Where to deploy AI for maximum martech impact


AI is rapidly reshaping various industries, and martech is no exception. As AI becomes increasingly integrated into business operations, marketers face a wealth of options and strategies to leverage these technologies effectively.

Let’s examine the benefits, deployment strategies and key considerations for integrating AI into your martech stack to drive better results and optimize customer experiences.

Harnessing machine learning and generative AI for marketing success

Machine learning techniques that have been around for a while consistently deliver impressive results. For example, brands using predictive analytics and targeting the right audiences on platforms like Meta often see 15% to 40% improvements in CPA, ROAS and CAC.

Many of these tools are affordable, with starting prices in the hundreds of dollars per month, and can be set up in just a few days. Achieving a 10:1 ROI or better is now common, even with just one use case.

Generative AI is equally disruptive, allowing marketers to:

  • Perform data discovery: Use AI to uncover insights and trends within your data.
  • Summarize meeting notes: Leverage AI to organize and condense notes from multiple sessions.
  • Utilize natural language queries: Ask natural language questions to your datasets and let AI generate the necessary SQL queries.
  • Create data visualizations: Generate complex charts and graphs quickly with AI tools.
  • Scale content creation: Deploy generative AI to produce content based on brand parameters, eliminating creative bottlenecks and enhancing personalization.

The tyranny of choice

AI solutions are now available from almost every tool in a brand’s stack, making this transformation highly accessible. For example, while our company is a Google Cloud Partner, we also have Microsoft Office365, GitHub, Google Workspace, AI-powered meeting recorders and HubSpot. Many of our employees have paid OpenAI subscriptions, all with extensive AI features. 

The problem becomes more complex when we look at our clients’ stacks. They have ad campaigns across all the big ad platforms, email service providers, cloud accounts, journey management, cloud providers and standalone data science providers, all frequently part of their stacks. The broad array of choices necessitates a focused AI strategy.

Dig deeper: How to transform martech and multichannel marketing for the AI era

Start with an AI strategy

As with anything complex in the martech stack, start with a strategy by asking classic questions about your goals, data, tech stack and process. Determine where the biggest opportunities are for incremental revenue or saved costs. Some of these things may be marketing. Some may be in other departments. Focus on where AI efforts will make the most impact. For marketing technology:

  • NLP and computer vision may help with content classification.
  • Generative AI will help create content from form fills to expedite processes to creative optimization.
  • Predictive AI and machine learning will help identify the best audiences and optimize your segmentation.

Experts can help score the potential opportunity for your organization. The outputs of the AI exercise look much like any other investment opportunity. They can be scored by impact size and time/cost to deploy. 

For one mid-sized retailer we worked with, a small $20,000 investment was projected to add $300,000 in bottom-line value — a 15:1 ROI on its first use case. Another enterprise media company estimated a $5 million return on a $500,000 Google Cloud project. Each project could be completed in 2 weeks to 4 months. AI offers significant value.

However, with all the options, where in the tech stack should a brand deploy it?

Dig deeper: 4 ways to achieve early wins with AI in marketing

Deploying AI in different contexts

There are tons of options for adding AI to the value chain. For this article, I will focus on cloud providers and marketing technology providers. 

Cloud providers

Each offers compelling AI functionality. For marketing applications, the leading providers, Google Cloud and Snowflake, have each introduced compelling offerings. But the marketing user isn’t typically a cloud user. 

Clouds are governed by IT or enterprise data teams who need to understand the use case, prioritize resourcing and build functionality that drives cost in their cloud infrastructure. This is the obvious choice for cutting-edge organizations where customer data and proprietary predictions are part of the product. This is how companies like Netflix and Spotify deliver personalized experiences across all channels.

  • Pros: The brand owns the AI and can deploy it anywhere to utilize its scores and outputs. This allows for centralized control of customer experience through its AI, which is ideal for companies using a composable data activation strategy.
  • Cons: This requires IT buy-in or external consultants to build the AI. Integrating it throughout your tech stack will introduce some complexity with martech and messaging tools downstream of your cloud provider.

Martech providers

Many tools can help drive customer experience, such as: 

  • Multi-channel messaging tools.
  • Ad platforms.
  • Site personalization.
  • App personalization. 
  • CTV targeting.
  • Direct mail. 
  • SMS.
  • Etc. 

Providers offer AI tools that leverage the data your brand shares with them and the data each provider collects. While an individual customer’s score or creative experience may differ, each tool can drive incremental lift by using AI features.

  • Pros: Each tool has relatively easy-to-deploy AI, often for free or at a small incremental cost. They can be turned on with relative ease, and you can test the incremental impact of the AI solution with zero to minimal development resources. 
  • Cons: AI and personalization will not be consistent across your stack if you deploy them in multiple tools. Depending on the complexity of a brand’s stack, this could lead to disjointed customer experiences.

A hybrid approach

We often see clients develop their proprietary AI and combine it with vendor-provided AI solutions. For example, one subscription publisher we worked with has extensive churn and propensity scoring directly integrated into their data warehouse, Snowflake. 

The scores were shared daily with their CDP. Within the CDP, marketing users built segments and triggers from proprietary scoring. Then, they drove creative experiences with generative AI and used CDP behavioral scoring to tailor channel and frequency. This client reported a 20% improvement in churn mitigation.

There is no one-size-fits-all approach. A brand’s circumstances matter, but there are some near-universal truths to consider:

Decide your strategy

Forecasting which AI deployments will help provide the most incremental impact may make the decision to build/buy AI obvious. This will also inform where AI should be injected into the value chain.

Consider who your users are

Do you have adequate resourcing to build AI in your cloud with full-time employees or consultants? Do you have the ability to deploy that AI across channels? Or do you have valuable data in your martech tools that marketers can easily leverage? Or do you have all of the above? 

Customize your AI

Leveraging OpenAI in blind faith is not a strategy. Here’s an example of the same article I wrote here, trusting only AI (even with decent prompting). 

If you’re using generative AI, give it tons of context and prompts to know your policies, brand guidelines, brand-approved content and rules. This will make for much more compelling content. 

Similarly, if you’re building predictive AI, consider your churn windows, propensity scoring windows, etc. Consider your buyer’s journey and how AI can benefit those specific moments.

Don’t overbuy AI

You likely have duplicative AI in your stack. Building AI in the cloud is not a free endeavor. Govern how much you’re paying for duplicative and redundant AI. This can also help to govern the customer experience with fewer potentially conflicting signals.

Train your marketers

Often, fancy AI is deployed at the last mile by early-career, hands-on-keyboard marketers. They need training to know where and how to use AI — generative or predictive. Make sure to have a playbook for using AI and set up evergreen journeys that enable seamless adoption of AI into the customer experience.

Dig deeper: How wisdom makes AI more effective in marketing

Contributing authors are invited to create content for MarTech and are chosen for their expertise and contribution to the martech community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. The opinions they express are their own.



Source link