Week 42 - Google Ads Automated Promotions Start Oct 12: OpenAI's ChatGPT Visual Ads, Google's Spam Update Explainer, and Prime Big Deal Days

MaxWelcome to the Checkout Point, your quick dive into this week's eCommerce buzz with me, Alex, powered by Blicket. These twenty minutes pack the latest trends, news, and insights. Let's navigate the digital marketplace together. Ready? Let's go.

RachelQuick question. When's the last time you personally wrote every word a shopper saw about your store? Because I think the robots have been quietly picking up the pen.

Speaker 3Quietly? Max, they've got a whole desk now.

RachelRight?

Speaker 3And I only care about one thing. If somebody else is making the decisions, who's checking the ROI on them?

RachelOoh, okay. So welcome to Week Forty-Two of the Checkout Point. I'm Max, and that's Rachel, who just made my coffee nervous.

Speaker 3Your coffee should be nervous.

RachelFair. So this week is really one question. How much of your ads and your search presence are platforms now deciding for you?

Speaker 3Which is a little uncomfortable if you're the one signing off on the budget.

RachelTotally. We've also got a shiny new ad surface and a spam explainer. And when a source is secondhand, we'll say so.

Speaker 3Loudly, possibly with a little siren.

RachelSo anyway, let's start with the one that could change your ads on October 12th.

Speaker 3Yes, let's start there.

RachelOkay, so the thing I keep chewing on is the email. Google Ads sent a note to some advertisers, not all of them, saying it will automatically find high-quality promotions on your website and apply them to your campaigns.

Speaker 3Automatically find, meaning nobody on your team clicks anything.

RachelRight. Barry Schwartz at Search Engine Roundtable wrote it up on October fifth, and he describes it as coming soon, starting in October.

Speaker 3Wait, some advertisers. So you could be in the group and not know until the banner shows up?

RachelWell, I can only say what the write-up says. Some advertisers got the email. I wouldn't promise anything beyond that.

Speaker 3Fair. So what does it actually grab? Because promotion can mean a lot of things on a landing page.

RachelGoogle's help page gives an example. Automated promotions pulls an offer like twenty percent off on black sweaters from your landing page.

Speaker 3Black sweaters, of course. Nobody ever discounts the beige ones.

RachelThe beige ones are doing fine, Rachel. Anyway, it attaches that offer to eligible Search and Performance Max campaigns.

Speaker 3Eligible how? I want the conditions because that's where the scope lives.

RachelTwo that the help page names, the campaign has location assets, and it has no manual promotions running.

Speaker 3Hmm. That second one matters. If you've already uploaded a promotion yourself, Google leaves you alone.

RachelThat's how I read it. Your own promotions come first, and the automation fills the gap when there's nothing there.

Speaker 3Which raises a housekeeping question. If somebody on your team uploaded a promotion months ago and forgot, does that count as manual promotions running?

RachelI don't know how the help page treats a forgotten upload, so I won't guess. But it's exactly why I'd open the account and look instead of assuming the answer.

Speaker 3Right. Half the time, the answer to, "Do we have promotions running?" is "Ask Dave."

RachelEvery team has a Dave, and Dave left in March.

Speaker 3So the default is if you haven't said anything, Google says it for you.

RachelYep.

Speaker 3Okay. But I worked in ops at big retailers, and I've seen what happens when two systems disagree about a price. What if the page says one thing and the real deal ended last Tuesday?

RachelOh, no.

Speaker 3I'm serious. An automated system reads a page. Pages go stale. Somebody forgets to pull the banner, and now that old offer is sitting in your ad.

RachelTo be clear, nothing I've seen says that's happening. It's a question, not a documented problem.

Speaker 3Agreed. It's a question you audit.

RachelEvery company has one of those banners, the summer blowout that's still up in November.

Speaker 3And now a machine reads it and says, "Ah, a promotion."

RachelThe zombie sale, brought back by software.

Speaker 3Which is funny until a customer shows up expecting it.

RachelTotally. So what do you do about it if you do care?

Speaker 3The help page has an answer. If you need exact control, for example, over start and end dates, you can upload promotions manually.

RachelStart and end dates. That's the whole zombie sale problem right there.

Speaker 3Right. A manual promotion has dates you chose.

RachelAnd I don't know how Google reads the page, and the sources don't say, so I'd look at it before it goes live instead of after.

Speaker 3Same. Read your landing pages like a stranger would and check which campaigns even qualify. Location assets, no manual promotions. That's your shortlist.

RachelLet me turn that into an actual workflow because go audit is how good intentions die. Step one, list every Search and Performance Max campaign you're running.

Speaker 3Step two, flag the ones with location assets. Those are the campaigns the help page's conditions point at, so that's where I'd spend my attention first.

RachelStep three, for each flagged campaign, check whether anybody has manual promotions running. If yes, you've already made your choice. If no, that's the gap the automation fills.

Speaker 3Step four, open the landing pages those campaigns point to and read the offers out loud. Not skim, out loud. If you'd be uncomfortable seeing that sentence in an ad, fix the page or upload your own.

RachelI like out loud because it catches the stuff your eyes skip. Things read differently when you have to say them.

Speaker 3Step five, write down who owns this decision. A setting with no owner is a setting that changes by default, and I've watched that happen in real operations.

RachelAnd keep it small. You don't need a giant spreadsheet. You need a list of campaigns, a yes or no on each of those two conditions, and a name next to each one.

Speaker 3A name and a date. Come back after the rollout has actually happened and look again so you can see what really shows up.

RachelWhich is a good point, because everything we've said comes from the email write-up and the help page. What appears in live accounts, we don't know yet.

Speaker 3That's the honest limit. The audit is about being ready, not about predicting what Google will do with your pages.

RachelSo anyway, Google is deciding what your offer sounds like. Speaking of platforms controlling what shoppers see, OpenAI just opened a new ad surface.

Speaker 3Oh, now that's a different animal. Go on.

RachelOkay, so this one landed on October fifth too. OpenAI announced a visual ad format for ChatGPT.

Speaker 3Visual how? Banners?

RachelBarry Schwartz at Search Engine Roundtable wrote that OpenAI will test it later this month in the US with an initial group of advertisers.

Speaker 3Initial group, so not you, not me.

RachelNot me, definitely. And the format shows up during image generation.

Speaker 3Wait, during the image itself? Like, I ask for a picture of a sofa, and a sofa brand pops in?

RachelKind of, yeah. That's the surface being tested, image gen.

Speaker 3Hmm. That's a very specific moment to catch someone.

RachelIt is. Think about what someone's doing when they generate an image. They're mid-idea, imagining something they want, a room, an outfit, a gift.

Speaker 3So it's closer to intent than a banner on a news site, or at least that's the pitch an advertiser would hear.

RachelRight. And I'm reading between the lines on that. The source just says the format shows up during image generation. What it means for buyers, nobody knows yet.

Speaker 3Good. I'd flag that as a guess. Interesting moment, unproven moment.

RachelRight. Now, there's a second write-up on Unite AI that adds some detail. It says the ads will be labeled and kept separate from the generated image.

Speaker 3Okay, that matters. Nobody wants the ad baked into the picture.

RachelTotally. And it says OpenAI states that ads don't influence answers.

Speaker 3Says. That's OpenAI's claim, and I'd like that tagged as a claim.

RachelFair. It's their statement, and the report passes it along.

Speaker 3I've heard we keep it separate in a lot of meetings, usually right before it wasn't.

RachelOuch.

Speaker 3I mean, come on. Separate on a slide is easy.

RachelSo what would make it real for you? Not a slide, something you can see.

Speaker 3Labels I can see with my own eyes, placement I can check myself, and I'd want to know how a shopper can tell the ad from the image without squinting.

RachelWhich, to be fair, the report says is the design, labeled and separate. Whether it feels that way in practice is the open part.

Speaker 3Right. The claim is the starting line. The test is the whole race.

RachelSame report also mentions expanded measurement partners.

Speaker 3Which is the part I care about.

RachelAnd brand safety pilots with DoubleVerify and Integral Ad Science.

Speaker 3Huh. So the brand safety names are already in the room.

RachelYeah, before most of us have even seen an ad.

Speaker 3Explain what a brand suitability pilot even is for the folks who don't buy media all day.

RachelPlain version, and as I understand the term, it's about whether your ad shows up next to stuff you're comfortable being next to, and I'm keeping it general because the report only names the pilots and the two companies.

Speaker 3And expanded measurement partners, meaning more outside eyes on whether ads did anything. That's the part a buyer cares about, but it's a mention, not results.

RachelExactly. A mention in a report. No numbers, no case studies, nothing to benchmark against.

Speaker 3Okay, but what's the actual ROI? There are no results yet. Nothing.

RachelZero data, right?

Speaker 3So if you're a merchant with budget, what do you need to see before you spend a dollar? Clicks, sales you can trace back, something about who's even looking at the image?

RachelHonestly, I don't know. Probably all three.

Speaker 3Yeah, I don't either. I'd want an answer before a test budget, not after.

RachelOkay, let me try to answer my own question without pretending I know. If I were running that test, I'd want to decide the yardstick before the money moves.

Speaker 3Say more. What's the yardstick?

RachelPick one number you'd actually act on. Sales you can trace back is the cleanest. Clicks are noisy, and people saw it is the weakest of the three.

Speaker 3I'd add a ceiling. Decide upfront the most you're willing to lose finding out. Then it's an experiment, not a leap of faith.

RachelAnd since only an initial group of advertisers gets in first, for most of us, the first move is just watching. See what the early group says, if they say anything.

Speaker 3Which is not nothing. Being second isn't a failure. It's a cheaper way to learn.

RachelAnd one more thing for the watching phase. Decide now what you'd want from an early advertiser's report, not the headline, the specifics, what they measured over what window compared to what.

Speaker 3Right, because it worked is a sentence. Here's what we measured and how is evidence. I'd trust the second one and discount the first, especially from anyone selling the format.

RachelFair. And even a solid report from the early group tells you about their store, not yours.

Speaker 3Which is why the ceiling matters. You're buying an answer for your own business, not borrowing someone else's.

RachelSame. Speaking of which, that's a platform telling us it's got things under control.

Speaker 3Mm.

RachelNow the other side of the Google story, how Google talks about deciding what deserves to be seen.

Speaker 3Spam, you mean.

RachelSpam. Yep. Okay, so Google's spam side. Barry Schwartz flagged a recap of a three-day event in Europe called Google Search Central Live Deep Dive.

Speaker 3Wait, flagged a recap, not the event itself?

RachelRight. Secondhand. A writer named John Campbell wrote it up at We Are Roast, and Barry pointed to it, so we're a couple of hops from the room.

Speaker 3Got it. So hold it loosely. What did Google actually say?

RachelThat it addressed why it has rolled out so many spam updates in twenty sixteen. That's all the evidence I got, and I'm not going to invent the reasons.

Speaker 3Refreshing. Most spam explainers I've read have a lot more certainty than sources.

RachelAnd it's worth saying how I'd frame it for your own team. If somebody asks, "Did Google say X about spam?" The honest answer is, "A recap says Google addressed it." Different sentence, different confidence.

Speaker 3That's the sourcing ladder. Original event, then the recap, then the post about the recap, then whatever your Slack channel turns it into.

RachelBy the end, it's a game of telephone with a ranking factor in it.

Speaker 3My favorite kind of game.

RachelTotally. There's also a second Barry post. Google's Gary Illyes reportedly presented new data on crawling, indexing, and serving timelines.

Speaker 3Numbers?

RachelNone in what I've got, so no figures from me.

Speaker 3Fine, but why does a merchant care? Spam filtering sounds like someone else's problem.

RachelDoes it, though? Think about the last three stories.

Speaker 3Hmm. Google writes your promotion. ChatGPT gets an ad surface. Now Google decides what counts as spam and what gets seen.

RachelThree different doors, same hallway. The platform decides more of how your store shows up.

Speaker 3And my supply chain brain says that's a bottleneck you don't own. If a default is wrong, who finds out first?

RachelProbably a customer or your revenue line weeks later.

Speaker 3Mm. Ouch.

RachelAnd note the scope. We can't say any of this hits your store. The recap says Google addressed its spam updates this year. Whether your pages feel it, only your own data tells you.

Speaker 3Which is why the audit has to include search, not just ads. Look at your own traffic and see whether anything shifted around update time, and don't assume the cause. So the move is boring and kind of great.

RachelSo the move is boring and kind of great. Audit the defaults. Open the settings you never touched and check what switched on for you.

Speaker 3And put it on a calendar because an audit nobody schedules is just a nice idea.

RachelConcretely, a short recurring slot and one owner with a list of what each setting is currently set to.

Speaker 3And write down what on purpose looks like so the next person doesn't have to guess.

RachelDocumentation, the least glamorous word, the most useful.

Speaker 3And note the date of each check so when something does change, you can say, "We looked on this day, and it was fine."

RachelBoring, and it saves you from arguing about what everybody remembers. Memory is a terrible audit log.

Speaker 3Which, speaking of calendars...

RachelOkay, let's bring it to the actual dates because holiday season is basically here. Okay, Prime Big Deal Days, October sixth, running through the seventh. NBC News, updated October fifth, confirms the dates.

Speaker 3So that's tomorrow, depending on when someone hears this.

RachelRight. And NBC calls it the unofficial start of the holiday shopping season.

Speaker 3Unofficial. Retail has been unofficially starting the holidays since August.

RachelSince the Halloween candy showed up next to the sunscreen.

Speaker 3But I'll give them this. I ran ops at big retailers, and a two-day event pulls forward demand like nothing else. Shoppers who see a deal on Amazon go check your price that same hour.

RachelWait, even if you're not on Amazon?

Speaker 3Especially then. Your conversion rate that week is partly set by someone else's event.

RachelOkay, make that concrete for me. I'm a small store. I'm not running any Prime event. What do I actually do on October sixth?

Speaker 3First, I'm working from experience here, not from anything in the NBC piece. It only gives the dates and the framing. But in my ops days, the week of a big competitor event, we checked our own prices and stock before anything else.

RachelPrices and stock first.

Speaker 3If a shopper arrives from a comparison and your best seller is out, you paid for a visit that couldn't turn into anything.

RachelThat's the thing people forget. Traffic isn't the win. A checkout that works is the win.

Speaker 3Here's a story. A team I worked with spent a whole week arguing over promotion copy, and nobody checked what the product pages said about shipping. The copy was great. The shipping message was wrong. Guess which one customers noticed?

RachelOuch. Okay, mine's a smaller example. Picture someone who's been eyeing a pair of headphones for weeks. A big sale hits, they see a price, and they open a few tabs to compare. You're one of the tabs.

Speaker 3And the tab that answers faster and clearest tends to get the click. Again, that's my read, not data from any source we have.

RachelRight, and I'm not claiming a number for how many shoppers do that. I'm saying the comparison behavior is plausible, and you can plan for it cheaply.

Speaker 3Cheaply is the point. Check the price, check the stock, check the shipping message, check that checkout works on a phone. That's a one afternoon list.

RachelHuh. Depends if your site is ready for the traffic. Speaking of which, quick hits, different flavor.

Speaker 3Go.

RachelBarry Schwartz found a new help document. Google Ads has an AI campaign localization tool for search campaigns. It's in beta.

Speaker 3Localization meaning translation?

RachelThe excerpt I've got just says the tool is there, and it's beta. I don't have the details beyond that, so I'm not gonna guess what it rewrites.

Speaker 3Good, because AI and your ad copy in another market is a sentence that should make a merchant sit up.

RachelTotally. Another one that quietly changes what shoppers read.

Speaker 3And that's the question I'd ask anyone thinking about trying it. If you sell in more than one market, who on your team can actually read the result and say, "Yes, that's how we talk"? A tool shouldn't be the only reader.

RachelRight, and again, I'm speaking generally. I don't know what the tool outputs. But a person who knows the market checking the copy sounds like a smart rule for any new automated thing.

Speaker 3Hmm. Beta, though, so somebody's gonna find out the hard way what the defaults do.

RachelHopefully not you.

Speaker 3Hopefully not anyone listening. Okay, my quick hit is a practical e-commerce piece, and Smarty wrote it, and it's about adding social profiles to Search Console.

RachelSearch Console and social. Odd couple.

Speaker 3The suggestion is simple: publish a video on YouTube, put related product images on Instagram, then track organic search performance for each.

RachelOh, I like that. It's an actual measurement plan instead of just posting and hoping.

Speaker 3Right. Because posting is easy. Knowing whether it showed up in search is the part nobody does.

RachelSo walk me through it. I post a video, I post some pictures, then what?

Speaker 3Then you look at how each one performs in organic search in Search Console. I don't have more detail on the setup than that, so I'd read the article itself before trying it.

RachelAnd what do you do with the answer?

Speaker 3Decide where your next hour goes. If the video shows up and the images don't, or the reverse, you've learned where to spend limited time. If neither shows up, you've learned that too, and it cost you almost nothing.

RachelThat's the part I like. It's a cheap experiment with an answer at the end, not a vague brand awareness thing.

Speaker 3Brand awareness. Dry. The two words that have funded a lot of bad decisions.

RachelFair. And one video, one set of images is small enough to do before the holiday rush.

Speaker 3And it feeds the same question we keep asking: Where does your presence actually show up, and can you see it?

RachelSo anyway, let's wrap this up with what to actually do. Okay, so the handoff said what to actually do. Rachel, you've got one action item, right? Just one?

Speaker 3One. Because if I give you five, you do zero. Open your Google Ads account and check your promotion settings before the rollout date.

RachelBefore October twelfth.

Speaker 3Yes. Then decide on purpose. Let Google pull offers from your pages or upload your own with dates you control.

RachelOkay, but can I push on that? You said one item. Is there nothing else a listener could do this week if they've got a spare hour?

Speaker 3I said one item that matters most. If you've got the hour, here's the short extension. Item two, look at your search campaigns and see which ones meet the two conditions we talked about, location assets and no manual promotions.

RachelItem three for me, if you're curious about the new ChatGPT ad format, decide your yardstick before anyone shows you a pitch. One number you'd act on and a ceiling on what you're willing to lose.

Speaker 3Item four, schedule the audit. One owner, a recurring slot, and a written list of what each setting is on purpose. It's the item that makes the other three stick.

RachelItem five is the cheap one. Prime Big Deal Days run the sixth and seventh, so check your prices, your stock, and your checkout on a phone before you worry about anything fancy.

Speaker 3And a bonus if you've got Search Console open anyway, try the video and images idea from Practical Ecommerce and see which one shows up in organic search.

RachelOkay, that's more than one. You said five does zero.

Speaker 3I said if I give you five, you do zero. So pick the first one. The rest are the menu.

RachelLove it. So the week in one breath. Google writes your promotions. ChatGPT gets a new ad surface. Google spam talk came to us secondhand. And Amazon throws a two-day party to kick off the holiday season.

Speaker 3Different stories, same question underneath. Who's holding the pen on your store?

RachelRight. And the answer keeps being, um, not always you.

Speaker 3Which is fine as long as you checked.

RachelAnd if you want one line per story, here's mine. Google's Promotions know your defaults and decide on purpose.

Speaker 3ChatGPT's visual ads, it's a test with an initial group of advertisers. So for most people, the move is watch and set your yardstick, not spend.

RachelGoogle spam talk, it came through a recap relayed by another write-up. So hold it loosely and look at your own traffic instead of trusting a headline.

Speaker 3And Prime Big Deal Days plus the quick hits. Someone else's event shapes your week. A beta localization tool is one more thing that changes what shoppers read. And measuring where you show up beats posting and hoping.

RachelThat's honestly a whole episode in four sentences.

Speaker 3Which is why it's a short show and not a four-hour one.

RachelOkay, my joke. If a robot is going to write my promotions, I want it to at least ask what my margins look like first. Humans don't, but I'd like to think a robot would be polite.

Speaker 3It won't ask.

RachelIt won't ask. So we ask for it. Before I forget, if you like the show, subscribe wherever you listen and swing by blicket.co.

Speaker 3Yes, blicket.co. Go look.

RachelQuick confession. Every time Rachel says audit, I feel a little guilty about the settings I haven't opened.

Speaker 3Good. Guilt is a free reminder. Open one settings page you've never touched this week. Just one.

RachelAnd if you find something odd in there, you've already paid for the episode.

Speaker 3The episode is free, Max.

RachelThen it's the best return on investment we've ever had. And Rachel, go check your own settings.

Speaker 3Already open in another tab, Max. Talk to you next week.