Overcoming Analysis Paralysis With AI

This data was originally featured in the June 19th, 2024 newsletter found here: INBOX INSIGHTS, JUNE 19, 2024: KEEPING PROJECTS ON THE RAILS, OVERCOMING ANALYSIS PARALYSIS WITH AI

In this week’s Data Diaries, let’s showcase one incredibly powerful, overlooked, underused application of generative AI, specifically large language models. Anyone who’s worked in analytics for more than a few minutes has heard of or even confronted that most dreaded of phenomena:

Analysis paralysis.

Yes, it’s organizational overthinking, the classic “don’t let best be the enemy of good”, waiting for the perfect plan or the perfect piece of data to take action. Analysis paralysis does exactly what the label says: it paralyzes organizations into inaction, letting critical opportunities pass you by.

Why does it occur? Mainly because of risk. If you work in an organization that doesn’t reward risk (and let’s face it, that’s most places nowadays) then it’s safer for your individual career to let an opportunity go by than it is to take a chance with “pretty good” information. There’s a clear neck to choke if you take a chance and it doesn’t pan out. There’s a collective shifting of blame when no one can be held accountable for not taking action.

So, how does generative AI play a role here? By being a second opinion. By offering a neutral third party’s opinion that, if you allow it, can nudge you into action OR tell you that standing pat is the right choice.

Okay, that sounds great. How do we bring this to life? Unsurprisingly, we’ll use the Trust Insights 5P Framework: purpose, people, process, platform, performance.

• Purpose: What is the decision we’re trying to make? What is the purpose of it?
• People: Who are the people involved? Critically, what role do they play in the consequences of the decision?
• Process: How do we normally make decisions of this nature? (And are those processes any good?)
• Platform: Where do we get information to make these kinds of decisions?
• Performance: And of course, was the decision the right choice?

Let’s look at how we would bring this to life. We’d start a session with the generative AI tool of choice, and start with the Trust Insights RACE Framework with a role that’s appropriate to the situation. Let’s say we are debating investing in ads on LinkedIn. We know it works for some folks, and we also know it costs far more than other kinds of advertising.

• Role: You’re an expert B2B digital marketing manager skilled at lead generation from both paid and unpaid marketing channels.
• Action: Evaluate the following circumstances and help make a decision about whether or not to advertise on LinkedIn.
• Context:
  • Purpose: We need to make a decision whether to advertise on LinkedIn or not for the purposes of lead generation for our sales pipeline.
  • People: I’m the digital marketing manager of Trust Insights, and I would need the CEO’s approval for this program, but building a case for or against LinkedIn ads is my responsibility.
  • Process: We’ve tried LinkedIn ads in the past, along with other digital ads channels. For this particular decision, we have an overall advertising budget of $500.
  • Platform: I’ll include some past results for LinkedIn, both paid and unpaid. I’ve also included Google Analytics data by source/medium so you can see what’s driving traffic and conversions already.
  • Performance: Help me make a decision.
• Execute:
  • If you have enough information to make a decision about whether or not we should buy ads on LinkedIn for lead generation, then please talk through step by step your decision-making process, then return your final decision.
  • If you do not have enough information to make a decision, please ask me followup questions to obtain that information.

What are the results of this?

Gemini rendered a verdict of no-go: we shouldn’t advertise on LinkedIn, given past performance. Instead, it recommends investing our limited budget in SEO or email marketing, while trying to figure out a working formula for LinkedIn with organic content first.

This is a useful technique to overcome analysis paralysis. Combining the Trust Insights RACE and 5P frameworks gives us a solid prompt to start making a decision; we could augment this with an ideal customer profile or more specifics, but if we just need a second opinion on something we’re already debating, generative AI can provide it.

How are you using generative AI to overcome analysis paralysis?