Editor’s Note: This is Part 2 of a four-part series on how AI can be incorporated into marketing automation platforms. Part 1, AI marketing automation: Here’s how it works and why marketers should care. here.
For most of 2023, AI hype has focused on generative AI content use cases (copy, images, video). While some question the ultimate impact of generative AI, mainstream adoption justifies much of the focus on content-focused features.
But an even more serious move is underway. The introduction of AI into society. every day Application of marketing technology.
For martech leaders, embedding AI into core stack components such as CRMs and marketing automation platforms (MAPs) increases accuracy and productivity. Within that, I have prioritized data management. We recognize that data management is also fundamental to most marketing operations leaders.
Data management: the first (semi) natural language process
Before the AI inflection point, data management was the earliest “natural language” change that fueled the growth of martech. how? Codeless transformations now allow the creation of new database fields that were previously given to IT departments. Digital engagement has been transformed with the ability to create internal and customer-facing fields integrated into landing pages and websites.
Even with automation, much of the input relies heavily on human interaction and system interfaces. And despite the ease of use of the tools, training was still a barrier to adoption for (good) data entry.Early AI Algorithms Influenced Various Data Cleaning Processes rear The data was entered incorrectly or was incomplete. But we all knew it was most efficient to prevent inaccurate data from entering the system that would result in erroneous results downstream.
We will use the general framework Garbage In, Garbage Out (GIGO) to illustrate.
“There is garbage”
1. Enter data
Martech leaders cringe when users say data entry is difficult. Especially when the interface has changed over time, it’s natural to resonate. (If you’re a Salesforce shop and still switching to Classic and Lightning, it’s a reminder of your empathy.)
Many major vendors, including Salesforce, recently predicted that the “prompt” revolution in generative AI would change user interfaces forever. All his UI now has to handle natural language, reducing friction (excuse me if you’re being cynical) when entering data for users.
For example, ChatSpot (HubSpot’s AI interface) leverages the GPT model in its user interface. (I’m vendor agnostic, but I’ve taken advantage of this tool and have been able to test it in public alpha releases, so I’ll just pick an example.)
Let’s start with the basics of adding a new contact.
Users don’t have to remember where to click “Add contact” in HubSpot’s standard interface. Instead, a simple prompt like this is used:
In its three-month alpha, HubSpot also added prompt templates that trigger actions based on common to-dos, so you can pick them from your favorites list like this.
2. Explore and add data about people and companies
Many MAPs got their basic customer information from their website. AI simplifies this task, allowing you to quickly create summary versions of key profiles to reinforce contact personas or supplement company company information. for example:
3. Embed in your spreadsheet
Nearly 70% of marketers spend 10 or more hours a week working with spreadsheets, according to MarTech’s 2023 Salary and Career Survey. These are the foundation of the martech stack.
In my March 2023 MarTech conference presentation, I talked about how these tools (and their formulas, VLOOKUP functionality, etc.) are still secret decoders for working across multiple data sources. Many large teams have full-time data analysts to support these efforts. Smaller teams often have data-savvy marketers with Excel expertise.
However, programming VLOOKUP is too technical for many people. Marketers are now using generative AI prompts to create formulas. Several AI plugin utilities bring AI-created prompts directly into your spreadsheet.
These natural language “no code” features will be the most powerful and most used additions. These are embedded directly into underlying knowledge work tools such as Google Workspace Labs and Microsoft Co-pilot. A user asks her AI assistant to extract domains from email addresses, extract first and last names, company names, etc., effectively creating structured data through natural language prompts.
“Garbage out”
Now let’s look at the other end of the spectrum. This is a use case where AI assists data output.
4. Natural language interface for analytics
We have all been there. You will be prompted to export the report in PowerPoint or Google Slides instead of visiting the platform. Being able to get reports from your application through natural language prompts is a game changer.
“<穴埋め> Would you like a report based on this?” prompt lowers the barrier to more people directly accessing analytics.
Over time, the more likely users are to enter data and make sure it’s reflected properly, the more likely they are to provide quality entries. Perhaps the user will fix it in the source instead of fixing the chart.
5. Injected visualization features
The creation of visualizations also incorporates functionality. Through a plugin/interface you will be able to request these visualizations from the platform.
Like many people, I’m looking forward to having access to OpenAI’s code interpreter capabilities. In the meantime, I’ve been following others piloting it. That includes his Ethan Mollick, who sneaked up on the feature in his One Useful Thing newsletter. Its content is excerpted in a recent newsletter post.
6. Accessible big data
All of these data input and output benefits are not limited to specific data being the CRM/MAP “source of truth”.
Lowering the barriers to entry to more data sources will allow access to other data augmentations and supplemental attributes through AI-based prompts, so that the output of one analysis can influence others in ways not previously considered. may be linked to the output of the analysis of good.
Governance and training still needed to avoid blind faith
Martech leaders must be careful not to rely solely on AI for data management and quality. Given that generative AI tools are immature and can impact data quality if not monitored, additional governance should be applied.
Data management challenges have a double impact. Prompts may not honor your organization’s guidelines for associating contacts with accounts. You may need to develop more advanced prompts that follow these guidelines.
Now anyone importing data into a spreadsheet is doing a sanity check after applying the formula. A typo can cause problems across thousands of records. But if users didn’t create good prompts the first time, flawed logic introduced by the AI could corrupt thousands of records.
what’s next? Part 3 of this series will dive deeper into the introduction of AI into the MAP campaign process.
Get Martech! every day. free. It’s in your inbox.
The opinions expressed in this article are those of the guest author and not necessarily those of MarTech. Staff authors are listed here.
