Four Forces Reshaping Marketing

As marketers adapt to profound trends—the arrival of generative AI and large language models (LLMs), the convergence of adtech and martech, and greater attention to privacy issues—the right technology stack is essential to great marketing.

In their new report, Modern Marketing Data Stack 2023, the marketing technology company Snowflake analyzed what its customers use most often to enhance marketing data analytics and more accurately measure performance and ROI.

1. ADTECH/MARTECH CONVERGENCE: WALLS COME DOWN

The technologies that support advertising and marketing activities, respectively, grew up independent of one another, just as advertising and marketing teams were usually separate, or at least not tightly integrated. The ecosystem of tools to manage paid channels was therefore separate from the tools to manage owned channels.

To manage paid channels, we got ad servers, ad exchanges, demand- and supply-side platforms, paid and social management platforms, and one set of measurement tools. To manage owned channels, we got CRMs, email marketing systems, SEO, and another set of measurement tools. Over the last decade or so, the two disciplines and their respective tools have increasingly converged. This convergence has been driven largely by a number of factors:

• The frenetic and fragmented media landscape means that marketers have to work harder to both identify and reach their target audiences. Guiding a new customer from first impression to loyal fan involves many touches across different stages of the journey. Doing it right requires a precise understanding of who your target is (when the data may be siloed in various martech or adtech tools) and where and when to deliver your message (which may be complicated if your marketing and advertising campaigns are not coordinated).
• There’s a lot of data out there to help with exactly that challenge. But that data is largely fragmented, too—siloed in specialized applications. A business may have a lot of information about a potential customer’s interests and behaviors scattered ineffectively across a dozen solutions used by the advertising, marketing, and sales organizations.
• Inefficiency costs money. Think of the technologies connecting a marketer who wants to advertise and the publisher who has the ad inventory as a supply chain. Each link in the chain has costs. A marketer with a dollar to spend on digital advertising will use multiple solutions to unify, enrich, and activate data on a potential customer on the way to that ad platform, and each step costs money that could be spent on more impressions. These technologies improve the quality of the campaign, but economic pressures drive marketers to get there as efficiently as possible—and drive vendors to improve the power of their own solutions to deliver value.
• Business leaders are always looking to cut expenses. Especially in chaotic economic times, the message is “do more with less” and quantify the results of every expenditure. That means that almost everything can be seen as trimmable fat, even when marketing has been charged as a growth driver. The need to better acquire, retain, and upsell customers while showing greater ROI and efficiency is exactly why a focus on a well-understood target audience is indispensable. In addition to the economic pressures to contribute more value, and prove it ASAP, the art and science of marketing is being reshaped by the rapid convergence of adtech and martech, the emergence of LLMs as the hottest area of AI, issues around data privacy, and the imperative to focus on a unified approach to data as an antidote to silos, security/governance headaches, and costly inefficiency.

What makes the years-long trend important now, and necessary to call out in the 2023 Modern Marketing Data Stack, is that we’re moving from the idea of convergence to actual implementation—and having the right tech foundation can accelerate that evolution. The larger worlds of adtech and martech are still far apart in significant ways, but for many of the customers we see in the Snowflake Data Cloud, the wall is beginning to come down.

Identity providers in particular want to bridge the adtech/martech gap, with vendors increasingly offering such classic martech practices as creating unified customer profiles, delivering personalized customer experience, and measuring the success of marketing campaigns to understand the full customer journey. It’s worth noting that many organizations within this category, such as LiveRamp, Neustar, and Experian, are building identity, enrichment, onboarding, activation, and other capabilities within Snowflake as Snowflake Native Apps (in preview as this report is published), enabling these processes to happen faster and with greater security. Essentially, this means there is no need to copy and move data between any of these solutions and the Data Cloud, adding agility and governance while reducing latency.

A broader example: Traditionally, adtech uses cookies and third-party data segments of cookies to profile audiences, while martech relies on lists of email addresses. As third-party cookies shuffle off this mortal coil, email lists increasingly power paid channels. That’s an instance of the adtech/martech convergence playing out within the customer data platform (CDP).

Each of these factors drive marketers to find actionable insights by creating a unified approach to a potential customer, from first contact through purchase and beyond. The goal is a clearer and more measurable picture of the buyer’s journey, and an ability to focus not on siloed email or ad activity, but on the overall customer experience as a seamless journey that’s persistent and consistent.

As publisher platforms and marketers have wrestled with these challenges, there’s another factor, driven largely from the adtech side: Rising privacy concerns have created serious disruptions for both martech and adtech, though challenges around the thirdparty cookie have been a major headache for adtech in particular. Advertisers rely on third-party data, such as cookies, to target their ads. The slow death of third-party cookies, increasing regulation, and other privacy-driven upheavals have meant that a business’s advertising strategy, like its marketing strategy, must rely more on first-party data—that which comes directly from the business’s relationship and interactions with its customers.

Thus, we have a thinner wall, if any, between modern advertising and marketing teams. Their more unified technology approach results in fewer silos; cleaner, more trustworthy data; better governance, compliance, and security; greater insight; more ability to take action on their data; and more bang for every ad dollar spent.

The adtech/martech convergence has moved from idea to actual implementation.

2. AI/ML: THE EMERGENCE OF LLMS

The transformational impact of large language models and natural language interfaces will be felt throughout the marketing data stack, and throughout the marketing organization. In particular, LLMs and generative AI are expected to automate and improve a wide range of marketing functions and goals, including:

• Ad campaign optimization: Insights to improve audience segmentation and ad placement will emerge from large data sets that can include tables with rows and columns but also images that LLMs can parse, improving conversion rates and campaign ROI.
• Content generation: If you’re looking for serviceable text, quick and cheap, generative AI tools will write your blog posts, ad copy, web pages, emails, and more. They can pretty much do all that already.
• Natural language processing: LLMs are built on advanced NLP techniques, meaning that marketing tools will better interpret and respond to human language. Chatbots that don’t infuriate your customers and virtual assistants that don’t frustrate you are just the beginning. NLP-powered sentiment analysis can also help marketers gauge public opinion and sentiment toward their brand or campaigns.
• Market research and competitive analysis: Throw it all into your LLM: tweets and online retailer reviews, market research reports, news, publicly available data sets. The algorithm can process all of it to give marketers deep insights into consumer trends and sentiment, the competitive landscape, and shifting market opportunities.
• Customer personalization: Similarly, LLMs can take everything you know about a customer, and similar customers, and tailor campaigns and interactions based on behavioral patterns and more. The outcome: more effective campaigns, happier customers, better campaign results.

The primary and most obvious effect on a marketer’s day-to-day will be in terms of productivity. Advanced AI capabilities will meaningfully upgrade pretty much every technology in the stack and increase the amount of data that marketers can leverage, the quality of the insights and outputs, and the speed of pretty much everything. Whereas querying data today involves going to a data science team that can code the specific queries or provide the algorithmic model to analyze the data, marketers with no coding skills will be able to ask more questions in natural language, and get the outputs directly.

Where data scientists do need to code, the process of writing queries or shaping a data model will be much faster, in part because LLMs don’t need nearly as much structure; with guidance, the model figures out the structure itself. That means that data science teams will also be faster and more productive.

But there’s a subtler, and arguably more important, change that the rapid, widespread adoption of LLMs and AI generally will drive: We’ll trust algorithmic outputs more. Right now, a lot of data outputs come to marketers in frankly mysterious ways. The process of working with data is complex, difficult, and obstruse.

We see the outputs, but we understand that the process of obtaining them doesn’t necessarily feel … organic. Algorithms are inexplicable, remote. When we can simply ask our data tool to explain something to us, and the answer comes to us in natural, conversational language, we’ll trust the output more because it will feel more human. Depending on how generative AI models develop, this heightened tendency to trust could be either helpful or problematic. But without a doubt, this trust will further accelerate the adoption of AI and change the way we work. Nontechnical and less-technical knowledge workers will find working directly with data as natural as we now find using search engines, email, and instant messengers throughout the workday. And the accuracy of recommendation engines, churn predictions, and marketing attribution tools will dramatically increase, delivering the speed of automation while increasing the effectiveness of human decision-makers.

The evolution of the modern marketing data stack is being driven by the sheer volume and variety of data out there. It’s possible for marketers to understand their customers much more deeply than ever before, and to act more quickly and precisely on those insights. But the aggregation of so much data, as well as the need to collaborate on that data with various partners, raises privacy concerns. As does the fact that the ecommerce era means that a business has a direct relationship with a customer’s data. A retailer, for instance, no longer buys ads from a TV station or newspaper promising their audience, on average, fits a certain age, gender, and/or income level. Now the retailer is using data that is, in part, provided directly by the consumer through past purchases and loyalty programs, among other sources. That retailer serves its message directly to the consumer through email or by placing ads on the sites that their target consumer visits. How the retailer uses that data, and which partners it shares the data with, can complicate the relationship between company and customer.

Yet if efficient and insightful data technologies have created this privacy challenge, they can also help solve it. Bringing all one’s data together to create a unified picture of a target customer can improve management of privacy responsibilities. Consumers both provide data and make consent decisions through a variety of channels. They check yes or no on an offer to receive marketing emails. They sign up for a loyalty program that trades data for discounts. They accept or reject the cookies that provide a more personalized online experience. All of these decisions and permissions live within various marketing technologies. Bring that data together and it’s easier to meet obligations to the customer and the relevant regulatory regimes.

Increasingly on Snowflake, we’re seeing our users do exactly that, unifying this data to create not only a holistic view of the customer and a clear view of that customer’s privacy preferences, but also to earn trust and simplify the ability to act on consumer data subject requests.

A unified data structure, with privacy and consent features built in, will also be essential to properly inform LLM training, allowing the removal of data for which consent has not been obtained. This structure also improves the consent-preserving data collaboration within the adtech ecosystem, as we are seeing with market frameworks that let advertisers, ad agencies, and publishers track consumer consent for web page visits.

Speaking of privacy-preserving data collaboration within adtech, we’ve seen increasing adoption of data clean rooms as a way to share customer data without exposing sensitive details. As brands seek increased collaboration to counteract the deprecation of cookies, data clean rooms allow collaboration and privacy compliance to coexist.

Efficient and insightful data technologies helped create a privacy challenge. They can also help solve it.

3. PRIVACY: NAVIGATING THE CHALLENGES

We’ve covered three macro forces shaping the marketing landscape: economic pressures to show ROI, the epochal arrival of LLMs in the AI space, and the convergence of adtech and martech. There’s a fourth force at play, even closer to the technology stack itself:

The need to build a technology infrastructure that by its nature prevents silos and maintains a unified approach to data. That’s been a pain point for years, but one that most organizations have yet to resolve. Looking ahead to our first category, we’ll consider the technology stack, analytics, and data capture, pausing to consider technologies such as Heap, Snowplow, Piano, Mixpanel, and Amplitude, all listed as leaders or up-and-comers. An essential difference between the technologies that lead the modern marketing data stack and the ones that aren’t on the list is that these solutions build directly on your data layer, operate directly from it, or leverage modern data sharing to work directly from your enterprise data platform with no friction. You don’t have to pull the data outside, creating a new copy to manage, so that your tool can get to work on processing the data. These tech makers (and many others we’ll meet in the following pages) are driven by that fourth force: the need for agility that brings the applications closer to the single, unified data store, equipped with native processing capabilities.

4. UNIFIED DATA: SINGLE SOURCE OF TRUTH—AND VALUE

That sounds like an in-the-weeds discussion of infrastructure, but it’s more. By working with tools that are closer to the data, you’re optimizing the expensive time of your data engineers. You’re cutting the time and expense of getting your campaigns right.

And with that, you’re able to be proactive, rather than reactive, in your marketing, and that’s the key to winning in an ever-faster, ever-more-competitive landscape. Marketers shouldn’t be hampered by data friction. Instead, they should be working on what they love to do: delivering differentiated campaigns with optimal speed, accuracy, and agility.

Conclusion

With all that in mind, you are urged to look at leading solutions that some of the most sophisticated, data-driven marketing organizations are adopting to modernize their marketing stacks.

ABOUT SNOWFLAKE: Snowflake enables every organization to mobilize their data with Snowflake’s Data Cloud. Customers use the Data Cloud to unite siloed data, discover and securely share data, and execute diverse artificial intelligence (AI) / machine learning (ML) and analytic workloads. Wherever data or users live, Snowflake delivers a single data experience that spans multiple clouds and geographies. Thousands of customers across many industries, including 639 of the 2023 Forbes Global 2000 (G2K) as of July 31, 2023, use the Snowflake Data Cloud to power their businesses.