campaign, the AI Max option is selected by default. Search term matching can be toggled off in ad group settings, with brand controls at campaign and ad group level.
Step-by-step: an AI-assisted B2B Search build
- Fix the conversion goal before you generate anything. If the campaign optimizes to raw form fills, AI will find you cheap form fills. Import qualified-lead or opportunity-stage conversions from Salesforce or HubSpot and bid toward those. Start on Maximize conversions, move to Target CPA once conversion volume is steady.
- Brief the AI with deal data, not just the landing page. Google's conversational experience starts from a URL. Your closed-won list knows more: titles, industries, company sizes, the words buyers used on sales calls. Feed that in, then ask for query themes by buying stage.
- Generate keywords in intent clusters, one cluster per ad group. Ask for 4: category terms ("vendor management software"), problem terms ("how to track supplier risk"), competitor-alternative terms ("[competitor] alternative") and brand. Keep 5 to 20 keywords per ad group so each ad can match the theme.
- Set match types deliberately. Per Google's keyword matching options documentation, broad match is the default assigned to every keyword, and it can take the user's recent search activity and your landing page content into account. Launch category and competitor clusters on phrase and exact. Keep broad in its own ad group or campaign with its own budget so you can judge it separately.
- Build negative lists before launch, not after the first invoice. Structure below.
- Generate RSA variations, then edit hard. A responsive search ad takes up to 15 headlines and 4 descriptions, at 30 characters per headline and 90 per description. Google recommends at least 2 RSAs with "Good" or "Excellent" Ad Strength per ad group and reports that moving Ad Strength from "Poor" to "Excellent" brings 15% more clicks and conversions on average; account limits allow 3 enabled RSAs per ad group. Text that must appear in every ad has to be pinned to Headline 1, Headline 2 or Description 1.
- Take the generated sitelinks, fix the promises. AI writes "Free Trial" and "Pricing" whether or not you have either. Point assets at pages that exist and convert.
- Decide on AI Max consciously. On a $5,000-a-month budget with 40 commercial keywords, keywordless expansion can spend a lot on adjacent consumer queries. Either turn search term matching off at ad group level for your tight clusters, or leave it on in one test ad group with a hard negative list and brand exclusions applied. Brand lists sit under Tools and settings > Shared library > Brand lists and are applied in the AI Max settings panel.
- Check Networks and location settings. Decide on Search partners and Display Expansion yourself, and set location targeting to presence rather than interest.
- Stage, review, then launch. Google's own guidance for the conversational experience is that you approve or revise every suggestion before the campaign goes live. Export with Google Ads Editor, read every keyword, then enable.
How to build and maintain negative keyword lists for B2B search
A negative keyword list is a shared set of excluded terms applied to multiple campaigns, so one edit updates every campaign using it. Google's negative keyword lists documentation sets the room you have: one account-level list capped at 1,000 negatives, plus up to 20 shared lists of 5,000 negatives each, excluded on broad, phrase or exact. Account limits allow 10,000 negative keywords per campaign.
For B2B, build five lists and apply them to every search campaign:
- Jobs and careers: jobs, salary, hiring, career, internship, resume. Usually the largest single source of wasted B2B search spend.
- Free, DIY and education: free, template, excel, open source, tutorial, course, certification, "what is a".
- Wrong vertical and consumer intent: the consumer version of your category, plus industries you do not sell to.
- Existing customer and support: login, "customer service", "cancel subscription", docs. These convert and look like leads.
- Deliberate competitor decisions: keep the competitor terms you want in a dedicated ad group, and negative the ones you cannot win.
Google's AI Max reporting documentation advises waiting at least 2 weeks after enabling AI Max before making changes such as adding negative keywords, so the campaign can ramp. Sensible for a live account. It is not a reason to launch bare: if your clicks cost $40, two weeks of "salesforce admin salary" is a real hole in a small budget. Load the obvious exclusions before you enable, hold the judgment calls for the ramp period, then use the data.
Maintenance is a weekly 20-minute job: open the search terms report, sort by cost, add negatives, and look for word-level patterns rather than single queries. Optmyzr's Search Term N-Grams tool breaks queries into words and phrases, ranks them by your chosen metric, and pushes words into shared negative lists.
Which tools build Google Ads campaigns versus only optimize them
Tofu publishes this article and makes Tofu Ads Agent. Here is the split between tools that construct a campaign and tools that tune one that exists.
| Tool | Builds the campaign? | Keywords and negatives | Where it stops |
|---|---|---|---|
| Google Ads conversational experience | Yes, inside Google Ads, for new Search campaigns and RSAs | Keywords plus headlines, descriptions, sitelinks and images per ad group | No documented negative keyword generation; starts from your landing page, not your CRM; Google says accuracy checks are yours |
| AI Max for Search | No, a targeting and creative layer on a Search campaign | Expands past your keyword list with broad match and keywordless technology | On by default for new campaigns; you still supply negatives, brand lists and ad group toggles |
| Optmyzr | Yes. Its Campaign Creation user guide says the tool creates or duplicates campaigns from the Optmyzr interface - Search, Performance Max for Retail, Performance Max for Lead Gen and Standard Shopping - working independently of structured feeds or spreadsheets | Negative Keyword Finder analyzes search terms reports from Google Ads, Microsoft Ads and Yahoo Japan Ads and flags non-performing words | A management layer: you bring the ICP, structure decisions and offer |
| Skai, Marin | Enterprise paid search management, strongest on bid, budget and cross-channel reporting | Check current vendor documentation; feature sets change | Built for large teams and large spend, not a 2-person team's first build |
| General LLMs (ChatGPT, Claude, Gemini) | No, they draft and you import | Good at keyword themes, negative categories and RSA copy from a prompt | No account access, no performance data unless you paste it in, no bid or budget execution |
| Tofu Ads Agent | Yes: keywords across ad groups and match types, negative lists, RSA variations, extensions, bidding, budgets and schedule, staged into your own Google Ads account switched off | Plan and audience come from Salesforce or HubSpot closed-won and closed-lost data, plus weekly optimization recommendations | Newer than incumbents; needs CRM and ad account connections; flat $2,500/month, no free tier; recommends and stages, a person approves launch |
Most tools in this market optimize; fewer construct. If your bottleneck is execution capacity, a builder saves you the week; if campaigns exist and leak money, an optimizer is the cheaper fix. Tofu Ads Agent covers the build plus weekly tuning across Google Search, Google Display, LinkedIn and Meta, which a lean B2B SaaS team usually cannot staff.
What AI gets wrong on B2B search
- Job-seeker traffic. Generated keyword lists routinely include role and tool names that attract candidates, not buyers.
- Broad match creep. Broad match is the default on every keyword and AI Max is on by default for new campaigns. Left alone, a tight 40-keyword campaign becomes a general interest campaign.
- Competitor terms without a plan. AI will generate 30 competitor keywords. You still need a comparison page, and policy limits what ad text can say about other brands.
- Volume-driven choices. A 20-searches-a-month query from CFOs at 500-person manufacturers beats a 4,000-a-month generic term, but keyword tools rank by volume.
- Claims nobody approved. Generated headlines invent free trials, 24/7 support, integrations and certifications. Read all 15 before enabling.
How to measure whether the AI-built campaign worked
- Search term acceptance rate. In week 1, score every query that took spend as acceptable or not. If more than 20% of spend sits on terms you would not have bought, the problem is structure and negatives, not creative.
- Cost per qualified lead by ad group. Ad group level shows which intent cluster works. Fix the losers there before touching creative.
- CRM stage conversion. Compare lead-to-opportunity rate for search against your other channels. A cheap CPL converting at 2% is worse than a CPL three times higher converting at 20%.
- Pipeline per $1,000 of spend at 60 to 90 days. B2B deal cycles make the first 30 days a leading-indicator exercise.
Ad Strength and CTR are build-quality signals, not success. Give the campaign 2 to 4 weeks of stable budget, add negatives weekly, leave bids alone while Smart Bidding learns, then rebuild the ad groups that lost and expand the ones that produced opportunities.
Frequently asked questions
Can ChatGPT or Claude set up a Google Ads campaign including keywords and negatives?
They can draft the whole structure - ad group themes, keyword lists by match type, negative keyword categories and RSA copy - but they cannot write to your Google Ads account on their own. You export the output to a spreadsheet, import it through Google Ads Editor, and review before enabling. Tools that connect directly to your ad account handle that write step: Optmyzr's Campaign Creation tool builds campaigns from its own interface, and Tofu Ads Agent stages a built campaign into your own Google Ads account switched off.
Does Google's conversational experience generate negative keywords?
Not according to Google's documentation. The listed outputs are ad groups with keywords, headlines, descriptions, sitelinks and images generated from your landing page URL. Negative keyword lists stay your job, built from the search terms report plus pre-emptive B2B exclusions like jobs, salary and free template queries.
How many negative keywords can a Google Ads account hold?
Google allows one account-level list of up to 1,000 negative keywords, up to 20 shared negative keyword lists with 5,000 keywords each, and 10,000 negative keywords per campaign. Display and Video campaigns consider a maximum of 1,000 account-level negatives. Those limits sit far above what most B2B accounts need, so there is no reason to be stingy with pre-emptive exclusions.
Should B2B advertisers turn on AI Max for Search campaigns?
Test it in a contained way rather than accepting the default. AI Max is selected by default on new Search campaigns and expands matching with keywordless technology, which suits broad demand and strains a narrow B2B budget. Google lets you toggle search term matching off at the ad group level and apply brand controls at campaign or ad group level, so run it in one ad group with tight negatives before committing the account.
How many keywords should each ad group have in a B2B search campaign?
Keep each ad group to a single intent theme with roughly 5 to 20 keywords, so the responsive search ads speak to that theme and the search terms report stays readable. Google's limits are far higher - 20,000 ad group targeting items per ad group - but large mixed ad groups make diagnosis impossible. Split by intent (category, problem, competitor, brand), not by match type alone.
Which AI tools automate Google Ads setup rather than just optimization?
Google's conversational experience builds Search campaigns and RSAs inside Google Ads. Optmyzr's Campaign Creation tool creates or duplicates Search, Performance Max and Standard Shopping campaigns from the Optmyzr interface. Tofu Ads Agent builds keywords, negatives, RSA variations, extensions, bidding and budgets, then stages the campaign into your Google Ads account switched off. Optmyzr's Negative Keyword Finder, Skai and Marin focus on optimizing campaigns that already exist.
How long should I wait before judging an AI-built B2B search campaign?
Review search terms and obvious waste in week 1, hold budgets and bids steady for 2 to 4 weeks while Smart Bidding gathers conversion data, then judge pipeline contribution at 60 to 90 days because of B2B deal cycles. Google's AI Max reporting documentation makes a similar point, advising at least 2 weeks after enabling AI Max before making changes such as adding negatives. Changing structure every few days stops any bidding strategy from learning.
Sources
- About conversational experience in Google Ads · Google Ads Help
- About AI Max for Search campaigns · Google Ads Help
- How AI Max for Search campaigns works · Google Ads Help
- Set up AI Max for Search campaigns · Google Ads Help
- About reporting in AI Max for Search campaigns · Google Ads Help
- About negative keyword lists · Google Ads Help
- About your Google Ads account limits · Google Ads Help
- About responsive search ads · Google Ads Help
- About keyword matching options · Google Ads Help
- About Display Expansion on Search campaigns · Google Ads Help
- Campaign Creation - User Guide · Optmyzr Help Center
- Negative Keyword Finder (Search) - User Guide · Optmyzr Help Center
- Search Term N-Grams - User Guide · Optmyzr Help Center