How GenAI Is Changing Search and Engagement-Focused Ecommerce (2 of 2) (Tech Strategy)

Posted on by

In Part 1, I laid out basic frameworks for:

  • Attention vs. infrastructure ecommerce
  • Search-focused ecommerce

Which brings me to the point, which is interaction and engagement-focused ecommerce.

Pinduoduo Is a Great Example of Interaction and Engagement Focused Ecommerce

Amazon (and Alibaba) were mostly doing the previous strategy when Pinduoduo broke out in 2016-2017. And it really had tremendous growth. Colin Huang launched PPD in 2015, rocked Alibaba in 2016 and took it public in 2018. Colin is on the short list of China CEOs you should follow closely.

Pinduoduo was really interesting in its early years because it came at ecommerce from a completely different angle. It was a real contrast to Amazon, Alibaba and search-focused ecommerce. It was 100% focused on interactions and engagements. If Amazon was a digital Walmart, Pinduoduo was a digital carnival.

Here is how Matthew described the difference in his white paper.

And the primary KPI for interactive ecommerce isn’t GMV. It’s DAU and MAU, which you can see in PDD’s early numbers.

Get Our Free “Digital Strategy and CX Insights” Emails

–

You also focus on satisfaction (NPS), happiness (people really enjoy using the app) and delight. Note: Delight is a combination of satisfaction and surprise.

And sure enough, when you looked at GMV, MAU and DAU, Pinduoduo was completely different than Alibaba and JD. It had dramatically larger DAUs but a ridiculously low GMV per MAU. People were coming to the app frequently to have fun. But were only occasionally buying something. Sort of like going to the carnival. Which is very different than going to Walmart.

It also helped that the smartphone had completely replaced the PC in China at this point. PCs are only used a few times per day (at home, at office). While smartphones are used all day long. So PDD’s focus on high frequency usage made a lot more sense for smartphones.

PDD also really changed the discovery process.

That to me is what is most interesting. Instead of customers doing an active search (which takes effort), they mostly just scroll a newsfeed, watch videos, or play games. It is a push model with users just passively (and often mindlessly) consuming.

So, they discover products in very different ways. Customers are driven more by curiosity. By emotion. And they get surprised. They’re not walking up and down the Walmart aisles with their shopping list. They’re exploring the carnival and seeing what’s there.

Another analogy Matthew uses is a sushi train versus a big restaurant menu. Instead of a massive menu of items to search through, you just sit as the sushi bar and items come by on the conveyor belt (a physical newsfeed). You get surprised by things. You grab what catches your interest impulsively (you have to move fast). And you make lots of small purchases.

Engagement = Recommendations + Entertainment + Community

So how do you get interaction and engagement in ecommerce?

You definitely want lots of features and games. The UI has to be lively and dynamic. And you always need to have new things (especially promotions). You want to customers to always be checking in to see what’s new.

The Whitepaper argues you need three things:

  • Recommendations. This is focused on passive consumption, not active searching. So that means machine learning-powered newsfeeds are key. Basically, just like TikTok.
  • Entertainment. That means features, games, and videos.
  • Community. This is about tapping into group behavior. You want active communication and sharing between users. And on social networks. You also want livestreaming with friends. PDD got really lucky by also being able to do group buying.

That’s a pretty good framework.

I look for two archetypes in interaction-focused ecommerce.

  • Type 1: Newsfeed based. That’s TikTok. That’s the sushi train. New items are continually presented to you. TikTok Shop is a good example of this archetype. It’s about video entertainment with lots of little purchases sprinkled in.
  • Type 2: Carnival based. That’s early PDD. It’s lots of games. It’s much more active. There is sharing and interacting with others.

The breadth of products is another interesting dimension to this.

For search-focused ecommerce, more products is better. The more you can offer a long tail of unique products, the better.

But early PDD (Type 2) was a lot about offering really cheap prices. It achieved this by doing C2M, which meant getting deals from the factory for a smaller number of products. That’s how they got the amazing deals that would show up in the newsfeed. That were in mostly in a small number of commonly used items.

In contrast, TikTok Shop is about enabling lots of individuals sellers / content creators to sell their goods. That are offering a huge, long-tail of unique content and product types. Which the algorithm then uses to match to unique user interests.

Keep this in mind. This is one of the big use cases for AI ecommerce below.

A Quick Aside About Temu

I haven’t really mentioned PDD’s wonder child Temu in all this. They launched Temu in September 2022 and shocked pretty much everyone. Like with PDD, it had shockingly low prices and a hyperactive user interface. Which it accomplished by connecting foreign customers directly to Chinese manufacturers (C2M). And by cutting out intermediaries. Customers love the cheap products and a fun interface. PDD also spent a fortune on marketing and promotions in the first years.

However, Temu also has a problem.

It takes 5-7 days to deliver from China to foreign countries. Which customers don’t like. Why would a customer wait a week when you can get it from Amazon tomorrow?

Cross-border ecommerce needs to give customers a compelling reason to wait. I think Temu accomplishes this in a couple of ways:

  • It does have really low prices.
  • It’s more fun.
  • A lot of these are impulse purchases. You’re buying because it’s just fun on the app. You just bought something on impulse. So, you don’t need to get it immediately, like you would with a needed purchase.
  • Temu has a big selection of long-tail products.

This last reason is why I shop on Temu. I use it when I need something unique and I can’t find many options on domestic ecommerce sites. I know Temu will have way more choices in smaller categories because it is tapped into China’s manufacturing base.

Anyways, this is an interesting variation on the long tail vs. common products question.

Ok. Let’s get to the last point.

Generative AI Is 100% Impacting Search Focused Ecommerce

I recently wrote an article about my visit to Alibaba Cloud. I wanted to understand how AI is changing ecommerce. Specifically, how it is changing the two strategies I just discussed. Here’s the article. It has a lot of content. But is not a great read.

Alibaba has long talked about combining shopping with entertainment and community. Pretty much what I just talked about.

And a great example of this is livestreaming in ecommerce. This is how you showcase new products. With lots of discounts. The host can answer questions in real time. Plus, other viewers also chat with each other. It’s interaction focused ecommerce. And it is a big use case in GenAI.

Use Case 1: GenAI is Democratizing Ecommerce Livestreaming

Pretty much everyone will now be able to live steam.

You sit at a desk and talk to your phone. You just focus on selling your products and chatting with customers.
The AI does everything else. It adds product descriptions. It puts up games, features, text and prices. It moderates the chat. It handles the transactions. It turns out generative AI is really good at generating content (including videos).

It makes livestreaming so simple anyone can do it. Which will dramatically expand the number and types of merchants selling this way. We see CEOs now doing livestreaming. And the sales ladies at department stores. And farmers in their fields.

How does this help you in interaction-focused ecommerce?

  • It makes it more entertaining. These are fun to watch.
  • It increases engagement. It’s social. People chat and comment. You definitely need content moderators though.
  • It is more trustworthy and authentic. You trust your favorite live streamer more than a corporation (usually). And you also get trust from the other people chatting. It’s social proof.
  • It encourages impulse shopping. That is really important. The more impulse shopping you can get the better.

So GenAI is democratizing and supercharging all of this. Live entertainment with embedded shopping is powerful and becoming really easy. Everyone is going to do it.

Here’s how Alibaba talks about this.

Use Case 2: Image and Multi-Modal Search Is a Big Upgrade to Search-Focused Ecommerce

Often a customer knows what he/she wants and there is strong purchasing intent. But he / she doesn’t know the right keywords to search for it. They try a couple and keep getting the wrong results.

This is not just a problem with the user’s vocabulary. It is also a problem in the product classifications. Text-based search depends on having standardized product categories and descriptions. But a lot of products are unique. They can be one of a kind. They can be artisanal products.

Enter image search. You just photograph what you want and it searches for it. Or you can generate an AI image of what you want.

And it is evolving into multi-modal search. You put in some text, a photo and even a voice note. It combines this into a smarter search of products.

Which leads to…

Use Case 3: Conversational Search Is a Game Changer

This is my favorite one.

Forget searching with keywords. You just have a conversation with the AI. You tell it what you are looking for.

“I am going to a party on Friday night and want a new black dress.

It can come back with options for dresses. But you didn’t really tell it what you wanted. You told it about a problem you had. And it can offer solutions to that outside of a keyword search.

“Would you also like some shoes? How about accessories that match?”

Conversational search (also called guided shopping) lets you operate outside of the search parameters entirely. Instead of putting in lots of keywords, you have a conversation with AI about your problem. It then makes suggestions. This lets merchants operate way outside of the search parameters. And even to change them with a conversation.

Here’s how Alibaba describes it.

Ok. There’s a lot going on here.

But is conversational search still type of search-focused ecommerce? Or is this something new entirely?

I tend to think it’s a new thing. I’m watching for adoption and usage this year.

That’s it for today. Cheers, Jeff

———

Related articles:

From the Concept Library, concepts for this article are:

  • Interaction / Engagement-Focused Ecommerce
  • Search-Focused Ecommerce
  • Generative AI

From the Company Library, companies for this article are:

  • Pinduoduo

———-

Leave a Reply