Universal Commerce Protocol (UCP) Explained: What Every Ecommerce Brand Must Know

1 - > SPEAKER_00: Hello and welcome to another episode of Growing 2 - > E-Commerce. 3 - > I'm one of your hosts, Mike Ryan. 4 - > Unfortunately, Chris can't join us today, but we'll be back in 5 - > the studio for normal production next week, and we'll look 6 - > forward to seeing you then. 7 - > Meanwhile, I'm going to revisit a topic that we first covered 8 - > back in February: UCP, Universal Commerce Protocol.

9 - > What is it? 10 - > How does it work? 11 - > This was one of our actually, it was our single most popular 12 - > episode of the year so far. 13 - > Um, it even beat an interview with Chinny Marvin about AIMAX 14 - > for search, which I we accidentally sent to the podcast 15 - > feed.

16 - > So that shows you how popular and this topic is, how much 17 - > interest there is. 18 - > And um, we'll get into a little bit about why that is, or let's 19 - > just jump into it. 20 - > I think the situation right now is one of fear and confusion or 21 - > also impatience. 22 - > There's a lot of emotions going on.

23 - > Organizations, broadly speaking, have this kind of AI mandate, 24 - > right? 25 - > There's uh across many levels like are our employees using AI 26 - > enough? 27 - > Are they using it too much? 28 - > But also when it comes to topics like the future of search, web 29 - > traffic, the impacts on our customer acquisition channels 30 - > and stuff like that.

31 - > I believe everyone saw what happened to publishers when AI 32 - > overviews rolled out and people are spooked. 33 - > You know, uh, no one wants to go dark or become invisible. 34 - > No one wants to get left behind or miss this. 35 - > So that is the urgency that's sort of driving interest here.

36 - > And then the next thing that comes into place is that the 37 - > topic is complicated, it's hard to understand. 38 - > You'll see wildly varying information about how to 39 - > implement it or what is really necessary. 40 - > And so, because of all this, I want to hopefully simplify this 41 - > topic. 42 - > I think it's worth revisiting.

43 - > We know a lot more now than we did back in February, and just 44 - > spell out what is UCP? 45 - > Really? 46 - > What is it? 47 - > Why does it matter?

48 - > And why is Google building this and everyone signing on? 49 - > And then we'll get into some of the basics of implementation. 50 - > I want to give you a reasonably detailed answer that can help 51 - > you understand what this is actually about. 52 - > But I I guess you understand in the scope of a podcast episode, 53 - > I can't get into every single implementation detail or 54 - > scenario.

55 - > But I and by the way, I'm not a developer, but I think that's 56 - > valuable here because I'm just out here trying to pay attention 57 - > in January. 58 - > I was reading all the documents and I've come back and looked at 59 - > them again more than once. 60 - > And so just I think we can talk about this at eye level and come 61 - > a bit further together. 62 - > That's the goal of this episode.

63 - > Um to kick that off, what the F is UCP? 64 - > There's the this in and of itself is the subject of a lot 65 - > of confusion, in my opinion. 66 - > Um like to describe what it is, though, I think first we have to 67 - > take a step back and look at this topic of agentic commerce, 68 - > agentic shopping. 69 - > Um, if you're a longtime listener of the show, you'll 70 - > know I don't particularly like that phrase.

71 - > Uh me, I would view this more as AI-assisted shopping. 72 - > It's a little less buzzwordy to me, and it describes more how 73 - > consumers experience this. 74 - > Consumers don't know what agentic commerce is. 75 - > Consumers know that an AI is helping them find products and 76 - > do stuff.

77 - > If you just go back a year, um the idea back then, the very 78 - > popular idea, and it's still in the public's mind, was that 79 - > computers would be literally navigating our e-commerce shops 80 - > and like literally scrolling through and adding things to the 81 - > basket and doing the whole checkout. 82 - > And there were big articles about this in all kinds of 83 - > publications. 84 - > Um that, you know, and the feedback was not good, or like 85 - > the it didn't because frankly, this was never how this whole 86 - > thing was going to work.

87 - > That's my take. 88 - > Maybe we'll get there one day. 89 - > But it's dangerous, it's slow, it's expensive, it doesn't 90 - > scale. 91 - > You'd hear these stories about an AI spending 45 minutes to buy 92 - > the wrong product with the most expensive shipping option or 93 - > stuff like this.

94 - > And the this got everything off on a very bad footing because um 95 - > this isn't really should shouldn't have been how we 96 - > imagined this happening. 97 - > And again, there's a lot of public awareness of this now, 98 - > which increases what we call the trust gap in terms of consumer 99 - > adoption of AI for shopping. 100 - > Um, because consumers think things like, I would never give 101 - > my credit card to an AI. 102 - > And that's good.

103 - > You shouldn't give your credit card details to an AI, in my 104 - > opinion. 105 - > I also wouldn't do that. 106 - > Um, but that's not how agentic commerce or AI-assisted shopping 107 - > would is actually working or will work in the future. 108 - > This is basically a misconception at this point.

109 - > Um so the alternative to this kind of would be the old school 110 - > plumbing situation. 111 - > And this is also not a proper solution to agentic shopping. 112 - > What you can do here is on the one hand, imagine an 113 - > indeterminately large list of all the AI surfaces where 114 - > consumers are and might use an AI assistant. 115 - > And this could be things like ChatGPT, um Google AI mode, or 116 - > if we look at the way the direction that Google is taking 117 - > this in, uh that could also be YouTube, Gmail.

118 - > You know, Google wants to embed a universal cart across all of 119 - > their surfaces so that you can shop agentically uh from 120 - > basically anywhere on Google. 121 - > Um, but it can also be other surfaces in the future like 122 - > AI-powered experiences in a marketplace or wherever. 123 - > Really, we don't even know yet, but an indeterminately long list 124 - > of surfaces, consumer services. 125 - > And then on the other hand, you'd have all the merchant 126 - > shops and the merchant catalogs behind that.

127 - > And you would need a connection, like an API or something like 128 - > this, between every single one of these with each other. 129 - > And this is something that I nicknamed the glass spider web 130 - > because that's what it would be. 131 - > It would be this unimaginably complex network of connections 132 - > and totally brittle and fragile, um, ready to break, built to 133 - > break practically. 134 - > So this was also not going to be the way that this was going to 135 - > work.

136 - > Um and what we have instead is the solution, a pretty elegant 137 - > solution in the end. 138 - > Well, you can that's very debatable. 139 - > It it is complicated, but once you wrap your head around it, 140 - > it's not bad. 141 - > And it's this thing co-developed by or co-founded and 142 - > co-developed by Google and Shopify called Universal 143 - > Commerce Protocol.

144 - > Um, and this is exactly to provide a third way here. 145 - > So it it is more like conventional plumbing, but it's 146 - > scalable. 147 - > And it does give AIs the freedom, but can also power 148 - > experiences that have nothing to do with AI. 149 - > Um, so you know, we have that whole list of consumer surfaces 150 - > and platforms, and then we'll have a whole list of business 151 - > platforms, um, retail shops and stuff like this.

152 - > And the other complexifying layer is that there's payments 153 - > in here, CRMs, ERPs, PIMs. 154 - > There's uh all kinds of things that need to happen in here. 155 - > I think let's universal commerce protocol. 156 - > I think we know what universal means, right?

157 - > It it means that it should be compatible. 158 - > Everyone should be able to use it, it should be completely 159 - > compatible. 160 - > Um commerce, I hope we all know what commerce is. 161 - > And then the the maybe the mysterious word here is a 162 - > protocol.

163 - > So just to give another example that might make this more 164 - > familiar, a protocol is is uh basically a way for systems to 165 - > communicate. 166 - > Let's look at um I work at a company that uses Google 167 - > Workspace. 168 - > Some of my friends or customers work at companies that use 169 - > Microsoft Outlook, Microsoft Teams, whatever. 170 - > And um, we're able to send emails to each other, despite 171 - > that fact.

172 - > And it's because there's an email protocol that standardizes 173 - > things and allows these different systems to communicate 174 - > with each other. 175 - > And so we're talking about a universal commerce protocol, and 176 - > it consists of these different layers that help bridge between 177 - > the consumer platforms and the business platforms. 178 - > Um the first layer of that is a services layer, and this is very 179 - > abstract. 180 - > In this case, we're talking about shopping because actually 181 - > commerce is a lot bigger than just buying products.

182 - > Um like the there's also um travel, there can be yeah, 183 - > hotels and booking, there can be a lot more that will be 184 - > supported here eventually. 185 - > But the first use case being tackled by UCP is shopping. 186 - > So that should, by the way, give you a glimpse into how big this 187 - > topic can be, perhaps will be, we'll see. 188 - > Um, but they want to reinvent not just shopping, but all of 189 - > commerce.

190 - > They want an infrastructure to handle this across these 191 - > different kinds of verticals. 192 - > Um but within the vertical surface of shopping or service 193 - > of shopping, that's where we are, uh it comes down then to 194 - > capabilities. 195 - > This is the next layer in the communication between these two 196 - > sides of the market, basically. 197 - > Capabilities, this is basically just about defining what is 198 - > e-commerce, actually.

199 - > What are the main things that need to be done? 200 - > What are the biggest Lego blocks that of which this whole thing 201 - > is is built? 202 - > And so this is things like catalog, carts, checkout, order 203 - > management, these core areas of activity that we're all familiar 204 - > with. 205 - > In order for AI to navigate this, we'll get to that, there 206 - > just needs to be these def these defined areas of activity.

207 - > And then there's another level to this, which are called 208 - > extensions. 209 - > And these are things like discounting and others. 210 - > But that's these are smaller things. 211 - > And the reason they want to, the difference between capabilities 212 - > and extensions, capabilities are basically the building blocks or 213 - > the cornerstones.

214 - > And extensions are the smaller things. 215 - > I think they recognize that there's a risk of scope creep 216 - > here. 217 - > And they don't want to kind of trivialize what a capability is. 218 - > They're keeping another more detail there of extensions, 219 - > which is for handling this other stuff that might come up as we 220 - > go along.

221 - > The purpose at a high level of these things, basically, you can 222 - > have something like AI mode, this agentic thing, and then 223 - > your web shop, for example, um on your built-on your e-commerce 224 - > platform, whatever it is. 225 - > And the cape, you're you will support certain capabilities, 226 - > the AI will support certain capabilities, and they talk to 227 - > each other. 228 - > This is a thing called negotiation, and they basically, 229 - > because we'll get into how that works in a little bit more 230 - > detail, but they look at each other and say, hey, I have this 231 - > capability and this one, this one.

232 - > Oh, I have these. 233 - > They find out what they have in common, and they know that is 234 - > how they can interact with each other. 235 - > So this is very important. 236 - > And then the final most technical layer of this whole 237 - > thing are what's called transports.

238 - > And this can be um APIs like a classic REST API. 239 - > This is a very this is uh architectural style for building 240 - > an API, and it's the way a lot of the internet is functions 241 - > using these. 242 - > And then there's these newer flashier ones like MCP, model 243 - > context protocol, agent-to-agent protocol, and uh and others. 244 - > So these newer, a more general purpose AI protocols that have 245 - > been built.

246 - > What Google and Shopify did here is they don't care which of how 247 - > you want to transport information, it's going to be 248 - > supported. 249 - > This is the point that it's universal. 250 - > Um so they've they've built this system that just helps 251 - > businesses and AI services talk to each other. 252 - > That's the point of all of it.

253 - > And there doesn't need to be a custom API between every single 254 - > thing, and the AI does not need to be going on your actual 255 - > effing website and clicking around on it. 256 - > That is the whole point of this. 257 - > That's what UCP is. 258 - > So I hope I've explained it clearly enough for viewers.

259 - > I'll there'll be a visual a visual on-screen to support 260 - > that. 261 - > But and the next thing that I want to mention about UCP, which 262 - > is very important because we, you know, we talked about this 263 - > idea of universality or it being universal, this is important. 264 - > For any kind of standard like this to succeed, it needs to be 265 - > accepted, it needs to be adopted. 266 - > Now we talked about this on past episodes.

267 - > OpenAI had a competing standard. 268 - > Actually, theirs came out first. 269 - > They were co-developing it with Stripe, and it was called ACP. 270 - > It still exists, I guess, but it's not going to make it.

271 - > They've lost. 272 - > At this point, across every layer, UCP has been endorsed and 273 - > accepted, which is very important to it actually 274 - > becoming a standard. 275 - > And if we look at this, it doesn't matter which category 276 - > you're talking about. 277 - > The big AI hyperscalers have all accepted it.

278 - > Amazon, Meta, Microsoft, of course, Google. 279 - > Google likes to joke that this is probably the only thing they 280 - > ever agreed on, which might actually be true. 281 - > Uh then marketplaces and retailers. 282 - > There have been big endorsements from Walmart in the US, Zalando 283 - > here in Europe, Etsy, Wayfair.

284 - > The list goes on and on. 285 - > Everyone who's anyone has endorsed this. 286 - > Then there are the commerce platforms, and there's probably 287 - > a little bit more work to do here, actually. 288 - > Uh, but Shopify, uh, of course, they're a co-founder of this, 289 - > Salesforce, Commerce Tools, it's seeing acceptance here as well.

290 - > Um, and this is where it gets a little complicated when it comes 291 - > to implementation, because not everyone is a Shopify shop. 292 - > And even for Shopify, it's not perfectly plug and play. 293 - > Um, but we know plenty of retailers and brands who are 294 - > using other Commerce platforms or have bespoke Commerce 295 - > platforms. 296 - > So we have to see how this will develop as well.

297 - > Um, and then also I just want to mention the payments layer 298 - > because this is also super important. 299 - > Visa, MasterCard, MasterCard, uh, PayPal, Stripe, Klarna, and 300 - > every other one that you can think of. 301 - > The probably the broadest adoption is among payment 302 - > processors. 303 - > They're all like, yes, get me on board.

304 - > If there are payments to be processed, I will be there. 305 - > Uh, they're very excited about this. 306 - > And that's important too because some of this will push you 307 - > toward like Google Pay, but ultimately every payment 308 - > processor is supported. 309 - > So let's talk about implementation a little bit.

310 - > Um, friend of the podcast, Joey Bidner, expressed some 311 - > frustration that there are videos out there on YouTube or 312 - > wherever saying that all you need to do is populate this one 313 - > field in your Google Merchant Center and you're good to go. 314 - > Um, and you know, then there's people like me mentioning that 315 - > if you would print out the entire UCP.dev documentation, it 316 - > would probably be a 300-page book. 317 - > It's a frustrating situation.

318 - > Like, where does the truth lie and what's going on here? 319 - > Um and actually, maybe before we get into a couple of details, I 320 - > want to answer that at a higher level with the classic answer. 321 - > It depends. 322 - > But, you know, the worst place you can start right now is 323 - > probably to go to ucp.

dev, which is the full documentation. 324 - > You know, this is really a topic where you don't want to boil the 325 - > ocean, so to say. 326 - > It will already the reason I say that is because that is designed 327 - > for every use case that could happen and every implementation 328 - > and all this kind of stuff. 329 - > It's the biggest and kind of the most unfenced, or you you, if 330 - > you look at it, it might not be clear to you what and this is 331 - > relevant to me and what is not.

332 - > So that's what's so I think challenging about that, or why I 333 - > wouldn't necessarily start there. 334 - > My recommended place to start right now, I mean, you might 335 - > want to go and have a look at Shopify, they have some 336 - > documentation here, but I actually recommend starting with 337 - > Google's documentation. 338 - > You might be getting a little confused. 339 - > You're like, wait, Google has documentation, but UCP has 340 - > documentation, but isn't it Google UCP?

341 - > And this is where you really have to untangle these things a 342 - > bit. 343 - > Again, UCP is the total framework. 344 - > It's this whole structure that I just described to you of all the 345 - > capabilities and every use case that is possible. 346 - > Google has their own implementation of UCP, which is 347 - > in like Google AI mode and Gemini and more places.

348 - > It'll be coming to more places. 349 - > And go, it's also manifest in Google Organic Search to an 350 - > extent. 351 - > So this is this is what Google's scope or implementation is. 352 - > And I would say start there because it gives you some 353 - > boundaries and it is right now the biggest opportunity in terms 354 - > of UCP.

355 - > Because we're so early here. 356 - > This is another thing about impatience that comes up. 357 - > Um this will be big in the next years. 358 - > It's going to take time.

359 - > Uh that it will depend on more implementations and use cases 360 - > coming forward. 361 - > It will depend on consumers adopting those implementations 362 - > or use cases. 363 - > This is not going to happen overnight. 364 - > Uh and and the, you know, for the people telling you that 365 - > you're missing out right now, there's a little bit of truth to 366 - > that.

367 - > Because I guess if you the things that you need to do to 368 - > succeed in UCP and your other visibility and AI services, like 369 - > it's all interconnected and it's all healthy and it's good to do. 370 - > All of it is valuable to do. 371 - > Um so there's it can't hurt you to start working on this, but 372 - > it's also perhaps not as necessary right now as some 373 - > people would say, or as this organizational AI mandate would 374 - > imply. 375 - > Um, you know, I see things jumping to the top of roadmaps.

376 - > This happens in every client call that I'm in, just about 377 - > these things are jumping straight to the top of the 378 - > roadmap. 379 - > And you need to not forget about the world that exists today and 380 - > business as usual, and what are the optimizations you should 381 - > have been doing last year or the year before as well. 382 - > Like it just needs to be balanced. 383 - > So sorry, I went a little off topic there.

384 - > Um, but let's get into the implementation a little bit. 385 - > Yeah. 386 - > So I would start with I would start with Google because again, 387 - > AI mode, it it they're the most advanced. 388 - > Like the Agentic Commerce features in Chat GPT, they got 389 - > rolled back.

390 - > They're all on pause right now. 391 - > Doesn't mean they're going to be paused forever. 392 - > Um maybe they'll come back again. 393 - > Let's see what happens.

394 - > Um, but like what perplexed Or what other services are you even 395 - > interested in right now? 396 - > Amazon is its kind of its own thing. 397 - > So right now, Google is basically the show. 398 - > And that's why I would focus on that.

399 - > To talk about the generalized infrastructure here of UCP and 400 - > implementing it. 401 - > There are a few, a couple of core steps you need to do on the 402 - > infrastructure side. 403 - > You have to publish something called a discovery profile. 404 - > And I believe this is one thing that Shopify, for example, is 405 - > pretty good at automating, but it still might not be perfectly 406 - > perfectly plug-and-play.

407 - > But basically, you have to host this discovery profile. 408 - > And it does things like say which UCP versions do you 409 - > support? 410 - > And it does also exactly what we talked about earlier. 411 - > Which capabilities do you have enabled?

412 - > For example, catalog, cart, checkout. 413 - > And this allows the AI, when it meets your business, to 414 - > negotiate and see what you have in common. 415 - > That's all done through the discovery profile. 416 - > There'll also be your transport endpoints, like what we talked 417 - > about, this REST API or MCP, whichever one, accepted payment 418 - > handlers.

419 - > This is all there. 420 - > So the next thing to do here would be to choose and secure 421 - > those transport bindings. 422 - > And this is how the platforms and shopping agents talk to each 423 - > other. 424 - > So as I mentioned, it's all supported.

425 - > A REST API is probably the best choice for a lot of developers 426 - > who want to go in this direction because they're probably already 427 - > familiar with REST API. 428 - > It's not new. 429 - > MCP is new. 430 - > And yeah, there's maybe pros and cons.

431 - > Again, I'm not a developer, but I suspect that they'll that REST 432 - > API will have a lot of gravity because of its it's it's it's 433 - > already so dominant. 434 - > And by the way, this seems to be also what Google prefers. 435 - > If you go in Google's documentation, they want you to 436 - > work with REST API. 437 - > So another reason.

438 - > If you go in Shopify's documentation, they're focused 439 - > on MCP. 440 - > And it's like, great, I love when mom and dad are fighting. 441 - > You guys develop UCP together. 442 - > And but it's, you know, maybe that's the beauty of it as well.

443 - > Because again, this universal thing, it I guess it doesn't 444 - > really matter because this the system is it's all it it's all 445 - > supported. 446 - > Anyway. 447 - > Then you need to configure your payment handlers. 448 - > So this is specifying the payment methods that you accept, 449 - > um, like which cards or wallets or payment processors.

450 - > Um if you are accepting raw credit card data, uh, I mean, 451 - > yeah, then you you probably know about this anyway, but there's 452 - > like these 12-step frameworks that you need to comply with. 453 - > Um, it's called PCI DSS compliance. 454 - > I don't, I don't, I don't know what that is exactly, but that's 455 - > the 12-step framework that you need to do. 456 - > And uh probably you're delegating this to a payment 457 - > provider, though, um, like Stripe, PayPal, whoever.

458 - > So um that is at a generalized level with the with the 459 - > infrastructure. 460 - > And that's there, I guess the other big topic that I didn't 461 - > discuss right there is to do with identity linking and OAuth 462 - > and stuff like this. 463 - > And and uh the reason I don't want to talk about identity 464 - > linking is because I'm based in Europe, and I think this is 465 - > gonna be a big sticking point based on, you know, what's going 466 - > on over here with GDPR and privacy regulation and all that 467 - > kind of stuff.

468 - > And I think there's gonna be a lot of extremely interesting 469 - > news headlines to read in a while. 470 - > So I'm not commenting on that. 471 - > That's for like a legal team. 472 - > Uh I'm I'm not a developer and I'm definitely not a lawyer.

473 - > Okay. 474 - > I I recognize my my boundaries. 475 - > So then there's implementing capabilities. 476 - > Uh and the the big three for you to kind of worry about, I think, 477 - > would be catalog, checkout, and order.

478 - > There's also carts, but Google is building universal cart, even 479 - > the catalog part, like this is again why I say start with 480 - > Google. 481 - > Catalog, what is this? 482 - > It's um these functions about searching your catalog, getting 483 - > products which allow the AI agents to discover and fetch 484 - > things like your pricing and their real-time product 485 - > availability. 486 - > Um that's that is pretty much handled by Google Merchant 487 - > Center.

488 - > Google Merchant Center has that stuff in there for you. 489 - > And this is a huge advantage that Google has from a market 490 - > standpoint, because they have, I don't know how many tens of 491 - > millions of merchants or however many there are in that in 492 - > Merchant Center already, whereas OpenAI has to start from scratch 493 - > and build up their merchant supply. 494 - > But it's also, yeah, it it makes life a lot easier. 495 - > Uh then there's checkout, and there is some work to do here 496 - > because and again, this can this will depend partly on your 497 - > e-commerce platform and what they're gonna do for you.

498 - > In the longer term, I think more and more of this stuff will 499 - > become plug and play, as it should. 500 - > Um, it shouldn't be rocket science to do this. 501 - > And in I don't know, I don't want to put a time horizon on 502 - > it, but in the intermediate term, it won't be rocket science 503 - > anymore. 504 - > Um but the the core job of checkout is there are these uh 505 - > create checkout, update checkout, complete checkout 506 - > fields.

507 - > You can imagine that's that's just what an a like it for the 508 - > AI to be able to do this, they need to know what's going on. 509 - > And so this needs to be continually updated during the 510 - > session. 511 - > Um, and this can be things like requires escalation or flags 512 - > like incomplete, ready to complete. 513 - > So this is all in the documentation.

514 - > And one of the reasons that the documentation is so long is that 515 - > uh they will duplicate this for every transport, for example. 516 - > So the content becomes very duplicative because they have to 517 - > provide a slightly different documentation depending on what 518 - > you're doing, since they're all supported. 519 - > But again, with Google, you just have to focus on the REST API 520 - > documentation because that's what they're looking for from 521 - > you.

522 - > Yeah, then on order, it this is basically supporting webhooks 523 - > and so that you know the status, the order status can be 524 - > retrieved or understood. 525 - > Like is it is it shipped, is it in transit, is it delivered, 526 - > what these kind of updates so that an AI platform is able to 527 - > provide that. 528 - > Um and as I mentioned, my biggest tip is to start with 529 - > Google. 530 - > Um they they really just to I'm just gonna read through their 531 - > bullet points what they want here.

532 - > Uh they need you to prepare your merchant center account, which 533 - > is your shippings, returns, and product feed. 534 - > So that's not rocket science, I guess. 535 - > Uh they want you to set up your business with Google Pay until 536 - > antitrust regulators come in. 537 - > Um but you can, yeah, there's other payment uh options are 538 - > supported here, but yes.

539 - > Um then you need to publish your UCP profile. 540 - > That's what I talked about before. 541 - > That's also pretty okay, I think. 542 - > Then you need to complete the native checkout integration, 543 - > which is also what I just mentioned, these three core rest 544 - > endpoints for session, creation, update, and completion.

545 - > Then you need to sync the order status, and that's the last 546 - > thing that I checked talked about there. 547 - > So really, Google has one of the not to say that it's that it's 548 - > easy, the team's gonna have to learn how to do this. 549 - > There can be challenges. 550 - > I'm not trying to trivialize it.

551 - > But if you go check out Google's documentation compared to what 552 - > you might hear online, like it's somewhere in the middle because 553 - > no, it's not as simple as just activating a single field in 554 - > your merchant center, but it's also not as complicated as 555 - > reading through 300 pages of documentation. 556 - > So that's the message I want to leave you with. 557 - > This is doable, but you know what? 558 - > I still have a little bit of time to keep recording.

559 - > And if you're interested and you want to keep listening, let's 560 - > keep going further, okay? 561 - > Let's talk about native commerce attributes in Google Merchant 562 - > Center. 563 - > So, and and this would be where the false claim comes from that 564 - > all you need to do to go live with Google UCP and be there for 565 - > universal card and native checkout is to populate the 566 - > native commerce attribute. 567 - > Yes, you do need to do that, but it's not the only thing, sadly.

568 - > Um so the the native commerce attribute basically is a per 569 - > product attribute in your feed that you need you need to 570 - > populate with a yes or no value. 571 - > It's actually a zero or one value, but this this determines 572 - > if the if that product is going to be allowed to have native 573 - > checkout. 574 - > And um not all of your products are eligible for native 575 - > checkout, perhaps. 576 - > So that's one thing you have to check.

577 - > Um and then there might be strategic considerations as 578 - > well, like because not everyone's perfectly happy about 579 - > AI or like these surfaces. 580 - > Basically, Google is becoming a marketplace here. 581 - > It's really a change in the relationship that we've had with 582 - > Google, where they were a traffic acquisition source. 583 - > Um, we got to have all the first party data on our websites, and 584 - > we got to own the customer relationship.

585 - > And now they get all the fun first-party data, and we feel 586 - > like we're just order takers. 587 - > This is a potentially a risk. 588 - > So you can decide on that kind of a basis. 589 - > You know, maybe there's think of this, maybe there's some 590 - > products you would list on Amazon, but not others.

591 - > That starts to be a little bit the relationship with Google 592 - > here, potentially. 593 - > Um, but as I mentioned, not all products are eligible. 594 - > So there's a, you know, you can read the comprehensive list, but 595 - > it's pretty common sense. 596 - > It's stuff like if you have custom products, if you have 597 - > age-restricted products, if you have subscription products, none 598 - > of this is supported by UCP yet.

599 - > It might be in the future, but UCP is not playing the age 600 - > verification game. 601 - > Their checkout is too simple right now for subscriptions and 602 - > stuff like that. 603 - > So there are some eligibility things you need to think about. 604 - > But otherwise, if you feel like it, you can opt in your whole 605 - > product feed and and see what comes back as this comes to life 606 - > in more markets, by the way.

607 - > So, yeah, then there is a consumer notice attribute that 608 - > you may need to populate. 609 - > And this would be things like if you, if there are, if there are 610 - > safety or legal declarations that you make in your checkout 611 - > on your website, those also need to be in the checkout inside of 612 - > AI Mode or Gemini or wherever. 613 - > Um, and so this is just a place to populate that. 614 - > Uh, and then there is a merchant item ID, which is a little 615 - > boring but not unimportant.

616 - > If you have a uh a different item ID in your checkout than 617 - > what is in your feed, that's a that's a mismatch and that's 618 - > confusing. 619 - > So uh the merchant item ID is a unified ID that you provide as a 620 - > supplement to that. 621 - > Speaking of the word supplement, all of these should be delivered 622 - > via your supplemental feed. 623 - > Google is is recommending that right now.

624 - > Just a couple other notes. 625 - > Again, to be eligible, your products need to be in stock 626 - > with complete shipping and tax setups. 627 - > And um you'll need a return attribute. 628 - > That's not a new attribute, but your return attribute or return 629 - > policy label at the merchant center level.

630 - > One or both of these need to be filled out. 631 - > That's what you need to do to for your products. 632 - > Like if you have set up your website, which we've gone 633 - > through in a high level of detail, but I hope you 634 - > understand it a bit better. 635 - > And if your feed is prepared in the following way, um, and if 636 - > native checkout is available in your market, then your products 637 - > can start appearing for native checkout.

638 - > That's the next thing is about optimizing all of this and what 639 - > happens next. 640 - > And so I want to start, I want to tease out a theme here 641 - > because I think I mentioned it earlier, but UCP is also just 642 - > useful infrastructure. 643 - > And in some ways, it's bigger than uh the agentic stuff or the 644 - > AI-assisted stuff. 645 - > Because right now, there are already buy buttons being 646 - > populated in Google organic results on the Google search 647 - > engine result page, that and it has nothing at all to do with 648 - > AI, but it's infrastructure that can be used.

649 - > And so this can power other experiences besides just the AI 650 - > stuff. 651 - > So, because you you might be an AI skeptic, consumers will never 652 - > do this. 653 - > This is never going to actually be that big. 654 - > You could be totally right, but it doesn't matter because the 655 - > infrastructure will also support existing experiences.

656 - > And so, in that regard, you might remember buy on Google. 657 - > This achieves a very similar result to buy on Google. 658 - > And buy on Google had nothing to do with AI. 659 - > Chat GPT didn't exist back then when they were rolling that out 660 - > and so on.

661 - > AI mode didn't exist back then. 662 - > Um, so this is something that's much broader. 663 - > And also, as I mentioned before, that is the bare minimum to be 664 - > eligible. 665 - > But you haven't done anything at that point to optimize your 666 - > discoverability.

667 - > And so this is the next horizon here. 668 - > And actually, for some of us, like in Europe, it's the first 669 - > horizon because native checkout isn't here yet, but we can 670 - > already start working on our discoverability in AI services. 671 - > So that's where it's complicated to wrap your head around because 672 - > like UCP is bigger than Google and it's even bigger than 673 - > shopping. 674 - > And then inside of that, there's AI mode and Gemini and stuff.

675 - > Um, but UCP is also overlapping with what we already know. 676 - > And then there's like what I'm about to talk about, some of the 677 - > feed optimizations that you can make, which support what happens 678 - > in UCP or like in AI mode, it supports also your visibility in 679 - > search. 680 - > And it's just there are kind of, I don't know, I'd have to like 681 - > Venn diagram at one time or build some kind of a diagram for 682 - > it. 683 - > But I, you know, I get funny questions like when is UCP 684 - > coming to Europe?

685 - > I'm like, UCP is already here. 686 - > You know, Zolando is building stuff with UCP, they're building 687 - > their own implementations or experiences or use cases with 688 - > that. 689 - > The question people are trying to ask is when will native 690 - > checkout come to AI mode in Google? 691 - > That's what they're actually asking.

692 - > And they say, when is UCP coming here? 693 - > And I don't blame them, but these are like different but 694 - > overlapping things. 695 - > I hope that makes a little sense. 696 - > So on to maybe our last uh topic for today.

697 - > Looking at the time, I just want to talk a little bit about 698 - > conversational attributes. 699 - > We've talked about them before on this show, but here we are 700 - > with me now. 701 - > Maybe you're new to the podcast. 702 - > Uh let's just get into it a little bit, okay?

703 - > Um, there are now six of these conversational attributes 704 - > available globally. 705 - > So that means even here in Europe, where I am, um, you 706 - > don't have to wait for all that. 707 - > You can start improving this stuff now if you want. 708 - > And these are things like the whole point of this, there are 709 - > these new ad formats that are coming to shopping, and not even 710 - > just to like native AI ads inside of AI mode, but there are 711 - > new shopping formats that will be coming to the classic Google 712 - > SERP.

713 - > And there are other things that Google will be doing, like just 714 - > writing new titles for you. 715 - > And so even if, again, if you don't believe in all the AI 716 - > stuff or you think it's going to take forever to get here, um, 717 - > these things can just help you have better Google shopping 718 - > titles that are highly relevant to the query that has been 719 - > searched, for example. 720 - > So the the first six are QA, document link, related products, 721 - > item group titles, variant options, and popularity rank.

722 - > And the first thing I want to say about all those is that five 723 - > of the six came from OpenAI's feed spec. 724 - > The only one that, or maybe, and maybe I'm wrong, document link. 725 - > I'd have to check. 726 - > Maybe that came from OpenAI's feed spec as well.

727 - > I'm not sure, but uh so what Google is doing right now is 728 - > just closing any gaps with what OpenAI has tried to offer or 729 - > could offer or stuff like that. 730 - > Uh the OpenAI is activating their feed again, not for 731 - > organic product discovery in Chat GPT, but for the product 732 - > listing ads that they're starting to build and sell. 733 - > Um but Google doesn't want to allow any daylight between 734 - > OpenAI's feeds back and theirs or Meta's feeds back and theirs.

735 - > Um and by the way, I think OpenAI did a good job writing 736 - > their feed specifications. 737 - > Uh but just to highlight a couple of my favorites here, 738 - > okay? 739 - > Uh question, question and answer, you can add up to 30 740 - > FAQs. 741 - > If you have FAQ data, then this is great.

742 - > That's an asset that you can use. 743 - > If you don't have FAQ data, this could be the wrong place to 744 - > start, of course, because what do you want to do? 745 - > How do you want to, how do you do you just want to yeah, that's 746 - > a big, a big ask. 747 - > Um, but document link even kind of gets privilege over FAQs.

748 - > It's Google's preferred preferred source here. 749 - > There seems to be an implicit hierarchy between these. 750 - > I'm trying to map them out. 751 - > Sometimes Google likes redundancy between information, 752 - > sometimes they don't.

753 - > And they also seem to have preferred sources. 754 - > So document link, imagine you have a product detail page, and 755 - > there are links on there to a spec sheet, instruction manual, 756 - > something like that. 757 - > Google's just gonna read that. 758 - > Their AI is gonna read that.

759 - > You can provide that to them. 760 - > Um, and I think a reason this is a preferred source from for them 761 - > is because it comes straight from the manufacturer. 762 - > There's very little room for error or loss of information 763 - > there. 764 - > Um, because something that has been kind of translated from the 765 - > manufacturer's manual to your FAQ, something might get lost 766 - > along the way or get wrong.

767 - > So I believe that's why this is their golden standard of 768 - > verified, detailed information. 769 - > And if you have these, of course, it's not relevant for 770 - > all categories. 771 - > Like, do you have spec sheets on your I don't know, like, but for 772 - > electronics, for example, it's a great, great source. 773 - > Um and then I ah, yeah, I won't call it related product too.

774 - > I love this. 775 - > You can specify things like part of a set. 776 - > Are there required parts? 777 - > Are things frequently bought together?

778 - > Is there a substitute? 779 - > Are there alternate brands? 780 - > I love that. 781 - > I mean, yeah, there's a data challenge here to deliver all of 782 - > that stuff.

783 - > This helps the AI support your average order value. 784 - > This helps do things like upselling and cross-selling, it 785 - > helps ensure a good customer experience to make sure that if 786 - > they bought some like part A and they needed part B with it, that 787 - > they got part B. 788 - > So I'm a big fan of that one too. 789 - > And then the lastly, yeah, I'm gonna give a shout out to 790 - > Popularity Rank as well.

791 - > It's an interesting one. 792 - > It's basically you can provide the percent uh percentage rank 793 - > uh of like your yeah, your sales. 794 - > Sales, for example. 795 - > So you provide your own bestseller data.

796 - > Let's see where Google goes with this. 797 - > But they can write copy like the number one bestseller or in the 798 - > whatever kind of copy like that. 799 - > And we know from like that's a label on Amazon that drives a 800 - > lot of volume, I think. 801 - > And so I'm optimistic about that as well.

802 - > One more, nah, yeah, I'm gonna keep you a little longer. 803 - > Sorry, it's gonna be a long episode, but there's just so 804 - > much to cover, you know? 805 - > And I want to do it justice uh because I think there's a lot of 806 - > shit content about this stuff or an absence of it out there. 807 - > And I'm I'm doing my part, I'm trying to do my part.

808 - > You can yell at me and say I did a bad job, but I'm trying. 809 - > So uh a couple of attributes that have been around for years, 810 - > but are finding newfound life. 811 - > These are product highlight and product detail. 812 - > And um again, speaking of Amazon, this was something that 813 - > Google introduced where I think they wanted some of their some 814 - > of the experiences that they offer with uh the ads and free 815 - > listings and stuff like that.

816 - > They wanted to better rival Amazon's experience. 817 - > You know, people like those skimmable bullet points on the 818 - > product detail page or Amazon, for example. 819 - > So the product highlight is a skimmable list of benefits, 820 - > basically four to six bullet points. 821 - > That's where you're selling your product.

822 - > Product detail is where you're telling about your product. 823 - > And so that's more like spec oriented. 824 - > But, you know, we just mentioned there's the document link. 825 - > This can be a source.

826 - > And Google in the past, like in when they launched it in 2019 or 827 - > 2020, they had they imagined that it's very spec-oriented. 828 - > You have a camera with this lens and this blah, blah, blah, blah, 829 - > blah. 830 - > Um, but now they are kind of inviting you to submit what they 831 - > call a wide variety of product facts. 832 - > So if there's a flavor or an occasion, uh they they want, 833 - > they're really opening that up as a place for you to add 834 - > details that can be reflected in the ad copy or in AI talking 835 - > organically about your product.

836 - > This is all going to be important there. 837 - > Last thing that I want to talk about, I talked a couple 838 - > episodes, so I won't recover this, but a hidden feature of 839 - > Google Ads, go look up that episode about the new share of 840 - > voice metric that they're rolling out in Merchant Center 841 - > to help you understand how visible you are in AI surfaces 842 - > like Gemini. 843 - > But what I have neglected to mention, there's something also 844 - > coming to Merchant Center called UCP Analytics, which is a bit 845 - > what it sounds like.

846 - > It's probably not what you would love to get, because as I said, 847 - > this is now Google's fun first-party data instead of your 848 - > fun first party data, but they're going to give you some 849 - > metrics in there. 850 - > Um, the number of clicks to buy, that's the number of times 851 - > shoppers clicked on the UCP buy button. 852 - > Purchases, the total number of successful transactions 853 - > completed, purchase rate. 854 - > This is basically one divided by the other.

855 - > It's the percentage of clicks to buy that resulted in a 856 - > successful purchase. 857 - > So purchase rate is your conversion rate for native 858 - > checkout. 859 - > Going to be a good one to watch. 860 - > Average order value.

861 - > Maybe you can see if that goes up when you add um you know 862 - > related products and things like that. 863 - > Cancellations, how many got canceled or refunded? 864 - > Account linking, this is basically your customer match 865 - > rate. 866 - > We talked earlier shortly about identity linking, but it's an 867 - > important concept because identity linking will power 868 - > things like um people getting their member pricing inside of 869 - > AI mode and not just on your website.

870 - > So account linking, they're gonna have like the six the um 871 - > how many times they tried to link and the success rate and 872 - > stuff like that. 873 - > So that is to me comparable to your customer match rate and 874 - > advertising. 875 - > Um, and your top-selling UCP products. 876 - > So a ranked list of the products driving the most sales through 877 - > your integration.

878 - > Um is it everything we could possibly want? 879 - > No, maybe there's more coming, but I think it's important 880 - > because you know, I haven't been perfectly satisfied with this 881 - > thing where it's like, oh, you remain the merchant of record. 882 - > I mean, yeah, no shit. 883 - > I have to do all the customer service.

884 - > Thanks. 885 - > Like, you go, like, um, yeah. 886 - > Um, so give me something back, and this is a first step in that 887 - > direction. 888 - > Um, but let's just recap it because I'm out of time and it's 889 - > been a long episode.

890 - > I hope you made it through, and I hope that UCP and some of the 891 - > things around it are a little more clear to you. 892 - > Because as I said, I think there's a lot of urgency here 893 - > and not a lot of understanding, and that's a very uncomfortable 894 - > situation for everyone. 895 - > Um so again, my key uh takeaways here don't boil the ocean, start 896 - > with Google. 897 - > It's just the biggest opportunity, it only makes 898 - > sense.

899 - > And they'll give you, they'll tell you exactly what you need 900 - > to do and focus on so you're not distracted by everything else 901 - > that could be. 902 - > That's step one of the journey. 903 - > Step two is improving your feed so that you can improve your 904 - > discoverability, not just for UCP-powered experiences, but in 905 - > general. 906 - > It's just something you should do anyway.

907 - > So, hope that makes sense. 908 - > Had a fun time talking this through with you. 909 - > And um, yeah, next week we'll be back in studio with Chris, 910 - > everything back on normal. 911 - > So looking forward to that.

912 - > Thanks again. 913 - > This has been another episode of Growing E-Commerce, brought to 914 - > you as always by Smarter E-Commerce, also known as SMEC. 915 - > You can learn more by visiting Smarter-e-commerce.com.

916 - > And if you enjoyed this episode, or if you think I got it totally 917 - > wrong, um, and you're angry at me, then you know, drop a 918 - > comment. 919 - > We love to hear from you. 920 - > Give us a review, but only if it's a good one. 921 - > Thanks.

922 - > We'll see you next time.