AI Decisioning: 3 Key Considerations for Successful Application

.
February 16, 2024
AI Decisioning: 3 Key Considerations for Successful Application

The pace of innovation with AI will not be slowing down. As this technology continues to penetrate your business, it will become critical to establish the ground rules and governance around the application of AI. From marketing to supply chain, and every function in between, there must be ground rules for how your business applies artificial intelligence. Aligning your entire organization in the right direction to make the best decisions around AI will ensure the application’s long-term success

At Blend, we are helping clients solve these dynamics by starting with careful consideration of 3 key areas:

  1. Managing Embedded AI vs Build vs Buy
  2. Understanding Use Cases and Expectations
  3. When to AI and When NOT to AI

Let’s jump in.

Managing Embedded AI vs. Build vs. Buy

The decision of whether to build or buy an AI tool is where most teams start their journey. There are hundreds of new tools on the market, with plenty more being released weekly. Building an intriguing widget is more affordable than ever with the cost and availability of processing power and data storage being at an all-time low. As a result, we should be honing in more on the management of embedded AI tools and features. This is often an overlooked element as it tends to live within a functional area and is often taken for granted. Sometimes they are overlooked simply because there is no ability to replace the solution.

For example, while historically creating customer bidders was possible, the new capabilities of Google’s search bidding cannot very easily be replaced. Even if you could, there is no way to truly apply that. Google Search isn’t going to become Open Source anytime soon. This is where the onus falls on a company’s leadership team. Management of the inputs and expectations are crucial decisions to make.

As Google, Meta, and the other platforms become more AI-enabled (especially as the data access barriers in digital advertising continue to rise), organizations need to have a hand on the wheel and guide these AI tools to maximize results. While these AI features tend to live within the business capability that uses the tool, as noted previously, it still is important that the overall AI governance plans and infrastructure have an eye on these capabilities to support the overall management of the data and application. Organizations need to make sure that all AI, whether a stand-alone product or an embedded capability within a function, is operating through guided hands

Use Cases & Expectations are a Must

AI platforms and technology are not “set it and forget it” tools. They need to be guided to deliver what it is that we’ve set them up to do. This is why use cases and expected outcomes are mandatory in ensuring the AI technology is successful. How does your team define these for yourselves? Lack of use cases and proper expectations have been a failure point for technology and measurement decisions for years. There are a substantial number of examples where CDPs, MTA platforms, and Decision engines, just to name a few, have all resulted in wasted investment. This is a symptom of failing to establish the “WHAT the tool is going to address” and “HOW that is going to change things” within the teams that deploy these tools.

You can count on artificial intelligence to fall into the same trap and potentially create even worse outcomes. AI is going to also be guided by poorly defined data inputs and is going to potentially make business decisions and drive unexpected customer experiences. It might work well under a few edge cases, but then fail when we need it most because the use cases and expectations were not well defined and so it was fed with inappropriate data to learn from. As the saying goes– garbage in, garbage out.

To AI or NOT to AI

This ultimately is a cumulation of the previous two areas, but just because you CAN use AI, doesn’t mean you should.

Reflect on the use cases your team has defined and ask:

These are “in the mirror” discussions that organizations need to have. These tools and technologies also come with some expected change management. Roles, processes, and entire functions may be outright removed. Is the business ready for that?

For instance, with all of the capabilities around generative content, the traditional role of the creative and SEO teams will change. They are going to shift more into providing inputs, whereas in the past they were building full creatives and content. Legal and promotion need to make sure expectations are properly set so that the generated content is allowable, and whether they must review every piece of content. Then, AI in this use case may not be prudent.

Success with AI Starts with Strong Planning

There is no arguing that AI is going to be a crucial component of how businesses operate in the near term. They are going to drive personalization, chat experiences, product recommendations, inventory forecasting, and much more. The important thing is to ensure that the hype around AI doesn’t drive the business to make costly investments or decisions that ultimately cause more harm than good. At Blend this is something that we are exceedingly passionate about. We want to help ensure everyone gets the full benefit of AI, but without falling into the common pitfalls that could create dissatisfaction and distrust of artificial intelligence solutions.

Learn more about Blend’s artificial intelligence practice.

The pace of innovation with AI will not be slowing down. As this technology continues to penetrate your business, it will become critical to establish the ground rules and governance around the application of AI. From marketing to supply chain, and every function in between, there must be ground rules for how your business applies artificial intelligence. Aligning your entire organization in the right direction to make the best decisions around AI will ensure the application’s long-term success

At Blend, we are helping clients solve these dynamics by starting with careful consideration of 3 key areas:

  1. Managing Embedded AI vs Build vs Buy
  2. Understanding Use Cases and Expectations
  3. When to AI and When NOT to AI

Let’s jump in.

Managing Embedded AI vs. Build vs. Buy

The decision of whether to build or buy an AI tool is where most teams start their journey. There are hundreds of new tools on the market, with plenty more being released weekly. Building an intriguing widget is more affordable than ever with the cost and availability of processing power and data storage being at an all-time low. As a result, we should be honing in more on the management of embedded AI tools and features. This is often an overlooked element as it tends to live within a functional area and is often taken for granted. Sometimes they are overlooked simply because there is no ability to replace the solution.

For example, while historically creating customer bidders was possible, the new capabilities of Google’s search bidding cannot very easily be replaced. Even if you could, there is no way to truly apply that. Google Search isn’t going to become Open Source anytime soon. This is where the onus falls on a company’s leadership team. Management of the inputs and expectations are crucial decisions to make.

As Google, Meta, and the other platforms become more AI-enabled (especially as the data access barriers in digital advertising continue to rise), organizations need to have a hand on the wheel and guide these AI tools to maximize results. While these AI features tend to live within the business capability that uses the tool, as noted previously, it still is important that the overall AI governance plans and infrastructure have an eye on these capabilities to support the overall management of the data and application. Organizations need to make sure that all AI, whether a stand-alone product or an embedded capability within a function, is operating through guided hands

Use Cases & Expectations are a Must

AI platforms and technology are not “set it and forget it” tools. They need to be guided to deliver what it is that we’ve set them up to do. This is why use cases and expected outcomes are mandatory in ensuring the AI technology is successful. How does your team define these for yourselves? Lack of use cases and proper expectations have been a failure point for technology and measurement decisions for years. There are a substantial number of examples where CDPs, MTA platforms, and Decision engines, just to name a few, have all resulted in wasted investment. This is a symptom of failing to establish the “WHAT the tool is going to address” and “HOW that is going to change things” within the teams that deploy these tools.

You can count on artificial intelligence to fall into the same trap and potentially create even worse outcomes. AI is going to also be guided by poorly defined data inputs and is going to potentially make business decisions and drive unexpected customer experiences. It might work well under a few edge cases, but then fail when we need it most because the use cases and expectations were not well defined and so it was fed with inappropriate data to learn from. As the saying goes– garbage in, garbage out.

To AI or NOT to AI

This ultimately is a cumulation of the previous two areas, but just because you CAN use AI, doesn’t mean you should.

Reflect on the use cases your team has defined and ask:

These are “in the mirror” discussions that organizations need to have. These tools and technologies also come with some expected change management. Roles, processes, and entire functions may be outright removed. Is the business ready for that?

For instance, with all of the capabilities around generative content, the traditional role of the creative and SEO teams will change. They are going to shift more into providing inputs, whereas in the past they were building full creatives and content. Legal and promotion need to make sure expectations are properly set so that the generated content is allowable, and whether they must review every piece of content. Then, AI in this use case may not be prudent.

Success with AI Starts with Strong Planning

There is no arguing that AI is going to be a crucial component of how businesses operate in the near term. They are going to drive personalization, chat experiences, product recommendations, inventory forecasting, and much more. The important thing is to ensure that the hype around AI doesn’t drive the business to make costly investments or decisions that ultimately cause more harm than good. At Blend this is something that we are exceedingly passionate about. We want to help ensure everyone gets the full benefit of AI, but without falling into the common pitfalls that could create dissatisfaction and distrust of artificial intelligence solutions.

Learn more about Blend’s artificial intelligence practice.

Download your e-book today!