ChatGPT and the Future of Customer Self-Service

It being Friday, and having some time on my hands, I decided to explore the ChatGPT capability that has been dominating our LI feeds over the last couple weeks from a Customer Self-Service perspective and get some associated discussion going among the CX/Support community.

Specifically, I was interested in validating how effective the technology would be when asked for information that was NOT general or academic in nature, but instead reflected the type of firm specific detail that many of us in the CX world want to share with our clients via curated online FAQ or accessible KM repositories. More specifically, I wanted to observe the results of requests for informational detail or resolution instruction related to the products and services offered by a business to anticipate what those of us responsible for 'cashing those checks' should expect when (rather than if!) this becomes the standard for broader user search.

I learned a couple of things that I believe will be valuable for those of us in the industry to start thinking about:

1) The tool was absolutely capable of finding results and answering questions that were company and product specific, even when I omitted the company name, but included the branded name of the product/service.

This is without question a positive for those of us in the Customer Service/Experience space because it means that the detail we want our customers to find is definitely within the discoverable access of the tool, and will not be buried in the current SEO rankings of similar standard Google search. It may also have significant implications going forward in both the strategic and tactical methods we use for exposing our self-service knowledge to our customers - particularly given that it bypasses any knowledge/access-based constructs we might implement to gatekeep or curate the customer journey. Will knowledge self-service implementations become simpler by just exposing the backend KM repository to a point of internet access? How do we replace the marketing impressions cross-selling on our most visited support pages?

2) The 'answers' generated by ChatGPT provided no attribution of the source(s), at least not in the current version.

This can have some very significant impacts on the effectiveness of corporate Customer Service as it removes the reader's ability to judge the accuracy of the answer based on where it came from, or when the information was published. Do the provided results represent the 'official' instructions/answers from the corporate source, or outdated/undesirable 3rd party suggestions? Retaining any degree of control over the customer journey in this arena will become even more difficult. Grooming the universe of information that may exist on your own organization's digital footprint for consistency takes on greater importance as does more actively managing information published about the support of your company's goods and services by those not directly related to the the organization. How effective will current channel management strategies be with this greater degree of disintermediation between the organization's official Knowledge and the 'answer' presented to an inquiry - particularly if multiple sources are 'intelligently' combined into a single ChatGPT response?

One thing for certain, those of us in this arena should begin reviewing our current tools and programs now and start planning on how to pivot as I don't expect it will be long before this becomes a standard in how our customers engage.

For those that haven't tried it, I strongly encourage entering some of your own company's support inquiries into ChatGPT and see what comes back - it may surprise you! I'd also love to hear in the discussion about any other observations or takeaways that you think would be beneficial to the community.

#customerexperience #digitaltransformation #customerservice #innovation