Google's AI Overviews went live for searchers across the United States this week, and the first clicks they take will be the easy ones. Definitions. Quick facts. The simple how-tos and "what is" posts agencies have produced for small-business clients by the dozen. Those pages were always the weakest part of a client's reach, because the person reading them rarely needed the client at all. They wanted an answer, got it and left.
So this isn't a story about search ending. It's a story about which traffic was ever worth much. Agencies that rebuild client content around searches that need a visit (bookings, quotes, tools, local intent) and that turn the visitors they still earn into email subscribers will finish the year with clients who rely less on any one results layout. The fix sits inside the work you already do.
This piece gives you a way to sort client pages by risk, a reporting approach for this month's calls, and a plan for where content effort goes next. It's written on Thursday 16 May 2024, two days after the announcement. Where the data doesn't exist yet, we say so plainly.

AI Overviews reached every US searcher this week, and the clicks most at risk are the ones agencies have been selling
On Tuesday 14 May, at Google I/O, Google announced that AI Overviews is rolling out to all users in the US, with more countries to follow later in 2024. It's the feature Google spent the past year testing as Search Generative Experience inside Search Labs, now switched on by default rather than as an opt-in experiment.
An AI Overview is a written answer generated by Google's model and placed above the usual list of results. It draws on several web pages at once and shows links to some of them next to or beneath the answer, so the searcher reads a finished reply before they see a single blue link.
You'll hear plenty of forecasts about the ai search summaries impact on website traffic this month, from mild to apocalyptic. Our view is narrower and more useful. The searches most exposed are the ones where someone wanted a fact and had no reason to stay once they had it. Agencies have been selling those pages as blog volume for years, often measured by sessions rather than by enquiries. Those sessions were the thinnest part of the report. They're now the part most likely to shrink.
Overviews won't sit on top of every search either. In the Search Labs test they appeared on some queries and not others, and Google hasn't published how often they'll show now that they're live for everyone. Searches with a clear commercial or local purpose behaved differently in testing from broad questions, and that difference is where your planning starts.
An answer that fits in a summary was never going to earn a visit for long
Here's a quick test for any client page. Could a searcher get everything they need from three or four sentences on the results page? If yes, that page is exposed. If they have to do something on the client's site, such as book, buy, request a quote, check availability or use a calculator, a summary can describe the step but can't take it for them.
That test sorts a typical small-business site into clear groups. The table below uses examples from the clients agencies tend to carry, so you can map it straight onto your own accounts.
| Query type | Example for a small-business client | Likely exposure | What to do with it |
|---|---|---|---|
| Definitions | "what is a dental crown" for a dental practice | High | Keep the page, add an email capture offer, stop writing more of them |
| Quick facts | "how long does a boiler service take" for a heating engineer | High | Point the answer to a booking page rather than ending on the fact |
| Simple how-tos | "how to bleed a radiator" for a plumber | High | Add a clear "call us if this doesn't work" route and a capture form |
| Thin comparison lists | "best flooring for a kitchen" for a flooring shop | Medium to high | Add real stock, prices and a showroom visit link, or merge into a stronger page |
| Original data or opinion | "what 120 kitchen refits cost us last year" for a builder | Medium | Keep producing, since a summary can quote it but can't replace it |
| Action searches | "emergency dentist appointment today" | Low | Build fast, clear booking and quote pages |
| Tools and calculators | "loft insulation cost calculator" | Low | Build one useful tool per client where the trade suits it |
| Local intent | "plumber near me", "dentist in Austin" | Low | Strengthen service-area pages, reviews and business listings |
| Brand searches | "Harper Dental reviews" | Low | Make sure the client's own pages answer these well |
Should you stop producing blog content for clients? No. Stop producing the kind whose only job is to answer a question someone could have got anywhere. A post built on the client's own jobs, prices, photos or opinions still earns its place. A 600-word explainer rewritten from the first page of results doesn't, and frankly it was struggling before this week.
Clients will see impressions hold while clicks slip, and that report needs explaining before they ask
Picture the search console view for a client with a busy blog over the coming weeks. Rankings look steady. Impressions hold or even rise, because pages still appear on the results screen. Click-through falls on the informational pages, since the answer now sits above them. A client who only sees the site-wide click total will think something broke. Nothing did, but you'll want to say that first.
The honest answer to "how much will we lose, and when?" is that nobody can say yet. Data from the first week is thin. Nobody outside Google knows how often Overviews will appear across different types of search, and there are no independent studies of the live rollout to point to. Anyone quoting a percentage this week is guessing. Compare like-for-like weeks over the next month or two before drawing any conclusions about the ai search summaries impact on website traffic for a given client.
What you can change today is how you report. Site-wide totals blur everything together, so split clicks into intent groups and report each one separately.
- Informational: definitions, facts and how-tos, where you expect pressure
- Commercial: comparisons, pricing and product or service pages
- Action: booking, quote, contact and tool pages
- Local: service-area pages and "near me" searches
- Brand: searches that include the client's name
Then move the headline of the report away from traffic. Leads, booked appointments, quote requests and calls are what the client pays for. If informational clicks fall by a fifth while bookings hold, that's a good month, and the report should read like one.
Move client content toward searches that end in an action on the client's site
The repositioning is practical, not theoretical. Effort shifts from pages that answer questions to pages that let someone do something. That means booking pages that load fast and work on a phone, quote forms that ask for the job details a tradesperson needs, and interactive tools such as cost calculators or availability checkers. It also means pricing explainers that walk through what drives the cost and then send the reader to a pricing page, local service pages for each town or district the client covers, and original data and opinion that a summary can quote but can't stand in for.
Take Dan, who runs a three-van plumbing firm and has been on a monthly content retainer for two years. His blog has forty how-to posts, from unblocking sinks to fixing dripping taps. They bring in a steady stream of visitors, very few of whom ever ring him, because they came to fix the problem themselves. His service-area pages are a single paragraph each, and his booking form sits three clicks deep.
The new plan for Dan's account stops new how-to posts entirely for a quarter. The hours go into eight proper service-area pages, one per district he covers, each with real job photos, typical response times, the areas within them he serves and a booking button above the fold. The quote form moves to every page and asks for a photo of the problem. One tool gets built, a simple guide to what an emergency call-out costs at different times of day. The old how-to posts stay live, each one ending with a line on when to call a plumber and a link to book.
A dental practice follows the same pattern. Fewer posts on what a filling is, more effort on the new-patient booking page, an emergency appointments page, a treatment pricing explainer and a page for each neighbourhood the practice draws from. We've covered how generative search treats local business visibility in more detail, and the short version is that a business with clear, specific local pages is in a far stronger position than one relying on general advice posts.
Being the source an Overview cites is worth chasing, but it is not a traffic plan
Can you get a client cited in an AI Overview? Sometimes, and the work that improves the odds is the same work that makes a page good. Use clear structure with headings that state what each section answers. Give a direct answer near the top rather than three paragraphs of warm-up. Name the author, with a real person and real credentials, so a dentist's page is plainly written by a dentist. Add original material, such as the client's own prices, case photos, local knowledge or survey results, that gives Google a reason to point to this page instead of the next one.
Clients will ask whether a citation is worth anything. It is, as visibility. The client's name and link sit beside an answer for the exact thing their customers search, and that builds familiarity over time. If you're weighing how much of this to sell, our piece on how generative engine optimization differs from traditional SEO sets out where the two overlap and where they don't.
Here's the limit the advice pieces tend to skip. A citation link sitting beneath a complete answer gets clicked far less than a top organic result used to, because the searcher already has what they came for. There are no reliable figures yet on how much less, and you shouldn't invent one for a client deck. Treat citations as brand exposure, report them that way, and never present them as replacement traffic. The traffic that replaces lost informational clicks comes from action pages, local pages and people who've already given the client their email address.
Every visitor a client still earns should get a reason to join an email list
Google decides how a results page looks, and it changed that layout for every US searcher this week. An email list works differently. Once a customer gives a client their address, reaching them again doesn't depend on which box sits above which link. So every page that still earns visitors should end with a clear reason to subscribe, and that matters most on the informational pages under pressure.
The offer matters more than the form. "Sign up to our newsletter" gives nobody a reason to type anything. Small businesses do better with something tied to what the customer will need next. A dental practice offers check-up and hygiene reminders. A heating engineer offers a nudge before the annual boiler service and first notice of autumn booking slots. A plumber offers a one-page checklist for protecting pipes before winter. A garden centre offers a month-by-month planting calendar. Each one is useful on its own, and each one leads straight back to a booking.
Put the forms where readers are about to leave. That's the end of a how-to, just after the direct answer, or beside the step where someone gives up and decides to call a professional. Those visitors came for a fact. The form gives them a reason to come back. Dan's forty how-to posts from the last section are exactly where his boiler-reminder form belongs.
How fast can a small business build a list? It depends on traffic and on how good the offer is, and we won't put a number on it. A site with thin traffic grows a list slowly. So add capture points away from search too, such as booking confirmations, invoices, the counter and the van. Every one of those is a customer the client has already served.
Keep the list clean from the first signup. Bot-protected forms stop junk entries filling the list with fake addresses that hurt sending later, and our guide to protecting lead capture forms covers the setup. Optional double opt-in asks each new subscriber to confirm by email. You'll get fewer signups, and every one you keep is a real person who said yes, with a consent record that holds up under GDPR. Our view is to switch it on for any form that's already attracting junk.
A list only protects reach if the mail lands
Three basics decide whether a small client's emails reach the inbox. Addresses are verified before sending, so dead and mistyped ones never get mailed. New client domains are warmed up gradually rather than pushed into a full-list send on day one. And campaign volume is spread across several brand-matched sending domains, so no single domain carries the whole load. SendTo handles all three for agencies automatically. If you're wondering whether the checks are worth the bother, our piece on list verification sets out what goes wrong without it.
Agencies that sell email alongside SEO can turn a traffic scare into a steadier retainer
Your clients will ask what you're doing about AI search. Some will ask this month. "We're monitoring it" won't hold for long. A combined plan answers the question properly. Content effort moves to action and local intent, capture forms go on the pages that still rank, and a monthly email programme turns those subscribers into repeat bookings. Each part supports the others, and the client can see where the leads come from.
SendTo lets you run that email programme as your own product. Clients log into a portal with your brand on it, and you set their monthly prices yourself. You pay SendTo a flat partner fee, and clients buy their own sending usage on top of the subscription you charge. Inside the platform you build campaigns, landing pages and automated sequences, and you clean lists before sending. An AI reviewer checks subject lines, structure, links and basic compliance before launch. Campaign designs come out as finished HTML emails from a short brief and the client's brand settings in seconds, so you can show a client a real campaign on the same call where you explain the plan. Reports leave out opens triggered by security scanners, so the open and click figures in the client portal reflect real people. For current partner fees, see the pricing page.
Be straight with clients about what email doesn't do. A new list starts small, so it won't replace lost informational clicks in the first month. It won't rescue a site with no traffic and no customers to invite. And it only pays off if someone writes a decent offer and sends on schedule, which is the work you're selling. Sell it as the reach that keeps building while search layouts shift, not as an instant fix.
What to do with client accounts before the end of May
Everything below fits into two weeks for a typical agency client base. Start with the accounts that have the biggest blogs, because they have the most informational traffic exposed. If you want a fuller method for the first step, our guide to keyword intent mapping walks through it.
- Tag each client's top 50 pages by intent group: informational, commercial, action, local or brand.
- Flag the informational pages carrying the largest share of each client's organic clicks. These are the pages to watch and the first to get capture forms.
- Add a specific capture offer to each flagged page, such as a reminder, a seasonal offer or a checklist, with bot protection switched on.
- Build or refresh booking, quote and service-area pages, starting with the client's busiest service and best-performing area.
- Change your report template so clicks are split by intent group and leads, bookings and calls sit at the top.
- Send every client a short written briefing this week explaining AI Overviews, which of their pages are exposed and how you'll judge results.
- Start list growth now so each client has a warmed domain and a growing list before the autumn peak.
The timing on that last point is real. Public interest in email marketing is lowest in June and peaks in November, December and January. A list started in May has the quiet summer months to grow and settle, and it's ready when customers are shopping and booking. Start in October and you're warming a domain and building a list at the same moment you need it to perform.
Creators and consultants writing about this shift can recommend the email side and get paid for it
If you write for agencies, whether as a blogger, a freelance consultant or an affiliate, AI Overviews is the subject your readers want covered this month. The practical answer to it includes owned email, and you can recommend SendTo as the platform that lets an agency sell that email programme under its own brand.
Here are the terms. You earn 20% of the platform fee, the partner subscription, that each agency you refer pays SendTo, every month that agency keeps paying it. You never earn a share of what that agency's own customers pay the agency. Attribution lasts 90 days from the first click, and the first touch wins. Commission is paid monthly by PayPal.
Write about AI search for agencies and earn 20% of the partner subscription each agency you refer pays SendTo, every month it pays.
Join the SendTo affiliate programme


