A new A/B testing framework makes broad matching a piece of cake

No more duplicating campaigns or manually syncing changes.

Google Ads just made it easier to test broad match keywords with a new A/B testing framework built directly into Smart Bidding Search campaigns.

Now, you can compare broad match performance against your current setup with a single experiment.

How it works: When you launch the experiment, Google automatically creates broad match versions of your existing exact and phrase match keywords and splits traffic 50/50 between the two.

All campaign-level updates apply to both, making your test nice and consistent, and any changes at the campaign level apply across the board.

However, it doesn’t work if you’re using portfolio bid strategies. Just FYI.

Where to find it: You can set up these tests from the Experiments section or directly through a relevant recommendation in your account. Built-in reporting lives in the Experiments tab, too.

By the way, Google suggests running the test for 6–8 weeks to get enough reliable data. If performance holds up, you’ll have a clear path to broader reach without the usual testing pain.

Pretty cool.

Individual asset reviews and a crackdown on spammy AI content

If your ad campaign is a Jenga tower, Microsoft is pulling individual blocks.

And you’re still standing: Microsoft Advertising now reviews ads at the asset level, meaning it evaluates headlines, images, and descriptions individually.

If one element gets flagged, the rest of the ad can keep running as long as enough approved components remain.

Hopefully Microsoft’s update cuts down on annoying downtime while keeping up performance, especially during edits or policy reviews.

Sadly, appeals and notifications for individual assets aren’t available just yet… but they’re on the way.

Just be careful about how you’re using AI: Microsoft is going all parent-y on AI-driven scams and shady ads. It removed or restricted over 1B ads in 2024 alone.

It also banned 75,000 advertiser accounts and demonetized 250,000 publisher pages that violated content policies via deepfakes, misleading claims, or fake endorsements.

Of course, this should make the platform safer and smarter for both you and your customers.

… So long as you play by the rules. We’ve said it before, but be good out there, OK?

AI is already absorbing your search traffic. Here’s what you can do about it

Ever asked ChatGPT to recommend “the best product” that actually fits your needs?

Chances are your customers—and shoppers everywhere—are doing the same.

Even if your content has great rankings, AI summaries are siphoning away the clicks you used to count on.

But don’t stress.

Inside our AI Search Data Story, we show you exactly why and how this happens, and how you can get a head start on adapting to it.

You’re about to see:

• How AI Overviews affect organic visibility, even for top-ranking pages.
• What types of queries and formats still drive real traffic.
• What the next six months look like for Gen AI vs. traditional search—and how that could impact your content strategy.

The good news is, most brands haven’t caught on yet.

That gives you a rare opportunity to turn the trend to your advantage. But only if you hurry…

Find out how people are searching now, and how your brand can keep showing up.

Where AI can’t (and maybe shouldn’t) replace humans yet

Whatever you can do, AI can do better.

… Or can it?

Kelly Ayres argues the opposite. In fact, there are multiple instances where leaving things up to AI isn’t just a gamble—it actively holds you back.

She points out 44 marketing initiatives that still demand human intervention. Here are just a few:

1) Approving on-the-record quotes from SMEs, executives, or customers. Consent and nuance matter. Otherwise, you can build entire strategies on the wrong signals.

2) Selecting hero imagery or video thumbnails. Brand, cultural, and accessibility sensitivities all come into play.

3) Running voice-of-customer interviews and extracting pains in their own words. Can an AI read body language cues?

4) Predictions, projections, and philosophical content ideation. AI is reactive, not predictive, meaning only humans can break new ground.

5) Updating statistics, legal references, or medical data points. AI makes things up a lot, so human verification is a must.

6) Multi-language copy or cultural localization. Things like idioms, taboos, and regional context can all be missed by literal translations.

7) Negotiating partnership placements or guest posts. Relationship building, pricing, and editorial standards require soft skills like empathy, persuasion, and judgment.

8) Root-cause analysis of traffic drops or ranking volatility. Models might not see market shifts and website quirks.

9) Assessing sentiment in analyst reports. Language is nuanced and so must be interpreted by strategists.

10) Mapping experiment insights back to product-roadmap priorities. Political capital, sprint capacity, and revenue forecasts should be weighed by humans.

11) Selecting real-world examples or anecdotes. Do you really want to leave this up to ChatGPT to pick?

… And those are just 11 of Kelly’s points.

Want the other 33? See the full list here.

Social media changes fast. Here’s how social teams at Netflix and Apple keep up

When you’re writing posts, managing UGC and influencers, and planning campaigns, it’s easy to miss important social media updates.

… Which is why social media teams at top brands like Netflix and Apple use Geekout to stay ahead.

It’s the free weekly newsletter that captures critical social media news, trends, and shifts in a super-stuffed, fun read.

If social media is part of your job, missing Geekout should be considered malpractice.

Just kidding. Or are we?

Stay sharp. Subscribe for free (30,000+ subscribers).

Google Ads vs. Facebook Ads: Which platform should your e-commerce brand use?

Maybe you’re launching a new product, or simply exploring new paid channels.

At some point, you’ll have to answer the big, big question: Which of the two giants should you start with—Facebook or Google?

As always, there are pros and cons to each…

The case for Facebook:

Facebook is great for brands that sell one-of-a-kind products. If your product is entirely new, and you need to create demand around it, then Facebook is the platform you should go with.

The case for Google:

If your product is well-established, meaning you A) have a lot of competitors or B) you’re solving a problem that’s widespread enough that people search for it, then use Google Ads.

Google Ads can also work well even if your competitors are few in number, as long as there’s enough search volume around the problem you solve.

Does Google ever work for new products at all? Yes. Google has demand-capturing campaign types, like Search and Shopping.

It also has demand-generating campaign types, like YouTube, Demand Gen, and Performance Max.

But that being said, Google is more intent-driven as a whole. It’s designed to focus on what users are searching for.

So Facebook still wins when it comes to driving demand for products.

AI EDUCATION: ChatGPT, Claude, Gemini, Midjourney… so many names, but what’s actually useful for you in your work? There’s a newsletter called The Deep View that exists to sift through all the noise and get you up to speed on what’s actionable with AI products, and it’s free. Join 452,000+ subscribers with one click and let AI empower you.*

AI MARKETING: Google’s video-generating model, Veo 3, is now open to all Google Cloud users via Vertex AI Media Studio. Veo 3 is scary good at generating videos from scratch, so if you’re a Cloud user and you’re in need of video content, it’s probably worth trying out.

E-COMMERCE: Buy now, pay later (BNPL) is the hot way to pay. Gen Z even views BNPL as safe and normal, rather than risky. But while this method can help you lift sales, it’s also creating losses—Klarna saw unpaid loans double in Q1 alone. Something to watch…

AI MARKETING: It seems agentic AI is struggling to live up to the hype, kind of like The Last of Us season two. Despite bold promises, most agents can only complete between 30%–35% of tasks, often fail at the basics, and raise serious privacy concerns. Keep an eye on those workflows!

*This is a sponsored post.

A man is born in 1946 and dies in 1947, yet he is 86 years old when he dies. How is that possible?

You can find the answer here.

POOLSIDECHAT

Cool tech, (funny) business, lifestyle and all the other things marketers like to chat about while sipping cocktails by the pool.

Trader Joe’s moves across the street from… Trader Joe’s?

This one reminds us of the Spider-Man meme…

Apparently a brand-new store just opened in California, right across the street from another Trader Joe’s. And naturally, the internet is baffled.

Yes, the new one is bigger. Yes, it has underground parking. But it’s still Trader Joe’s. Same foods, same Hawaiian shirts.

So… is this peak convenience? Or just late-late-stage capitalism doing its thing?

Interested in advertising in one of our newsletters?

Connect with over 100,000 of the world’s best marketers who read a newsletter by Stacked Marketer.

Learn more.