Lazy Giuseppe Cocktail & How Generative AI Will Transform Commerce & Marketing (Part 2 - Experiences & Product Discovery)
Issue No. 6 - March 17, 2023
I hit two hundred subscribers this week! Thank you if you are among them, I very much appreciate you being here. I want to see this newsletter continue to grow - it is a sign it is adding value. Please share it and subscribe if you have yet to. Thanks in advance!
In today’s issue, I continue my series on the impact Generative AI will have on digital commerce and marketing. If you missed part one, see Part 1 - Personalization & Content - where I paired that with a great Umami Martini.
And for a cocktail, in this issue we are making a low-ABV cocktail - and look at what low-ABV means.
This Week’s Cocktail: The Lazy Giuseppe
This issue’s cocktail is what is often referred to as a “shim”. In cocktailing, a shim is a drink that has a lower alcohol content than a traditional cocktail - and certainly less than the boozy one’s I am prone to make after a great dinner.
While you may not be all that familiar with the word ‘shim’ in a drinks context, you have no doubt had cocktails that belong to this family of low ABV cocktails - including champagne cocktails, spritzes, cobblers, and many coffee drinks. Mixology is swiftly moving in the direction of low or no alcohol - as I wrote about in Issue No. 4 with the Billows & Thieves Mocktail - and there are plenty of reasons why, but of course most especially because there is less (or no) booze and that is generally going to be better for us.
Shims and other low ABV cocktails may be lower in alcohol, but they pack flavor complexity and nuance. That is often what mixologists are really after, but we do not need to get plastered to enjoy the craft. Often a shim can be made by flipping the spirit and the modifiers - think a Manhattan with mostly vermouth and only one to .75 ounces of rye. In this case, however, we are making a cocktail with no hardcore spirit, Cynar is half the cocktail. Cynar is a low ABV bitter amari made from artichokes. The sweet vermouth in this drink is also low ABV. The only modest level of ABV ingredient in this cocktail is the Meletti, but it is also still relatively low. (I put the ABV’s on the spec below so you can see for yourself.)
You might have heard the term ‘low ABV’ and not be quite sure what that means - unless you are of course into the craft beer scene, in which case you probably should take mixology more seriously (There is a whole new world of beer cocktails awaiting you!) ABV refers to ‘Alcohol by Volume’ and is a standard measurement for the percentage of alcohol found in a beverage. It is essentially the ratio of the volume of alcohol present to the total volume of the liquid. (I am going to spare you the math, because I am not that into it. Yet.)
Shims are often designed to be drunk in the afternoon - basically, so you don’t have to be. (I decided not to edit that out because I still think it is funny.) Shims are also great before dinner, especially this one as it is packed with bitters to get your appetite and gastrointestinal - um - plumbing… humming.
A PSA on Bitters - Bile! Gut bacteria! Cocktails!
Bitters have a long history of being used as digestive aids. When bitter flavors hit the tongue they trigger the production of saliva, gastric acid, and digestive enzymes - which help break down food and facilitate nutrient absorption. Bitters are of course designed to be an infusion or extraction of bitter flavor and healthful herbs, and there is a long, wonderful history and culture to explore in bitters. There are a lot of benefits to the liver, bile production, breaking down of fats, etc. - but I really think we should avoid mention of bile and fat in a cocktail blog, don’t you?
This shim is an adaptation of the “Bitter Giuseppe” found in the excellent book The Art of the Shim, by Dinah Sanders. Sanders was at the leading edge of popularizing this class of cocktail about a decade ago and she has a lot of great ideas and specs for shims in this book. The spec was originally created by Stephen Cole from Chicago’s legendary bar The Violet Hour, and Brad Thomas Parsons’ featured it in his excellent book Amaro as well.
Just because I felt like it, I spiced it up the spec for the “Bitter Giuseppe” a bit to have some fun and added some salt. I also played with the ratios and lowered the sweet vermouth and upped the Meletti a touch to get a bit more nuance in the flavors. In the end, this cocktail is surprisingly refreshing and light. And in case you are wondering, I decided to call my variation a ‘Lazy Giuseppe’ since we are also talking about Generative AI in this edition of the newsletter. (Note that Giuseppe asked me to tell you that ordinarily he is very industrious, but today he is saving time by using LLMs so he can drink more shims!)
The total ABV of this cocktail comes out to be 13.9%, essentially the same ABV of a glass of red wine, which is typically in the range of 12% to 15% - which also happens to be the bar a cocktail should really be under to be considered low ABV. So this ‘Lazy Giuseppe’ more than meets that, but to be honest, I don’t think mixology is a pursuit or craft where we should be too caught up in the details of what does and does not exactly qualify for these types of classifications and such (we are mixing after all, and let’s get to it!)
The spec, serves one:
2 oz - Cynar (ABV = 16.5%)
.75 oz - Carpano Antica Formula Vermouth (ABV = 16.5%, sub Cocchi Vermouth Di Torino or other quality sweet vermouth)
.30 oz - Meletti Amaro (ABV = 32%)
.25 oz - Fresh lemon juice (if you have any ABV in the lemon juice, discard and find a new lemon)
2 drops - Angostura cocktail bitters (let’s not worry about these)
3 drops - Infuse Bitters Tres Amigos cocktail bitters (See note on Angostura bitters)
Lemon twist or coin for garnish. (See note.)
Note: A “coin” basically refers to a slice of peel with a bit of the rind and meat of the lemon still attached. Don’t get too worked up about being precise, and a simple twist is fine if you are not sure.
The process:
Add all ingredients to a cocktail mixing glass (beaker or pitcher for substitute) then add ice and stir until well chilled. Strain into a rocks glass over ice, and garnish. (See note.)
Note: I may try shaking this one next time. Could be a shakerato opportunity here.
Analysis: The Impact of Generative AI on Commerce & Marketing - Part 2 - Experiences & Discovery
In part one of this series we explored what Generative AI is and how Generative AI will impact personalization and content in digital commerce and marketing. In part two, we will focus on experiences and product discovery.
“Just Google it.”
When Google launched in the late 1990s, it disrupted how we found information online by introducing what was at the time a radically different approach to online search - primarily its PageRank algorithm, simple and user-friendly interface, and speed.
By around 2002-2003, Google had become the leading search engine, sticking forks in Yahoo!, AltaVista, Excite, and Lycos - while others like Microsoft’s Bing could simply not keep pace with the innovation, quality of experience, and the share of search traffic Google rapidly built. Google's ascent not only impacted the search market but also changed the landscape of online advertising and the way people have accessed information and interacted online ever since. Today Google has approximately a ninety to ninety two percent share of the global online search market.
In e-commerce it is of course important to note that Amazon quickly reinforced this pattern, first by partnering with Google in a multi-year partnership in 2000 that involved Amazon using Google's search technology to power the search functionality on Amazon's website and displaying Google's sponsored ads on its product pages. The partnership helped Amazon capture a larger share of online traffic by improving its search capabilities and ensuring that users could find products more easily on its sites and in having its products index high in Google’s search results. As both companies have evolved, their relationship has shifted, with Amazon now competing with Google across a wide range of areas and certainly in search advertising - which I covered in my recent article on Amazon’s traps.
Google and Amazon established the search paradigm, but it has degraded.
Because Google and Amazon together established a pattern of user behavior that created the model of how we discover information and products online for the first twenty plus years of the web and e-commerce - which of course bleeds into offline behaviors as well. We search for stuff with keywords and then navigate. Google and Amazon search experiences - and everyone who has “fast-followed” - focus on helping users find information and products through keyword-based searches.
But the quality of those experiences is not what it used to be. Each has been corrupted in many ways by ad clutter, overwhelming choices, misleading links and information, a pay-to-play model, and commercial bias - hindering users and customers from finding the most relevant, accurate and unbiased information or products.
And while voice assistants like Siri, Alexa, and Google Assistant have already significantly changed how many discover information online - by introducing a hands-free and conversational approach - the impact on e-commerce and information search has largely been modest. (The exception could certainly be for those with visual impairment, where the impact has likely been more significant for those individuals.)
But that is about to change.
The way we search and interact online is set to change forever
If you have already played with Chat GPT directly - or through Bing - you have no doubt felt the impact on how you think about doing research online. Instead of a hopeful keyword search and then picking through links, sorting the sponsored from the not - and feeling like something is amiss as SEO jockeys manage to get irrelevant content to rank high - you get a nice neat summary. Of course, when using these you need to use your judgment and check other sources to ensure Chat GPT - and Bing, etc. - are not hallucinating (i.e. making stuff up) or basing the results on unreliable, outdated, or just plain wrong information. But the speed and quality is remarkable - as is the ability to have a conversation and ask for clarification, expansion, or adapt your queries in a conversational way.
And while these systems are not tuned to e-commerce and product discovery, the user expectation for how to engage these kinds of systems to solve research and discovery use cases is already changing. Fast.
We can expect to see these patterns reinforced - in part as investment in these systems and vertical applications continues to ramp up. According to Pitchbook, Venture Capitalists have increased investment in Generative AI by 425% since 2020 to $2.1bn in 2022, and with the press and scale Chat GPT has already achieved, we can expect that to increase in 2023 even as the venture market weathers a winter of high interest rates and fall-out from Silicon Valley Bank’s failure.
And of course the large players in search and cloud computing are all chasing this suddenly wide open market. Microsoft has announced investment US $10B in OpenAI, while Google has released Bard and Amazon and others invest heavily and with urgency. Already we see Open AI’s latest model GPT-4 is a step-change improvement over their previous release - based on a larger model with more parameters.
All that should lead to rapid evolution of the quality of the core LLM technology and experience of the users. All kinds of applications will be built upon these solutions and everything from searching for the right daycare provider to where to get a tire repaired to - you get it, just about everything - will change.
And while the e-commerce application of these public chat-based search solutions is limited today - in part because the index is not regularly updated in the same way Google’s or other conventional keyword search is done today, the experience they provide will already have an enormous impact - leading to the rise of conversational experiences online and in e-commerce.
The rise of conversational commerce - a step change in the discovery experience - is here.
What is conversational commerce? Think guided selling applications that are powered by natural language - text and voice - and of course can now be extended to a whole range of devices and interfaces. Instead of a keyword search, we will see the rise of conversational engagement becoming the norm. It is hard to say how long this will take - and certainly we have all been wrong before - but I think it is fair to say two to five years.
Instead of a search like: Blue summer dress
We will see searches like: I am looking for a blue summer dress that I can wear to brunch.
And instead of interacting with filters and adding keywords we will see a conversational approach like: Something that looks appropriate in the Silver Lake area of LA. Something I can also wear to work. I like the one on the left, do you have more like that?
Interactive experiences like “more like this” will become increasingly expected, but instead of it being a carousel of products on a product detail page, it will be a core part of the experience. And yes, this is an extension of visual search. And while those are not new ideas or fundamental technologies - visual search first emerged in the late 2000s and early 2010s - it suddenly will work more effectively and be expected in this context by consumers. There are lots of examples of slowly develioping user experiences and technology taking a long time to hit mainstream, such as the QR Code, which took over two decades to hit the mainstream after it was invented in Japan in 1994. Now - post pandemic - everyone understands and is comfortable engaging with QR codes. (Though I will be honest, I am not a fan of the QR Code menu anywhere outside of fast-food. We should put away our phones when we sit down to share a cocktail or break bread together! I want the paper menus back!)
Whether it is via text chat or voice, we can expect a conversational approach to search and discovery to come on strong. That will open up new channels as well. (I can hear e-Com leaders everywhere shouting, “Just what we need, more channels!”) That smart and connected TV sitting in your living room - that already has voice interaction capability - will suddenly become a compelling shopping and discovery experience. In-app and site chat, SMS and messaging apps, voice, even email will support a more natural, human, conversational shopping experience.
That will work because it will suddenly feel natural, a throwback to how shopping was when you could go into a store or call a call center and speak with an actual product expert. But while human associates who know the ENTIRE catalog of products EVERYWHERE is a pipe-dream, conversation commerce systems enabled with LLMs will deliver that - and assist those humans standing in front of you in a store too.
What makes this really impactful is that these systems will scale, be automated, and be self-learning - assisted by some initial training and maybe a bit of curation. We are very close to seeing these types of solutions in the field.
For example, at Bloomreach - where I work - we already have functioning prototypes of our APIs interacting with Chat GPT to enable our industry-leading commerce search capability to be extended to have a conversation with a customer. Soon customers will be testing these experiences in the wild. (Yes, I was tempted to show an early demo here - but chose not to as the engineering team adapts to the latest LLM models.)
Guided Selling is a concept set for a renaissance
Guided selling is also not new, and I know you have interacted with some of these shopping tools on sites when you are looking for a new laptop or TV, but let’s be honest, they have not been great experiences. A big reason for that? They are parametric systems built on top of lots of if/then rules - and you get the sense of that as you interact with them. And part of the reason they have tended to suck in the past is that they were a real pain in the ass to maintain, and thus often were not. These shopping tools may have been decent when they launched on a site, but over time, not so much and consumers essentially started to ignore them unless they were forced to use them.
Generative AI and its ability to handle complex, natural interactions and turn that into parameters that interact with a powerful search index will make these types of experiences very natural to us humans - and much, much easier to maintain and manage the backend.
Conversational commerce will quickly leapfrog the guided selling experiences of the past. Old guided selling tools will be replaced, the maintenance challenges mitigated. Even more importantly, customers will get the experience of navigating choices - from configurations to customizations to options and bundles - in a much more natural way that just feels like it was always supposed to work.
The implications of these kinds of conversational experiences stretch far beyond basic retail or business-to-consumer (B2C) experiences and into complex, high consideration products - like cars, appliances, and vacations - where conversations with experts were still often the norm (though it has often felt you knew more from all the online research you did that the so-called “expert” standing in front of you.)
Beyond consumer-focused e-commerce, the impact of Generative AI powered conversational commerce extends to B2B in a big way too.
As much as we all focus on B2C, we also know that B2B e-commerce is a much larger market than B2C. The global B2B e-Commerce market was close to US$18 trillion in 2021, over five times the size of the B2C market (Source: Statista). And of course, that is despite the legacy of channel conflict that B2B sales people and leaders often put in place, or the challenging - and at times clunky - experience delivered in B2B e-Commerce. E-commerce of course already dominates in B2B areas like distribution - dominated by buying commodities, consumables, and replacement parts - where efficiency and accuracy are key, but in other more complex sales online experiences have lagged. Often that is because of the need to configure and discuss with an expert. Those experts often require years of domain experience to become good, and training and enablement are notoriously challenging to do well.
But now we are on the cusp of a revolution in the B2B e-commerce experience and you can easily see that - in everything from machinery, software, office furniture (remember that?), and other complex considered purchases - the opportunities to deliver reliable expertise, scalable built-in learning, and multi-channel and digitally enabled conversational experiences will become compelling. Just as we explore how coupling that with visual search in the examples above, you can foresee everything from finding a replacement part to outfitting your new manufacturing plant will become better (over time) supported by LLMs and Generative AI.
The impact of Generative AI on how we shop can not be overstated - but it is still early
Conversation experiences and interactions with AI-backed systems will become normalized very quickly. Not only will customers become accustomed to it, they will be frustrated if they have to revert to what feels like an old type of experience where they have to do the work all themselves. Navigating a website and typing in a search keyword will feel like sending a fax does to us today, or a telex felt twenty years ago.
And haven’t we in a sense been trying to jump the “third wall” of e-commerce for a while? TV-sized digital commerce experience and form factors will emerge. Will they be apps? Will they be something else? That I am not sure about yet. In-store digital experiences will also emerge, both for discovery but also in site content and personalization. I will cover ‘phygital’ experiences in a later issue, along with all the other dimensions of how Generative AI will impact commerce and marketing. (There are a lot, so subscribe!)
How to prepare for conversation commerce
I know I risk being called out for bias, but the one thing I am very sure of is that you will need a great search index of your products - highly attributed, using semantics and machine learning - in order to be successful. You will also need a rich understanding of your customers, and see Part 1 of this series for an exploration of that.
While the user expectations - and implications of that - mean that the existing paradigms of product discovery - search, category pages, product recommendation carousels - will fall by the wayside while the underpinning capability of rich product data and highly performant and accurate search will become even more critical to all businesses - not just those who have large catalogs.
Do not count on these LLMs to get it right. e-Commerce is, well, different - and domain specific capability, accuracy, query interpretation, integration and responsiveness are all things that Chat GPT and LLMs will take time to improve on, or to focus on in the first place.
I wanted to make sure I asked an expert, so I asked Chat GPT itself its opinion:
Me: why will chat GPT not be a replacement for ecommerce search on commerce sites?
Chat GPT: ChatGPT is a powerful language model that can assist in various tasks, including search and information retrieval. However, there are several reasons why it may not be a complete replacement for traditional eCommerce search on commerce sites:
Domain-specific knowledge: eCommerce search engines are tailored to handle specific attributes, such as product names, brands, categories, and prices. ChatGPT may not have the same level of domain-specific knowledge, which could lead to less accurate search results.
Query interpretation: Traditional eCommerce search engines are designed to understand and process user queries specific to shopping needs. ChatGPT might not always interpret shopping-related queries as accurately, especially when they involve unique or specialized terminology.
Database integration: eCommerce search engines are integrated with the site's product database and can provide real-time information, such as stock availability and price changes. ChatGPT does not have direct access to this real-time data and may not provide up-to-date information.
Scalability: eCommerce search engines are designed to handle a large number of simultaneous queries and can scale efficiently with traffic. While ChatGPT is powerful, it may not offer the same level of performance under heavy loads.
Personalization: eCommerce search engines can offer personalized search results based on a user's browsing history, preferences, and other factors. ChatGPT, as a general-purpose AI model, does not have the same level of personalization.
Legal and ethical considerations: Using ChatGPT to handle eCommerce search queries may raise privacy and data security concerns, as well as compliance with relevant laws and regulations.
User experience: Shoppers on eCommerce sites may prefer a traditional search interface with filters, sorting options, and visual product displays. Implementing ChatGPT might not provide the same level of user experience.
While ChatGPT has the potential to enhance certain aspects of eCommerce search, it is not likely to completely replace dedicated search engines on commerce sites. Instead, it could be integrated as a supplementary tool to improve search functionality and provide additional assistance to users.
So product discovery and how shoppers engage in shopping digitally is set to change forever. How exactly is TBD, but preparing now is critical.
There are many other dimensions to the impact Generative AI will have on commerce, marketing, merchandising, and service which I have not really even touched on yet. Please subscribe so I can share those thoughts with you - along with a cocktail! - and let me know if there is anything related to Generative AI you would especially be interested in my thoughts on.
Join me and meet at these upcoming events:
Commerce Experience Live - Los Angeles - March 16, 2021. We will be gathering for cocktails I get to spec, and celebrating digital commerce leaders in the best way we know how – a party in a beach house in one of our favorite cities. Hope to see you there!
Shoptalk - Las Vegas - March 26-29, 2023. Vegas Baby! Meet me for a cocktail or mocktail at Commerce After Dark Meetup or at the MACH Alliance Cocktail Reception on Monday March 27.
Modern Retail Commerce Summit - New Orleans - I have never been to this show, but there are a lot of great bars in NOLA! Sazerac anyone? Peychaud's? Looking forward to it! April 11–13, 2023
If you are looking for me online, you can find me here, here, and here.
Be well, be safe, and here is to good business! Cheers! - Brian