Way back in 2008, Starbucks was in trouble. The recession was hitting hard, prompting a lot of people to decide they didn’t any longer wish to spend eight bucks on a cup of coffee. Profits had crashed by 28% in Q1 that year. By the end of 2009, Starbucks had shut 900 stores worldwide, and laid off nearly 7000 employees.

Management scrambled to figure out how to restore Starbucks to its former glory as the most popular coffee chain on Main Street and the largest coffee-house in the world. Former CEO Howard Schultz was brought back to the C-Suite to meet the challenge. Among his initiatives was one which seems incredibly obvious, but had not been tried: ask customers what they want.

Schultz launched “My Starbucks Idea.” Today we’d call it a social media platform. The app asked customers to propose then vote on changes in areas ranging from product flavours to the genre of the canned music. An astonishing 1.3 million Starbuckets proposed 93,000 suggestions.

A great many of the ideas were really very simple. Obvious, even. Not much had to be done to make individuals come back to the happy mermaid coffee place.

Starbucks implemented 277 customer ideas. They made the chairs more comfortable, and introduced free Wi-Fi. They added skinny and non-dairy milk options, expanded the range of “Frappuccino” varieties, and gave regulars a free cuppa on their birthday. Plus, since people asked for them, they added the little green doodads that plug the hole in the cup so it doesn’t splash on your tie whilst your walking to the office (I voted against that).

There’s an insurance point to this. IVANS, a company that operates a network connecting US insurance agents, MGAs, and risk carriers, did something similar lately. They asked insurance brokers and MGAs what they’d like to see from carriers in terms of automated capabilities. I just read the short report they published to reveal their findings.

If anyone were to take a stab at guessing agents’ top four functional automation priorities, they might well get at least three of them wrong. That’s not because they are incredibly niche or advanced, but because they couldn’t fathom that, in 2025, such obvious areas where tech can help are still done manually.  

Agents’ priority areas for automated process support are:

1. Commercial submission
2. Real-time data upload
3. Book transfer between carriers
4. Claims loss runs

Interpreting the many free-text responses, IVANS summarised What Agents Want as:

• Automated exchange of claims and policy information
• AI tech that automates manual tasks
• Agency management systems that connect with preferred carriers
• Quoting tools that are efficient and easy to use

GREAT NEWS! Not only can insurance technology companies deliver all of that, they can do it really easily, because the tech is already on the shelf. The automation of processes that agents most desire is mature. It has existed in multiple versions for many years. We clobbered “challenges” like the automation of claims loss runs years and years ago.

Which leads me to an Uncomfortable Truth. Insurance does not have a technology problem. It has an adoption problem. That’s not because insurers don’t spend a lot of time thinking about tech. They do; they just talk about the wrong stuff.

AI is a board-level topic for most insurance organisations. Millions of hours have been expended (already, no doubt) by insurance leaders reading about AI, asking about it, budgeting for it, and generally making it far more important in their thinking than it deserves to be.  

Another Uncomfortable Truth: if insurance leaders had spent half those millions of hours asking key trading partners about their greatest pain points, then talking to vendors about curing them, they’d already have covered the cost thorough process savings.

The knowledge gained by such conversations allows insurers to select and dine-on the low-hanging technological fruit. Supplying agents with real time data upload may be (a bit) more difficult for the potential gain available, but loss run automation is a doddle, and it is US agents’ NUMBER FOUR AUTOMATION PRIORITY.

(Sorry for shouting, but I am frustrated that everyone hasn’t nailed this yet.)

Forget the low-impact, high-effort technology applications. Chances are you just don’t need them. All your agents want is WiFi and a comfy chair. It’s no big ask, and you won’t even need AI.

Thanks for reading my diary. See you in 2026.