A prospect sees your Google ad at 9:47 p.m. They have three questions about price, timing, and whether you serve their area. Your office is closed. The prospect doesn’t want to fill out a form and wait until morning, so they keep searching.

Google wants an AI agent to handle that conversation inside the ad.

At Google Marketing Live in May 2026, Google announced Business Agent for Leads, a Gemini-powered agent that lets a person click “Chat,” ask questions, and get answers based on the advertiser’s website. Google says the goal is to create a conversational lead qualification experience on Search and send higher-intent prospects to sales teams.

The opportunity is real, but so is the risk. If the agent learns from vague, outdated, or contradictory pages, it can only work with what you gave it. Before you worry about getting access to the new ad format, make your website a reliable source of sales answers.

What Google Business Agent for Leads actually does

Google describes the format as a smart brand agent that sits inside an ad. A searcher can ask questions, receive immediate answers grounded in the advertiser’s website, and move toward a prefilled lead form when ready. Google’s own launch example shows a prospective student asking a university about programs, schedules, and eligibility before submitting their information. You can see the product example and announcement from Google.

This differs from a standard lead form asset. Google’s existing lead form assets let people submit contact information directly from an ad. Business Agent for Leads adds a conversation before the handoff.

That difference matters. A static form collects what a prospect types. A conversational agent can answer objections, ask qualifying questions, and shape what the prospect expects before your team speaks with them.

As of July 2026, availability is limited. Google lists the product as an announcement rather than a generally available self-service feature, and industry coverage reports it as a pilot for select advertisers. Google has also confirmed a beta rollout in India. Don’t tell your team it is available in every U.S. account until you see it in yours or receive confirmation from Google.

Why your website becomes part of the sales system

Most small business websites are written for human visitors and traditional search results. They use broad statements such as “solutions for every need,” hide pricing behind “contact us,” and leave operational details scattered across old posts, PDFs, and service pages.

That creates friction for people. It also creates weak source material for an agent that is supposed to answer direct questions.

Imagine a commercial HVAC company whose homepage says it serves “the greater metro area.” Its contact page lists one city. An old blog post mentions a county it no longer covers. When someone asks, “Do you service rooftop units in Westfield?” there is no single reliable answer.

The fix isn’t to stuff pages with AI keywords. Google says content shown in its AI search experiences still relies on its established technical requirements and quality practices, and that no special AI file or new markup is required. The practical job is simpler: publish accurate answers in plain language, keep them consistent, and make important information easy to find.

The seven-page readiness audit

Start with the pages most likely to contain answers a buyer needs before contacting you. You don’t need a 100-page rebuild. You need a clean source of truth.

1. Service pages

Each core service needs its own page. State what the service includes, who it is for, common exclusions, the typical process, and the next step. Replace “custom solutions” with specifics.

For example, a bookkeeping firm should say whether monthly reconciliation, payroll, sales tax filing, and catch-up work are included or sold separately. If the answer varies by plan, show the difference.

2. Pricing or cost guidance

You don’t need to publish a fixed price when every job is different. You do need to explain what drives cost.

Give a starting price, typical range, minimum engagement, or worked example when possible. If you cannot provide any number, list the variables used to quote the work and explain what information a prospect should bring. “Request a quote” is a call to action, not an answer.

3. Service-area page

Name the cities, counties, states, or distance limits you actually serve. Explain remote versus on-site availability. If travel fees apply outside a core zone, say so.

Keep this page consistent with your Google Business Profile and contact information. Google recommends keeping Business Profile information current for visibility in its AI experiences, alongside maintaining useful page content and visual assets, in its official guidance for AI features in Search.

4. Process page

Show what happens after a lead raises a hand. Cover the first call, estimate, deposit, scheduling, production, review, and support. Include realistic time frames, who attends each step, and what the customer must provide.

A buyer asking “How soon can we start?” needs more than “We respond quickly.” If your standard is a reply within one business day and projects normally begin two to four weeks after approval, publish that.

5. FAQ page

Build the FAQ from real pre-sale conversations. Ask sales and customer service staff for the questions they answer every week. Group answers by service so the page doesn’t become a random wall of text.

Don’t write ten versions of the same keyword question. One clear question and a direct answer are more useful than padded copy.

6. Policies and limitations

State cancellation terms, warranties, financing options, emergency availability, minimum orders, supported platforms, and anything you won’t do. Limitations can improve lead quality because they let poor-fit prospects self-select out.

7. Contact and handoff details

Your contact page should identify the right channel for sales, existing customers, billing, and urgent service. Show business hours and expected response times. Confirm what happens after form submission.

Google’s current lead form documentation says leads may be downloaded as a CSV or delivered through a webhook or supported integration. It also warns that downloadable lead data is only retained for a limited period, so review the lead form delivery requirements before relying on a manual process.

Build an answer inventory before rewriting pages

Open a spreadsheet and list the 30 questions prospects ask before buying. Use columns for the approved answer, source page, owner, last review date, and whether the answer changes by location or service.

Start with these five groups:

  • Fit: Who is the service for, and who is it not for?
  • Cost: What does it cost, and what changes the price?
  • Timing: How fast can you respond, start, and finish?
  • Risk: What are the contract, cancellation, warranty, and guarantee terms?
  • Handoff: What information is collected, and what happens next?

Now search your site for each answer. Mark anything missing, conflicting, or older than one year. Assign an employee who can approve the correct version. This small exercise often exposes sales problems that existed long before AI ads arrived.

Decide what the agent should qualify

More leads aren’t automatically better. Your sales team needs enough information to separate an urgent, profitable opportunity from a poor fit.

Choose three to five qualification points that materially affect whether you can help. A commercial contractor might need location, building type, project scope, target completion date, and budget status. A law firm may need practice area, jurisdiction, deadline, and conflict information, while taking special care not to invite confidential facts before an engagement exists.

Avoid turning the conversation into an interrogation. Ask only for information your team will use. Don’t request sensitive personal data merely because a form allows it.

Google’s broader guidance for lead generation recommends using first-party conversion data and enhanced conversions for leads to help its systems distinguish valuable leads from form fills. That means your preparation should include both the questions you ask and the outcome data you send back.

Create a measurement plan before launch

If the format appears in your account, don’t judge it by lead count alone. Compare it with your existing Search campaign and lead form process.

Track these numbers:

  • Conversation starts, completed leads, and the conversion rate between them
  • Cost per lead, cost per sales-qualified lead, and cost per closed customer
  • Invalid, duplicate, outside-area, and otherwise unqualified leads
  • Time from submission to first human response
  • Revenue or estimated pipeline value by lead source

Use a dedicated source label in your CRM. Preserve the Google click identifier when your setup supports it. Send qualified lead and closed-sale outcomes back through your approved Google Ads conversion setup. Google explicitly recommends importing high-value offline outcomes such as qualified leads or closed deals in its Performance Max lead generation practices, and the same business logic applies when evaluating a new lead format.

Run the test long enough to include your normal sales cycle. A campaign can look efficient after two weeks because it generated cheap conversations, then look very different when none of those prospects become customers.

Put guardrails around answers and handoffs

An AI answer isn’t a signed proposal. Your pages and sales process should make that clear.

Identify claims that require a human confirmation, including final pricing, availability, legal advice, financing approval, medical guidance, compatibility, and project deadlines. Put accurate qualifiers on the relevant pages. Keep regulated or high-risk information under review by the right professional.

Create an escalation rule for questions the published content cannot answer. The handoff should collect contact details, summarize the unresolved question, and tell the prospect when a person will reply. Test that notification path after every integration change.

Assign a website content owner, too. When prices, territories, staff, policies, or service packages change, updating the website should be part of the operational checklist, not a marketing task someone remembers months later.

A practical 30-day preparation plan

During week one, export the last six months of sales questions from call notes, emails, chat transcripts, and proposals. Remove customer details and group repeated questions.

In week two, audit the seven page types above. Fix contradictions first. Then add missing cost, timing, fit, and process answers.

In week three, define qualification criteria with sales. Set CRM source fields, lead stages, and offline conversion reporting. Test the path from ad or form submission through the first human follow-up.

During week four, have someone outside the company read the pages and try to answer the 30-question inventory. Anything they cannot answer clearly needs another edit. Then ask your Google Ads representative or account manager about eligibility and current regional availability.

Don’t wait for the ad format to fix the website

Business Agent for Leads may take time to reach your account. The preparation still pays off now. Clear service, pricing, process, location, and policy information helps human visitors decide whether to contact you. It also gives your sales team one approved place to point prospects for routine answers.

The businesses most prepared for conversational ads won’t be the ones with the fanciest chatbot. They’ll be the ones whose websites tell the truth clearly, whose sales teams agree on what makes a good lead, and whose tracking connects inquiries to revenue.

If your website cannot answer the questions your prospects ask before buying, talk with Your Web Team about fixing the gaps.