What's Working in Google Ads for Ecommerce in 2026: The Playbook We Run for DTC Brands

· PPC

What's Working in Google Ads for Ecommerce in 2026: The Playbook We Run for DTC Brands

Key takeaways

  • AI Max is now the default for many Search campaigns. Between September 1 and 30, 2026, Google auto-upgraded Search campaigns using campaign-level broad match or automatically created assets to AI Max. Dynamic Search Ads follow in February 2027.

  • The landing page is the biggest lever left. AI Max reads your site to pick landing pages and write headlines. Every page you build for a specific intent becomes a new way to match a query.

  • In Shopping, your feed is your keyword list. Titles and attributes decide which auctions you enter. Images decide whether anyone clicks once you are there.

  • AI Mode is a separate shelf. Shopping ads started showing inside AI Mode on August 31, 2026. Reported research found only 1.28% of products in the standard Shopping carousel also appeared in AI Mode for the same query.

  • Demand Gen is where net-new customers come from at scale. In our Q2 2026 YouTube Ads Report, 98 of 100 YouTube-driven orders left no trace in the click path platforms use for credit.

  • Platform ROAS misleads in both directions. It undercounts YouTube and overcounts branded Search. Judge both on incrementality and new-customer CAC.


What changed in Google Ads in 2026?

Google finished handing targeting and ad copy to AI across Search and Shopping, opened ad inventory inside AI Mode, and gave Demand Gen the measurement it was missing. For ecommerce brands, 6 changes carry real budget consequences. Here is each one and what I would do about it this week.

Change

When

What I would do this week

AI Max auto-upgrade for Search campaigns using campaign-level broad match or automatically created assets

September 1 to 30, 2026

Open every Search campaign's settings. Confirm which AI Max features are on. Add brand and URL controls before spend drifts.

Dynamic Search Ads migrate to AI Max

February 2027

Map your DSA page feeds to AI Max URL inclusions now, while you still have DSA data to compare against.

AI Max for Standard Shopping

Rolling out from mid-2026

Test it on 1 Shopping campaign with a holdout. Keep the rest on your current setup until the test reads.

Shopping ads inside AI Mode

Confirmed August 31, 2026

Fill conversational feed attributes like product highlights and product details.

Demand Gen campaign type attribution and view-through conversion optimization

Announced April to May 2026

Turn on the reporting. Run view-through bidding only with a holdout in place.

Merchant Center image minimum of 500 x 500 pixels

January 31, 2027

Audit image sizes across the catalog now. Google recommends 1500 x 1500.

The thread through all 6: Google now reads your feed, your site and your creative to decide where you show up. The accounts scaling this year feed it better inputs. The rest of this playbook is about those inputs.

Why do landing pages decide how far your Google Ads can scale?

The ad earns the click. The page decides whether that click pays back. The default setup sends every Google campaign to a product page or a collection page. That works for shoppers already ready to buy. It loses money on everyone else, including the shopper comparing 3 brands and the YouTube viewer who had never heard of you 30 seconds ago.

Pages built for 1 intent change the math. They let you bid profitably on comparison queries, problem queries and cold video traffic that a product page cannot convert.

2026 raised the stakes. AI Max final URL expansion chooses the landing page for each query from your site, and text customization writes headlines from your page content. A site with 1 product page per SKU gives the system very little to work with. A site with a comparison page, a buying guide and an advertorial gives it 3 more ways to match a query.

Every page has to match 3 things:

  • The query or creative that sent the click.If someone searched "best running shoes for flat feet," the headline answers that exact question.

  • The awareness level. Cold traffic needs education before it sees an offer.

  • The angle in the ad. If the ad sold durability, the page opens on durability.

Across our accounts, intent-matched pages converted cold Search traffic at [X]% against [X]% on the product page over [timeframe].

The trade-off: every page costs design and dev time, and each new page splits conversion data. I would build pages for your top 3 intents first, prove them, then expand. Launching 20 pages at once gives you 20 pages without enough data to judge.

The 7 landing page types we build for Google Ads

Each page type serves a specific traffic temperature and campaign.

Page type

Traffic temperature

Best campaign match

What the page has to do

Advertorial

Cold, problem-aware

Informational Search, Demand Gen, Discover

Teach first. Introduce the product as the fix.

Listicle

Cold to warm

Informational Search, YouTube in-feed, Discover

Give numbered reasons or mistakes. The product closes the list.

Us vs. them comparison

Warm, comparing brands

Competitor and "vs" Search

Show a head-to-head table and say who your product is wrong for.

Roundup

Warm, commercial

"Best [category]" Search, Discover

Rank 4 to 6 options on stated criteria, with ownership disclosed.

Long-form sales page or VSL

Warm to hot

High-intent problem Search, YouTube

Explain the mechanism, stack the proof, present the offer, answer the FAQ.

Quiz funnel

Undecided

Demand Gen, "which [product] is right for me" Search

Ask 4 to 6 questions and land on a personalized product page.

Offer page

Deal-driven

Demand Gen, promotional Search

Put 1 offer above the fold and cut everything else.

How to build an advertorial for Google Ads

An advertorial reads like editorial content and follows a direct-response structure underneath. It works for informational searches like "how to fix [problem]" or "does [solution] work," and for YouTube and Discover traffic.

  1. Open with a headline that mirrors the search or the ad. "Why [audience] still struggle with [problem]" or "The real reason [common fix] stops working after 30 days."

  2. Name the problem in the first 2 sentences. No throat-clearing.

  3. Teach something useful.A study, an expert view, a data point. The page has to be worth reading even if the reader never buys.

  4. Introduce the product as the recommendation. Explain why it solves the problem you just described.

  5. Explain the mechanism. Show how it works and connect it to the education above.

  6. Stack the proof.Reviews, ratings, third-party testing, before and after where compliant.

  7. Close with 1 specific offer. A first-order discount, a bundle, free shipping or a guarantee.

How to build a comparison page

Comparison pages catch shoppers searching "[competitor] alternative," "[you] vs [competitor]" and "products like [competitor]." These shoppers are close to buying and still deciding who gets the order.

  1. Hero with both products side by side. "[Your product] vs. [Competitor]: an honest comparison."

  2. Say why the page exists. "Customers ask us how we compare to [Competitor], so here is a side-by-side."

  3. Build the comparison table. Materials or ingredients, price per use, reviews, guarantee, shipping.

  4. Add a "who this is wrong for" section. Admitting where the competitor wins makes your other claims believable.

  5. Show switcher proof. Reviews from customers who came from the competitor.

  6. Close with the offer and an FAQ.

Guardrail: every claim about a competitor needs a source you can produce. Google's misrepresentation policy and US false advertising law both apply.

How to build a roundup page

A roundup ranks the top products in your category like an independent buying guide. It fits "best [category]," "top [product] for [use case]" and "[product type] reviews."

  1. Headline that matches the query. "The 7 best [product type] for [audience] in 2026."

  2. Explain how you ranked.Criteria first, so the ranking reads as earned.

  3. Put your product at #1 with the reason.

  4. List 4 to 6 competitors with real strengths and real weaknesses.

  5. End with a comparison chart.

Guardrail: disclose at the top that the page is brand-owned. The FTC treats undisclosed brand-owned reviews as deceptive, and Google will disapprove ads that misrepresent who is behind the page.

How do you get more coverage in Google Shopping?

Coverage means 2 things: showing up for every relevant, profitable search, and ranking in the top positions once you do. Shopping has no keywords. Google matches queries to your product data, so the feed is your targeting. A feed set up once and left alone loses coverage to competitors who keep improving theirs.

Step 1: Rewrite titles from your search terms report

Pull 30 to 90 days of search terms from Shopping and Performance Max. Find the queries that convert. Put their language in your titles.

Titles can run 150 characters, and most placements truncate them. The words that matter go first: product type, then the attribute people search, then brand if people search your brand name.

  • Weak: BrandName Premium Office Chair, Black, Model X3

  • Strong: Ergonomic Office Chair with Adjustable Lumbar Support, Breathable Mesh, Black | BrandName

The strong title matches "ergonomic office chair," "lumbar support chair," "mesh office chair" and "black office chair." The weak one matches almost nothing a shopper types.

Step 2: Fill every attribute Google uses to match and filter

AI Mode and AI Max for Shopping pull from these fields to write conversational ad copy and match natural-language queries. A blank attribute is a query you cannot match.

Attribute

What to do

title

Rebuild from converting search terms. Product type and key attribute first.

description

Reinforce the title keywords. Answer the questions buyers ask. Lower priority than title, still worth keeping clean.

product_type

Use your own category path up to 5 levels deep. It powers your campaign segmentation.

google_product_category

Set it manually. Google's auto-assignment is often wrong.

gtin, mpn, brand

Submit real values. They connect your listing to product reviews and the right catalog entry.

identifier_exists

Set to "no" only for custom or handmade products with no barcode.

color, size, material, pattern, age_group, gender

Fill every one that applies. These drive filters and variant matching.

product_highlight, product_detail

Add the specific benefits and specs people ask about. These feed conversational surfaces.

additional_image_link

Up to 10 images: angles, detail shots, scale.

lifestyle_image_link

Up to 5 in-context images: the product in use or on a person.

sale_price

Only when your price history supports the discount. Google suppresses sale badges it cannot verify.

availability

Keep it accurate. Out-of-stock clicks waste budget and trigger disapprovals.

Step 3: Fix everything Merchant Center flags

Merchant Center grades your store on shipping, returns and site experience signals. Open the store quality view and fix every flag it lists. Then turn on product ratings, seller ratings and promotions. All 3 show on your listings and lift click-through rate against listings without them.

Step 4: If you sell in the EU, use a third-party CSS

Every Shopping ad in the EEA, the UK and Switzerland runs through a Comparison Shopping Service. If you never chose one, you are on Google Shopping Europe, which applies a margin of roughly 20% to your bids. A third-party CSS removes it. Agencies report a real-world advantage of 16% to 18%.

It applies to Standard Shopping, the Shopping side of Performance Max and Local Inventory Ads. It does nothing for Search or YouTube. The trade-off: some CSS partners charge a percentage of spend that eats most of the saving. Read the fee schedule before you switch.

Step 5: Duplicate your top products by intent

A single product usually serves several buyers. Take a running shoe:

Search

Buyer

running shoes for flat feet

Runner managing overpronation

wide running shoes

Runner who cannot fit standard widths

marathon training shoes

Runner building mileage

running shoes for beginners

New runner who wants comfort over speed

The usual move is 1 title built for the highest-volume term. The listing still shows for the other searches, but Google reads it as a partial match. Lower relevance means a lower position, a higher CPC or both. A competitor with a listing built for "wide running shoes" will outrank you there.

The fix is product duplication. You create multiple feed entries for the same product, each built around 1 intent, each with its own title, description and landing page. Google treats them as separate listings, and each can rank for the search it was built for.

  1. Start with your top SKUs. Pick products with search volume that serve more than 1 type of buyer.

  2. Map 3 to 4 angles. Use case ([product] + use case). Demographic ([product] + audience). Benefit ([product] + outcome).

  3. Build each variation in a supplemental feed. Same product and price, new item ID, title and description built around 1 angle.

  4. Align the landing page.The title, the description and the page should tell the same story.

  5. Separate the variations. Put them in their own campaign or product group so you control budget and traffic.

  6. Track and cut. Watch impressions, CTR, conversion rate, ROAS and search term quality. Move budget to the variations that earn it.

Guardrails: every variation must be the same product at the same price with a compliant landing page. Duplicates compete in the same auctions, so start with 3 to 5 variations on your strongest SKUs and expand once the data says which angles work. Watch Merchant Center diagnostics after launch.

AI Max for Shopping's text customization is Google's automated version of this play. It rewrites titles and copy per query from your feed. I would run manual duplication on the top 5 SKUs where you already know the buyer intents, and test AI Max for Shopping on the long tail against a holdout campaign before rolling it out account-wide.

Step 6: Treat AI Mode as its own shelf

Google confirmed on August 31, 2026 that Shopping ads now appear in a carousel inside AI Mode responses. Research reported in September found AI Mode surfaces 3.9 products per query against 27.8 in standard search, only 1.28% of products overlap between the 2 on the same query, and AI Mode prices average 21.6% higher.

Eligible campaign types include Performance Max, AI Max with search term matching, Shopping and broad match. To compete there, complete your product highlights and product details, write descriptions that answer buyer questions, and confirm AI Max for Shopping is on for the campaigns you want eligible. I would hold judgment on AI Mode performance until you have 60 days of data. Reporting on these surfaces is still thin.

How do you use Shopping ads to reach new customers?

Standard Shopping sends every click, cold or warm, to the same product page behind the same plain product image every competitor uses. For prospecting traffic, change 2 things: the image and the page.

Change the image

Search almost any category on Google Shopping and the images look identical. Same background, same angle. When nothing stands out, shoppers decide on price, reviews and small details, which is a hard fight unless you are the cheapest.

Google allows more than the typical feed uses:

  • The main image_link has to show the product accurately, with no promotional overlays or watermarks.

  • lifestyle_image_link takes up to 5 in-context images.

  • additional_image_link takes up to 10 more.

  • AI-generated images must keep their IPTC metadata tag. Stripping it breaks policy.

Within those rules, test main images that show the product with packaging, the product in use, the product with its key components, or a close-up of texture and detail. Run the tests through a supplemental feed and compare CTR after 2 to 4 weeks.

Change the page

The common assumption is that Shopping traffic has to land on a standard product page. Merchant Center's actual requirement is narrower. The landing page must show the product's title, image, price, currency, availability and a buy button that match your feed, and the advertised product has to be the primary focus. Category pages and search pages are not allowed.

That leaves room. A long-form product page keeps a compliant buy box at the top and adds what cold traffic needs below it: an advertorial section, a comparison table, a "how it works" block, reviews and a pre-sell. The shopper who was not ready to buy gets a reason to believe before you ask for money.

Quiz pages and pure advertorials without a buy box will get your products disapproved in Shopping. Keep those for Search and Demand Gen.

Test it without breaking what already works

You cannot split 1 Shopping listing across 2 URLs, and replacing your product page risks the revenue it already drives from organic and Shopping. Use duplication instead. Keep the original listing on the existing page. Create a duplicate that points to the long-form page. Compare conversion rate and CPA over 4 weeks.

The trade-off: the 2 listings share auctions, so run the test on a product with enough volume to reach a clear result in a month.

How does full-funnel Search find new customers?

A typical DTC Search account runs branded and product terms and stops there. Search can reach a buyer at every stage: while they are researching a problem, while they compare options and when they are ready to buy. Match the landing page to the intent, and top-of-funnel queries become profitable because they draw fewer bidders and cheaper clicks.

Intent

Example query patterns

Landing page

What to watch

Branded

"[brand]", "[brand] + category", "[brand] + product"

Homepage, collection page or product page, matched to the query

Incrementality. Most branded conversions for mature brands would have happened anyway.

Transactional

"[product] sale", "buy [product]", "[product] free shipping"

Product or collection page. Test a long-form product page in crowded categories.

CPA against contribution margin

Comparison

"best [category]", "[competitor] alternative", "[you] vs [competitor]"

Roundup or comparison page

Longer conversion lag than transactional

Informational

"how to fix [problem]", "does [solution] work", "[ingredient] benefits"

Advertorial or listicle

Judge on a 30-day window or longer

Branded Search: check incrementality before you scale it

Branded Search reports the best ROAS in most accounts and deserves the most skepticism. In our analysis of branded search lift, results typically land between 10% and 50%, which means 67% to 91% of attributed branded conversions would have happened without the ad. The trade-off in cutting it: competitors can bid on your name and take the top position. Run a geo holdout before you cut or scale branded spend.

Where to find the queries worth building for

  • Search terms report, last 30 to 90 days. New converting queries, high-ROAS queries, long-tail searches you are not targeting, competitor queries, and the terms AI Max and Performance Max found for you.

  • Google Keyword Planner. Seed it with your product, category and problem terms. Filter results into category, comparison and informational buckets.

  • Google Search Console. The queries your site already earns organically show demand you can buy into.

  • Google Trends. Catch rising terms before they show volume in keyword tools.

  • AI for angles. Give it your product, customer, objections and use cases, and ask for queries by awareness stage. Then check every idea against real search volume. AI drafts the list. Search volume decides what you build.

Why intent-matched pages lower your CPC

Quality Score has 3 components: expected CTR, ad relevance and landing page experience. A page built for the query improves all 3. Ad Rank weighs quality alongside your bid, so a higher Quality Score buys the same position for less.

What is AI Max for Search, and how should you set it up?

AI Max is a set of AI features layered onto existing Search campaigns. Google describes it as an optimization layer on top of the campaigns you already run. It has 3 parts. Search term matching expands your reach with broad match and keywordless matching based on your landing pages and assets. Text customization writes headlines and descriptions from your site and existing ads. Final URL expansion sends each click to the page it judges most relevant.

Google reports that campaigns using the full feature suite see 7% more conversions or conversion value at a similar CPA or ROAS, compared to search term matching alone. That is Google's number measured on Google's attribution. Test it on yours.

AI Max also matters beyond the standard results page. AI Max campaigns are eligible for ad placements in AI Overviews and AI Mode. As of September 9, 2026, exact and phrase match keywords in standard Search campaigns can also serve in AI Mode for direct-intent queries.

Where AI Max earns its place

  • Keyword discovery. AI Max surfaces searches you would never have written down. Search terms it finds carry an "AI Max" label in the search terms report. Promote the converters to exact and phrase match in your core campaigns.

  • Sites with deep content. The more strong pages you have, the more AI Max has to match against. This is where the landing page work pays twice.

  • Top-of-funnel angles. When AI Max sends a problem query to a blog post and it converts, build a dedicated advertorial on that angle and fund it in Search, Shopping and Demand Gen.

Where AI Max goes wrong

  • It matches low-intent or irrelevant queries.

  • It sends traffic to pages that do not sell, like blog posts with no path to buy, policy pages and out-of-stock product pages.

  • It writes copy you did not approve. For supplements, skincare and health brands this is a compliance problem, and I would keep text customization off until someone has reviewed what it produces.

How we set up AI Max

  1. Start on a campaign with steady conversion volume. [30] or more conversions in the last 30 days is my floor. Below that, the system is guessing.

  2. Set brand controls. Brand inclusions keep branded and competitor campaigns tight. Brand exclusions stop non-brand campaigns from bidding on your own name.

  3. Set URL inclusions and exclusions. Exclude careers, policies and content with no path to purchase. Include the intent pages you built.

  4. Use locations of interest at the ad group level when you only sell or ship to specific markets.

  5. Write 7 to 8 strong headlines yourself. Text customization builds from what you give it.

  6. Launch as an experiment.Since September 2026, AI Max experiments support multi-campaign tests and keep your brand and location controls in place.

  7. Review weekly. Read search terms by match source and add negatives. Check the landing page report and pull any page that should not get paid traffic. Check the asset report and remove inaccurate or off-brand copy.

Low-volume searches get grouped in reporting, so read the trend across weeks rather than reacting to every row.

Why is Demand Gen the most scalable campaign type on Google?

Demand Gen is Google's visual campaign type for finding new customers. It runs on YouTube in-stream, in-feed and Shorts, plus Discover, Gmail and, since 2026, Google Maps. Channel controls let you pick placements. It reaches people before they search, which is the only way to grow past the demand that already exists for your category.

YouTube still gets filed under awareness. Our data points the other way.

What our Q2 2026 YouTube Ads Report found

We built the report with Stella, Northbeam and Fairing. It combines 92 incrementality studies, 45.8M post-purchase survey responses and more than $300M in tracked ad spend.

  • Median incremental ROAS was 2x.

  • 98 of 100 YouTube-driven orders were invisible to click attribution. Of 1.7M orders where buyers said YouTube got them to buy, 98 of every 100 left no trace in the click path platforms use for credit.

  • Across our own portfolio, YouTube spend grew 190% in Q2. New customer rate held at 99%, and basket sizes ran 14% above Meta.

The misses are in the data too. The low end of incremental ROAS was 0.17x, and 14% of tests never reached significance. Run a holdout before you scale it.

What Google shipped for Demand Gen in 2026

Google's own figures, useful as direction. Your holdout decides.

  • Creator assets:adding creator content to Demand Gen raised conversion lift by 20% on average.

  • Product feeds: Demand Gen campaigns with large catalogs saw 33% more conversions after adding feeds.

  • Campaign type attribution: Demand Gen gets its own conversion view instead of being deduplicated against Search and Performance Max.

  • View-through conversion optimization: Demand Gen can now bid toward view-through conversions on YouTube.

  • New customers:Google reports Demand Gen drives an 18% higher share of conversions from new customers than the paid media average.

The placements and what wins on each

Placement

Format

What wins

How we run it

YouTube in-stream

Skippable 16:9 video before or during a video

The first 5 seconds. Name the viewer's problem immediately.

The backbone for scale

YouTube in-feed

Thumbnail and headline in search and browse results

The thumbnail and headline carry the whole ad. Audio does not matter until the click.

Test thumbnails as hard as hooks

YouTube Shorts

9:16 vertical video in the Shorts feed

Native, UGC-style content under 60 seconds

Cut from 16:9 masters. Cheaper CPMs, lower retention.

Discover and Gmail

Image and headline in feeds

Creative that reads like editorial

Pair with editorial landing pages

Google Maps

New in 2026

Local intent near a store

Test only if you have a physical retail footprint

Creative formats that scale on YouTube and Demand Gen

  • UGC.A creator walks through the problem, what they tried, the product and what changed. It works in almost every category and is strongest where the result is visible.

  • Offer-led.The deal is the hook. Give the viewer a reason for the offer, then sell the product so the discount is not doing all the work.

  • Editorial or buying guide. Teach what separates good products from bad ones in the category, then show that yours passes every test.

  • Authority VSL. An expert walks through the problem, the hidden cause, the mechanism and the product. It performs best in skeptical categories like supplements, skincare, health and pet.

  • Stitched. Podcast clips, UGC and product demos cut into 1 narrative. The changes in face and setting hold attention.

Our standing rules for Demand Gen:

  • 16:9 is the backbone.In-stream and in-feed are the powerhouse placements. We cut 9:16 versions from 16:9 masters and brief 16:9 first.

  • Start with proven Meta winners. Recut them for 16:9. It is the fastest way to get tested concepts live on YouTube.

  • I do not use Google's new customer acquisition goal in Demand Gen. [Reason to confirm: I would rather measure new-customer rate off-platform through post-purchase surveys and hold bids to a real CAC.]

  • Retargeting stays on. I keep it running even when its nROAS looks weak, because it works as part of the funnel prospecting feeds into.

Landing pages for Demand Gen

Cold YouTube and Discover traffic does not know your brand yet. A product page asks for money before the viewer has a reason to spend it. Match the page to the creative:

  • Problem-led creative goes to an advertorial.

  • "Which one is right for me" creative goes to a quiz.

  • A mechanism VSL goes to a long-form sales page.

  • Editorial creative goes to a roundup or comparison page.

  • Offer-led creative goes to an offer page.

Keep the product page in the test as a control. Offer-led creative can convert best sent straight to the product page.

How should you measure Google Ads performance in 2026?

ROAS in Google Ads is a checkpoint. It undercounts Demand Gen, since 98 of 100 YouTube-driven orders leave no click trace, and it overcounts branded Search. Before a major budget decision, I look at 4 things:

  • Incrementality.Geo holdouts or conversion lift studies on Demand Gen and branded Search.

  • Post-purchase surveys. Ask every buyer where they first heard about you.

  • New-customer CAC and payback. Measured against contribution margin, the number that actually hits the P&L.

  • MER. Total revenue over total ad spend, at the business level.

If a channel only looks good in the platform, I would not scale it on that alone.

The 30-day plan to put this into your account

  1. Week 1: Audit and control. Check AI Max settings on every Search campaign. Add brand controls, URL exclusions and negatives. Pull 90 days of search terms from Search, Shopping and Performance Max.

  2. Week 2: Fix the feed. Rewrite titles for your top 20 SKUs from converting search terms. Fill missing attributes. Add lifestyle images. Clear every Merchant Center flag. Move to a third-party CSS if you sell in the EU.

  3. Week 3: Build for intent.Pick your top 3 intents and launch 1 page for each. Duplicate your top 3 to 5 SKUs by angle. Launch comparison and informational Search campaigns. Start an AI Max experiment on your strongest campaign.

  4. Week 4: Open Demand Gen. Launch with 16:9 recuts of your proven Meta winners, matched to the right landing page. Set up a holdout and a post-purchase survey question before spend scales.

"Won't AI Max and Performance Max do all of this for me?"

They handle the bidding and the matching. The feed, the pages and the creative are still yours to build, and the system scales whatever quality it gets. A thin feed and a single product page give AI Max and Performance Max less to work with than your competitor's full catalog of intent pages. Automation raised the value of good inputs.


Frequently asked questions

What is the best Google Ads campaign type for ecommerce in 2026?

It depends on the job. Our standard DTC build uses Shopping or Performance Max to capture product demand, Search with AI Max for non-brand and comparison queries, and Demand Gen to find new customers on YouTube. The split depends on search volume, margins and how much creative you can produce.

Should I turn AI Max off after the September 2026 auto-upgrade?

Not by default. Check which features were turned on, add brand controls, URL exclusions and negatives, then review search terms weekly. If performance drops after 30 days against a comparable period, run a proper experiment before switching it off. For regulated categories, keep text customization off until copy is reviewed.

Can I send Google Shopping ads to a landing page that is not a standard product page?

Yes, within limits. Merchant Center requires the page to show the product's title, image, price, currency, availability and a buy button that match your feed, with the advertised product as the primary focus. A long-form product page with a compliant buy box works. Category pages, search pages and pages without a buy box do not.

How long should a Google Shopping product title be?

Titles can be up to 150 characters. Use the space, but put the product type and the attribute people search for first, because most placements truncate titles. Build titles from the language in your converting search terms. Internal product names rarely match what shoppers type.

Is Demand Gen good for conversions or only for awareness?

Demand Gen drives conversions. In our Q2 2026 YouTube Ads Report, median incremental ROAS across 92 studies was 2x, and 98 of 100 YouTube-driven orders were invisible to click attribution. Measure it with holdouts and post-purchase surveys, because in-platform ROAS will undercount it.

Do Google Shopping ads show in AI Mode?

Yes. Google confirmed on August 31, 2026 that Shopping ads appear in a carousel inside AI Mode responses. Eligible campaign types include Performance Max, AI Max with search term matching, Shopping and broad match. Product overlap with the standard Shopping carousel is low, so fill conversational feed attributes to compete there.

When do Dynamic Search Ads stop?

Dynamic Search Ads migrate to AI Max in February 2027. Move your DSA page targeting into AI Max URL inclusions before then, while you still have DSA performance data to compare against.

What is a third-party CSS, and does it lower Shopping costs?

A Comparison Shopping Service submits Shopping ads in the EEA, UK and Switzerland. Google's own service applies a margin of roughly 20% to bids. A third-party CSS removes it, and agencies report a real-world CPC advantage of 16% to 18%. It only applies to Shopping inventory, and some partners charge fees that offset the saving.

Sources and methodology

This playbook reflects how we run Google Ads for DTC ecommerce brands at Taikun Digital as of September 2026. Platform facts come from Google documentation and industry reporting, listed below. Figures labeled as Google's come from Google and are measured on Google's attribution. Taikun figures come from our Q2 2026 YouTube Ads Report, built with Stella, Northbeam and Fairing.

  • Google, "We're upgrading Dynamic Search Ads to AI Max," April 15, 2026. blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026

  • Google, "Convert faster on YouTube with April's Demand Gen Drop," April 23, 2026. blog.google/products/ads-commerce/demand-gen-drop-april-2026

  • Google Merchant Center Help, landing page requirements. support.google.com/merchants/answer/4752265

  • Google Merchant Center Help, image link and lifestyle image link specifications. support.google.com/merchants/answer/6324350 and support.google.com/merchants/answer/9103186

  • Search Engine Roundtable, Merchant Center product data specification updates for 2026 and AI-generated image labeling.

  • Improvado and Common Thread Collective, AI Max September 2026 auto-upgrade timelines.

  • Common Thread Collective, "Every Google Ads Change in 2026," for AI Mode Shopping ads and the AI Mode product overlap research.

  • PPC News Feed, AI Max experiments and AI Max for Shopping rollout, 2026.

  • WordStream and Strike Social, Google Marketing Live 2026 announcement recaps.

  • Cobiro, Google Shopping CSS guide, for CSS margin and real-world advantage figures.

  • Taikun Digital, Q2 2026 YouTube Ads Report, with Stella, Northbeam and Fairing.

Work with Taikun

Taikun Digital is a Google Premier Partner that runs Google Ads and YouTube Ads for DTC ecommerce brands. If you want your account audited against this playbook, we will show you where your feed, pages and campaign structure are leaving new customers on the table.

Contact us here.



Back to all articles