Artificial Intelligence and the Bad Data Problem

Insurtech tools and agency management systems are game-changers for prospecting leads, closing deals, cross-selling and assisting clients. But sales and servicing tools are only as good as the data they contain.

The saying “garbage in, garbage out” applies to artificial intelligence (AI) tools, no matter how advanced they are.

Let’s say you train your AI model using client information you haven’t reviewed for accuracy. You ask your AI to pinpoint which leads would most likely respond to an employment practices liability insurance (EPLI) prospecting campaign. Your prompt it to list contacts who need EPLI based on their operational risk.

Your AI scours your system data for information on industry risk, including North American Industry Classification System (NAICS) codes. But your agents don’t always input NAICS codes for leads. Only 25% of the leads have NAICS codes, leaving your AI to guess on or omit 75% of them. When you ask for a strategy to harness your EPLI prospects, your AI returns an inaccurate assessment.

Don’t get caught in the AI hype and rush to launch tools you’re unprepared for. Launching AI tools using bad data isn’t a good idea. Instead of solving your data problems, it could perpetuate them. Learn why cleaning your data will be worthwhile in the long haul.

Good data, bad data and the realities of AI

AI can be a catalyst for your agency. But you’ll need to train it first. It takes employee oversight to get your data in shape. And it takes training to get AI to work in your workflow.

AI doesn’t understand “good” or “bad” data until you train it to differentiate between them. If you use bad data as the source information for your AI tools, AI will inherit it and output variations of that bad data. AI-generated bad data might seem more convincing than your old data. Some AI tools communicate their findings so convincingly you might not question them.

AI isn’t a catch-all solution for solving a data problem. Before implementing AI, review the quality of your data. Spot-check your data against reliable files. An AI tool may produce misleading and inaccurate business insights, preventing you from achieving organizational goals.

Outdated or bad data left to linger in a system leads to more bad data. No tool will help until you clean the data. But it takes a village to keep your data in check.

Train employees on good data hygiene

Train your employees on the importance of client data. Engage them and give them best practices on data entry. Explain how clean data will increase opportunities and improve service and client retention. Make staff accountable for their account data. Employees are more likely to participate if you’re transparent about why you want them to do a task and ways it will improve their workday.

Give your employees information they can act on immediately:

Input client details, even for leads

Include details like email, primary address, phone number, first and last name, nickname and NAICS code. Comprehensive deep data helps you make decisions on targeted email marketing and cross-sell opportunities.

Update clients’ data regularly

Staying current with contact information, name changes, life changes, birthdates, business partnerships and operational changes gives you reasons to reach out with relevant information.

Make data maintenance part of your client interactions

Verify contact information at the end of phone calls. Make account confirmation part of all service and sales calls. If you use a client service portal, encourage clients to log in and review their account information. This also drives awareness about your client services, creating value for your clients.

Update bounced emails and remove clients who opt out of emails

This increases client engagement with newsletters and cross-sell campaigns since your emails are getting to the clients who want them. Information gets to your clients’ inboxes instead of a spam filter.

Remind staff that cleaning data isn’t a one-and-done event

Ongoing maintenance is crucial for successfully implementing any system, including AI. Data collection practices change as technology changes. There will always be something new to manage, but if you keep staff informed you can stay ahead of the curve.

Create data entry guidelines and train your staff on them

Stay attuned to staff feedback on data cleanup initiatives. If they’re not doing it, find out why. If other staff members are successful in their cleanup, have them share their tips with the rest of the team.

Use AI tools to clean your data before launching other AI models

The good news is that AI tools can help you clean and organize your data before you enlist other AI models to help you strategize. These data-wrangling AIs can help your team with tasks like:

• Scouring databases for duplicated client data
• Mining and organizing disorganized data (Disorganized data doesn’t follow a specific arrangement or isn’t logically stored or connected. Spreadsheets or siloed systems that don’t synchronize are examples of disorganized data.)
• Flagging missing or inconsistent data
• Resolving data discrepancies (with employee oversight)
• Transforming data into uniform field entries (normalized data)
• Cleaning and extracting unstructured or unformatted data (Unstructured data doesn’t have astructured field entry. Account activity notes, social media posts and survey results are examples ofunstructured data.)

Some agency management systems offer tools to help you manage data cleanup. Call them for advice.

Make data hygiene a habit across all your systems

When data is unreliable, client marketing loses its impact. It seems small, but a misspelled name could make a client feel uncared for, leaving them doubting your commitment. At renewal time, they might go with the first reasonable offer from the competition.

With reliable data, client management and marketing tools are a superpower that can increase sales and engagement.