01 The market data
The AI-referred visitor shows up ready to buy
Shopify’s second-quarter data: AI-referred sessions tripled, and in research-heavy categories they convert at double the organic rate.
Shopify published a second cut of its commerce data on August 11, and the growth held. AI-referred sessions to merchant storefronts grew 197% year over year in the second quarter, with orders roughly tripling alongside. Organic search still sent more traffic than every tracked AI platform combined, and grew 12% itself.
The sharper finding is what those visitors do. In specification-heavy categories, AI-referred shoppers converted at about twice the rate of organic visitors, and half of all AI-referred sessions landed directly on a product page. The shopper described the need, weighed the options, and settled the choice inside the conversation.
That visitor is a diner who read the menu and the reviews before walking in. The persuading your website used to do is finished before the page loads.
One more number: when the AI system drew on structured catalog data rather than scraped feeds, the shoppers it sent converted at twice the rate. The honest caveat is that this is Shopify’s own platform data, with no disclosed merchant count. Trust the ratios more than the totals.
What to do about it
Your practice-area page is the product page in this story. A visitor from an AI conversation arrives with the comparison already made, so the page’s job is confirmation: say plainly who the service is for and what the first conversation involves.
The structured-data finding carries over. Schema markup describing your firm, attorneys, and frequently asked questions is the machine-readable layer that doubled conversion for merchants.
Source
Shopify, Kyle Risley, “AI and organic search are doing different jobs: What Shopify’s data shows,” August 11, 2026 ·
shopify.com
· Secondary: Search Engine Land’s August 13 coverage
02 The research
The AI Overview took blame it had not earned
A panel linking millions of real searches to real AI conversations finds most of the missing clicks were never there to lose.
Scrunch, an AI visibility firm, published a study on August 13 built on millions of search events from an opt-in panel that links people’s AI conversations to the same people’s web browsing, February through June 2026. The surface numbers confirm the fear: news searches showing an AI Overview, the machine-written summary at the top of Google results, sent readers to a publisher about 20% of the time, against roughly 30% without one.
Then the authors held the search itself fixed. Comparing the same query with and without the overview, the gap shrank to about 2 percentage points, too small to distinguish from zero. The reason: Google mostly places AI Overviews on searches that produced few clicks to begin with. Weather queries drew one 59% of the time; sports scores, the strongest click driver, 5%.
Blaming the overview for the missing clicks turns out to be like blaming the umbrella for the drought. It stands over ground that was mostly dry already.
The conversational side ran the other way. After a news conversation with an AI assistant, readers were 20.5 percentage points more likely to visit a news site that week, yet only 1.1% of those visits carried an AI referrer; about three quarters registered as direct traffic. One finding for smaller brands: mid-sized publishers were named in 17% of AI answers but captured almost none of the follow-through visits; 97% went to household names. Being named is not the same as being chosen.
What to do about it
Check which of your own queries trigger an AI Overview. Search the questions clients bring you and note where the summary appears; the harm is decided query by query, not by headline.
The recognition gap is the lesson. A mention converts when the reader already trusts the name, so track share of voice weekly against named competitors and keep building the brand beyond AI.
Source
Scrunch, Michael Iannelli, “Surviving the clickpocalypse: What millions of search events say about AI eating the news,” August 13, 2026 ·
scrunch.com
· Secondary: Search Engine Land’s summary of the findings
03 The platform
Google is illustrating its answers with machine art
AI-generated images are now appearing inside AI Overviews, and readers may assume the cited websites drew them.
Google began showing AI-generated images inside AI Overviews this week, Search Engine Roundtable reported on August 14, a month after the feature was announced in Google’s Images anniversary post. The first sightings are on recipe searches.
The recipe site that caught it, Inspired Taste, raised two objections: the cartoons flatten away the safety and technique detail its tested photography carries, and a searcher might assume one of the cited websites produced the illustrations.
The answer page has hired its own sketch artist. Your name stays on the record beneath the picture, and the picture is the machine’s guess.
For a firm the risk is a step removed but worth watching. A generated diagram simplifying a legal process could misstate steps your cited page gets right, and a reader has no way to tell whose work is whose. Machine-made images also carry none of the compliance review your own materials go through.
What to do about it
Run your core practice questions through Google and note where AI Overviews now carry images. If your pages are cited beside generated art, screenshot it. A dated record of what the answer showed is cheap now and hard to reconstruct later.
Keep publishing your own photography and diagrams with clear captions and image markup, so the record shows what you did and did not produce.
Source
Search Engine Roundtable, Barry Schwartz, “Google AI Overviews With AI-Generated Images In The Wild,” August 14, 2026 ·
seroundtable.com
· Secondary: Google’s original announcement
04 The deadline
The language dial comes off in September
Google Ads will remove manual language targeting from Search campaigns and decide for itself who understands your ad.
Google updated its language targeting documentation this week: starting in September 2026, Search campaigns lose the campaign-level language setting. Ads will match automatically based on the language of the ad itself, judged against the searcher’s own language signals. The change also covers the Search portion of Performance Max, Google’s automated cross-channel campaign type, while YouTube, Display, Discover, and Gmail placements keep the manual control.
Google says the goal is removing redundancy and that no advertiser action is required. Two details are worth keeping anyway: campaigns built one per language keep working exactly as before, and language exclusions do not exist today and will not exist after the change.
The thermostat is coming off the wall. The room still has a temperature; you just stop setting it. What replaces the dial is the wording of your ads and landing pages: the only language signal left under your control.
For a firm serving a multilingual county, the question is whether English and Spanish ads are cleanly separated. If they share one campaign, the system now decides which a given searcher sees.
Before September
Audit each Search campaign so every ad and its landing page sit consistently in one intended language. Separate campaigns per language remain the reliable structure; mixed ad groups get sorted by the machine from September.
State bar advertising rules apply to this new mechanism exactly as they did to the old one, and they apply in every language your ads run. Translated ads go through the same compliance review as the English originals.
Source
Google Ads Help, “About language targeting,” updated August 2026 ·
support.google.com
· Secondary: Search Engine Roundtable, August 14
05 The money
Google booked the moving truck for your campaigns
The AI Max migration now has hard dates: legacy broad match and auto-created assets move September 1, Dynamic Search Ads follow in February.
On Wednesday we covered Google auto-upgrading some Search campaigns to AI Max, its bundle of AI-driven matching and creative tools. The developer team has now published the full timetable. Creating the legacy setups was blocked on August 3, and the automatic migration runs September 1 through 30.
The mechanics are gentler than the headline. Campaigns migrate in place with equivalent settings, and existing brand inclusions and exclusions carry over automatically. Dynamic Search Ads, where Google writes the headline and picks your landing page, are excluded until February 2027, with in-account warnings from September and a reminder on January 15.
The moving truck is booked either way. The question left is whether you pack the boxes or let the movers guess what is fragile.
Follow the money underneath the logistics. Each migration moves another spending decision, which searches your ad matches and which page receives the click, into Google’s automated layer. The ability to create new Dynamic Search Ad groups disappears permanently in February.
Before September 1
List any campaigns using campaign-level broad match or standalone automatically created assets; those move first. Confirm your brand exclusion lists are current; they are what keeps your ads off competitor-name searches, where advertising rules are least forgiving.
Put February on the calendar for Dynamic Search Ads, and remember state bar advertising rules judge the ad that ran, not the tool that assembled it. Machine-assembled headlines still need your compliance review.
Source
Google Ads Developer Blog, Bob Hancock, “Migrate Campaign-level Broad Match and Automatically Created Assets to AI Max,” August 2026 ·
ads-developers.googleblog.com
· Secondary: Search Engine Land, August 14
