Five tips on AI prompt writing (with real world Customer Success examples)

Are you interested in using AI to make your Customer Success job easier? If so, this article is for you! We’ll walk through how to write prompts to help you better serve your customers, with examples you can add to your CS toolkit. Top tips include how to consider the customer perspective, use consistent language and formatting, and the best structure for efficient, effective results. We’ll also throw in some advice on optimizing and testing prompts for accuracy (to ultimately automate and save you time). By the end of this post, you’ll have the confidence and knowledge to create effective CS AI prompts. Let’s get started!

After playing around with our new CS AI tool and gathering product tips and tricks from the industry at large, the above is what we produced as an introduction for this blog. It was a team effort (strong prompts and follow-up instructions on our end, quick and accurate copy on CS AI’s end) and included some minor editing tweaks. Overall, we’re pleased with this generative AI’s recommendations.

Prompts make perfect

The day-to-day of a CS professional is more dynamic and varied than one blog post prompt, but the basic tenets of AI writing are the same. The key is to create a set of prompt writing muscles and develop a routine for automating tasks. Generative AI will provide content and help you connect ideas and concepts.

We’ve found the following five prompt writing techniques to work best.

Be specific: Think about how you would request something from a coworker. “Respond to that customer email” doesn’t offer much by way of next steps. “Draft a 50-word response to the customer email above that confirms we received their request, lists next steps for when we will fix the product issue, and offers some days and times for a follow-up call based on their preferred office hours” sets everyone up for success. This addresses both input (original customer email) and output (response, format, content needed, word count).

Use examples: Generative AI relies on preexisting data for context, so it’s helpful to guide them in the right direction from the start. If you need to summarize a quarter’s renewals compared to last year, then include a previous example with guidance such as “format this year’s summary of renewals data to match this example in tone, structure, and length.”

Include context: While chat-based AI tools may seem human and dynamic, they are really more like a smart, quick encyclopedia. Guiding them with background or related information can get you to your end goal faster. This can include customer demographic data, current events, or details on a competitor or similar business.

Provide a word count: Since CS AI is text-based, this is key guidance to include every time. Tying back to specificity for our customer email example, it can mean the difference between a five-word response (too short, not thoughtful) or a 500-word extrapolation (too long, difficult to understand next steps).

Refine, refine, refine: It’s rare to write anything perfectly on the first go. Just because generative AI seems like magic, doesn’t mean we should throw good habits like re-reading, proofing, and editing out the window. Review AI’s response and see how you can dig a little deeper (keeping in mind things like specificity, examples, context, and word count). Consider follow-up prompt language like “Let’s think step by step for X paragraph” or “Write this in the style of [famous person] or [industry profession].”

Three real world Customer Success prompt examples

Now it’s time to apply our general prompt writing tactics to Customer Success-specific tasks and projects.

Example 1: Email content generation

Email is an integral communication tool that just so happens to be extremely distracting. So let’s automate part of this daily practice! CS AI can help us save time on drafting and templatize common responses/tasks.

Prompt examples:

• Write an email to Jane at Acme to remind them that their renewal of [App Name] will be on July 15 for $24,000. Offer to meet if they have questions.
• Suggest some strategies for responding to an email regarding cancellation of a software product due to budget constraints.
• Write an email welcoming a new point of contact for your account [account/business type, contact details]. Suggest a few ways to connect.

Example 2: Customer and market research

According to Acuity Knowledge Partners’ recent survey on market researchers and ChatGPT, almost half of respondents anticipated it would have a positive impact on the industry and expected its usage to increase in the near-term. How can we use the CS AI tool to bolster our CS research needs?

Prompt examples:

• Competitor battle card – Provide a 500-word competitive analysis on X company that includes the products they build, who they sell to, and how they are different from [our own company].
• Customer segmentation – Categorize the following customer data responses into [X, Y, Z] buckets.
• Customer feedback – Segment this customer feedback into the following themes: product features, technical inquiries, customer service, subscription renewals.
• Content analysis – Summarize this customer call transcript in 350 words.

Example 3: Video/phone scripting

This is where the flexibility of content comes in handy. CS teams can use what they already have (email communications with clients, market research, call transcripts, content such as blogs, guides, or other videos) and feed this into CS AI for summarizing or reformatting. In this case, let’s see how it does with drafting scripts.

Prompt examples:

• Draft a call script for a customer referral that uses the same structure as HubSpot’s cold calling script.
• Write a script about the importance of having a consistent “Account External ID” to unify your customer data.
• Convert this blog post on the top tips for customer expansion into a 90-second/200-word video script.

Keep in mind, a lot of these prompts transfer well for other needs, like training resources or strategy ideation (e.g., “Write an outline on using ABC product, based on the following summaries” or “Review the following usage reports and provide a list of where time was spent within the product”).

Refine your Customer Success prompts

As we discussed in the first section regarding crafting a good prompt, the goal isn’t a perfect output on the first request. Rather, we hope for content we simply refine with CS AI’s help.

Similar to offering feedback to team members or editing your own work, this requires analysis of both what was generated and our original request. How can we get closer to our ultimate goal of a perfect email/summary/script?

Below are some seemingly standard refinement examples that can apply across content types and needs.

Refinement examples:

• Provide a shorter version [between X and XX words].
• Make it more friendly/formal/funny [in the tone of <famous author>].
• Include the benefits of [your app; CS AI may actually know!].
• Remove/add references to [this].
• Change [this word] to [that word].
• Now add a list of 3-5 next steps for how to implement.

Integrate generative AI into your Customer Success strategy

Despite how human CS AI’s responses appear, it’s important to remember that generative AI still has limitations. It may lack expertise in certain industries or topics or the ability to interpret some data. Because it’s still built by humans, it also may display certain biases that should always be checked and considered.

With those caveats and this guide to better prompt writing at the ready, it’s time to dive in and ask CS AI for the content and ideas you need, on demand and in seconds. Copy and paste never felt easier, right?

CS AI™ is an industry first: the first generative artificial intelligence built into a Customer Success platform.

Try it free at: https://churnzero.com/csai/.