We’re not breaking new ground when we say, there’s a lot of noise surrounding the future of artificial intelligence (AI) and the impact it’ll have on life as we know it. And sure enough, your favorite business function customer success will change as a result. Screw it, it is changing.
Since OpenAI’s generative AI model, ChatGPT, burst on the scene in November 2022, the world hasn’t quite been the same. And for four solid months, ChatGPT was relatively peerless. This changed in March 2023 when the world’s largest search engine entered the game: Google AI released Gemini (formerly known as Bard).
But what do these two titans of technology have in common, what are their key differences and, perhaps most importantly, which generative AI model will make the biggest impact on Customer Success Managers (CSM)?
Carry on reading this article if you’re curious to know:
- What ChatGPT is,
- What Gemini is,
- The key similarities and differences between ChatGPT and Gemini,
- How to craft great prompts for generative AI, and
- How ChatGPT and Gemini can be used by customer success teams.
Let’s get on with it, shall we? (Before a new conversational generative AI chatbot emerges and steals the limelight. 😂)
What is ChatGPT?
By now we're pretty sure you've caught wind of ChatGPT, the impressive AI model fashioned by OpenAI. Its primary function? Engaging in conversation, interpreting cues, and crafting strikingly human-like text responses. There’s honestly not been anything else like it used on this scale.
Even if you haven't interacted with it personally, its buzz is almost impossible to escape. It's a rare spectacle to witness a platform skyrocket its user base as much as ChatGPT did, reaching an unprecedented 100 million monthly users in January 2022, a mere two months post-launch.
ChatGPT is a large language model (LLM) that generates human-like text based on the input it's given. You can use it to replace or replicate the creative process for tasks that would ordinarily take more time and concentration:
- Drafting and editing content: You can use ChatGPT to draft emails, write short pieces of text, or edit existing content.
- Learning and teaching: ChatGPT can be used to explain traditionally complex concepts in straightforward and digestible terms. Traditionally, understanding these concepts would take much longer to research manually or via Google’s search function. You can also use it as a tool for learning about topics you’re unfamiliar with.
- General conversation: Although somewhat dystopian, ChatGPT can be used as a tool for interaction, to simulate human conversation, or even to brainstorm ideas.
ChatGPT offers two tiers: a free tier, and a paid tier.
The free tier is excellent, but has its restraints; you’re subjected to access restrictions during peak times due to high demand. OpenAI has circumnavigated this foothold by introducing ChatGPT Plus (its paid tier) allowing subscribers priority access. ChatGPT Plus radically reduces the response time for answers, providing a quicker and more efficient service.
OpenAI’s ChatGPT has undergone several iterations, with GPT-4 being the latest version at the time of writing.
What is Gemini?
Much like its slightly older peer, Gemini is an “experimental” chatbot that uses a massive dataset of text and code to generate text, translate languages, create all different kinds of written content, and attempt to answer your questions in an informative way.
Here’s the technical stuff: Gemini was initially based on the LaMDA (language model for dialogue applications) but since May 2023, it has been powered by the PaLM (Pathways Language Model 2), allowing it to converse with users.
For the time being, you can only use text-based prompts with Gemini, but this is set to change in the near future due to the capabilities of its training PaLM 2. That means eventually, you'll be able to provide Gemini with images and it’ll be able to analyze them!
Gemini's strength lies in its real-time internet scouring capability. In layman’s terms, Gemini is quite literally plugged into the internet – meaning it can pull every record on Google’s search engine within seconds.
Google's re-brand: Bard to Gemini
In February 2024, Google announced it's rebranding its Bard AI chatbot as Gemini, a name change reflective of the advanced AI technology powering the platform.
While users can still access the URL bard.google.com, a redirect has been implemented as the core experience is now rebranded to Gemini. But this goes beyond a mere name shift, as Google's rolling out a new Gemini app for Android users to replace their Google Assistant product. This move will integrate Gemini into Workspace apps like Gmail and Docs, with Google set to launch a premium "Gemini Advanced" tier powered by their more capable Ultra 1.0 model.
For customer success professionals who regularly leverage Bard's generative AI capabilities for tasks like customer research and drafting content, this rebranding will require adjustments to get accustomed to the new Gemini naming across interfaces and outputs.
However, assuming core functionality remains intact during the transition, their day-to-day workflows should be able to continue with minimal disruption once the Gemini rebranding is fully adopted, provided Google communicates changes clearly.
The simplified Gemini naming aims to clarify Google's AI offerings compared to competitors' more fragmented monikers.
The key similarities and differences between ChatGPT and Gemini
Since the birth of the internet and up until the rise of generative AI, when most people need information fast, they simply type a few precise keywords into Google. Then, you’d have a list of results across multiple pages (the SERPs).
Now, you can ask for rich, details answers formulated to answer exactly like you would if you asked another human. 9.9 times out of 10 the chatbots manage to answer your query successfully, a probability catastrophically higher than most of the population (excluding Mensa members!).
Here are the main similarities and differences between OpenAI’s ChatGPT and Google’s Gemini.
Access to information
Unfortunately, ChatGPT has a knowledge cutoff of the world and its events pre-September 2021, which means it cannot tell you anything that happened before September 2021. This flaw is perhaps all too human-like, unintentionally mimicking the flaws and blindspots in human knowledge – a mishap far from the point of OpenAI’s chatbot.
Despite its explosive advances on the AI scene, and perhaps shockingly, ChatGPT wasn’t initially connected to the internet. Since March 2023, this has been rectified.
If you’ve had a play around with ChatGPT, you’ll be all too familiar with the disappointing sinking feeling when you ask it to appraise the contents of a URL you copied and pasted into your prompt. ChatGPT responds with something like this:
“I'm sorry for any confusion, but as an AI language model developed by OpenAI, I don't have the ability to browse the internet or access external databases in real-time, including specific web pages or links you provide. My training only includes knowledge up until September 2021, and I generate responses based on that training.
“If you're looking for a summary or analysis of a specific article, I recommend providing key points or specific information from the article. Then, I can help answer questions or provide an analysis based on that information.”
How does it work then? Well, its answers are based on its knowledge base. But all is not lost – there’s a way around this. ChatGPT now incorporates plugins that help fetch updated third-party knowledge from the web. Got a PDF you want to read? Use a PDF plugin to pull that information into bite-size chunks.
On the flip side, Gemini has live internet connectivity, allowing it to provide real-time information. This is called “web scraping,” a technique that figuratively scrapes, or pulls data from a variety of web pages, like news articles, Wikipedia entries and academic papers – a function that ChatGPT lacks. "It involves sending HTTP or HTTPS requests to a web server and then sifting through the response to pluck out the desired data," according to ScrapeHero.
Medium of input and output
For users crunched for time, Gemini offers speech recognition for a hands-free user experience. Much like you can use speech-to-text recognition features in search, Google has understood the necessity for accessibility with its conversational chatbot. At the time of writing, ChatGPT only offers text-to-text questions and answers.
If you have a vast amount of customer data and want to put it through ChatGPT or Gemini, you might want to rein it in. You can’t go copying and pasting War and Peace (1867) in there, that’s for sure. The input limit for Gemini is 4,000 characters, which is equivalent to about 800 words. For ChatGPT, the character limit is even less, pulling in at around 2,048 characters, or “tokens.” This equates to somewhere under 500 words.
ChatGPT is trained on a dataset of text and code that includes over 500 billion words. This dataset includes a wide variety of text, from books and articles to code and scripts.
Gemini, on the other hand, is trained on a dataset of text and code that includes over 1.5 trillion words. That’s “trillion” with a “T”.
Now, we know it’s not all about the paint job – we’re not that shallow. But when it comes to tech, especially tech that’s adopted by 100 million daily users (and counting), the way things look does play an important role.
Gemini's interface is clean and easy to use, following the familiar style and layout associated with Google’s products. ChatGPT, on the other hand, uses smaller text and has an unfamiliar interface.
Both ChatGPT and Gwmini allow the user to dark mode to reduce eye strain and conserve battery life, handy if you’re going to be using a generative AI model to the maximum.
Accuracy of generative AI
Unfortunately, LLMs like ChatGPT and Gemini are only as good as the information fed to them. If you write a clear, succinct yet thorough prompt, you’ll reap the rewards.
Both models can be culpable of generating inaccurate or factually incorrect answers. It’s worth noting that both AI models are still under development, and most of these problems have been ironed out since their respective launches.
Like with any LLM, the more information, prompts and text you feed these tools, the more they’ll learn and evolve, and as a result, so will the quality of their responses.
How CSMs can craft great prompts for generative AI
In order to get the best out of ChatGPT or Gemini, it all starts with your prompt. The best way to ensure your AI-generated response is top tier is to start with the end in mind: what is your outcome? When writing your prompt, you’ll see success if you’re specific and concise, while being able to provide as much context as possible.
Here are the four simple steps you can take to draft a great prompt:
- Tell ChatGPT who to be, i.e. the persona.
- Provide as much context as possible. This could be company background information, a customer profile, customer usage data, a transcript of a call, or your product information.
- Be crystal clear about the end result you need the AI model to create. Be direct with your expectations – does it need to be a script, an email or a brochure?
- Don’t play by the book. If you don’t enforce creativity, it will churn out a general response lacking in flavor. Be clear with your instructions and give them a specific tone of voice and style to emulate.
Continue your AI education
The field of generative AI is swiftly advancing and holds immense potential to transform our customer interactions. Through the application of generative AI, companies can craft bespoke experiences that elevate customer satisfaction, loyalty, and retention.
How can ChatGPT and Gemini be used by customer success teams
It might seem obvious that a person working in marketing might use a conversational chatbot like ChatGPT or Gemini to aid their writing, whether it’s an email campaign or a blog post. But how does this fit in with customer success? In what ways can CSMs utilize these white-hot tools to make their lives easier?
First of all, the main USP of these AI models is to save time. Yes, efficiency’s well and truly the name of the game here. When used effectively, you can streamline countless processes to free up your time to focus on what really matters: providing exceptional customer experiences.
Here are the top key ways customer success teams can harness the power of generative artificial intelligence:
Customer onboarding and training
CSMs can utilize ChatGPT and Gemini to develop interactive customer onboarding and training materials, injecting energy into the process.
Let’s say you’ve got an introduction call with a new client and you want to learn as much as possible about their company. Usually, this would be a time-consuming and repetitive task, conflating several different websites and at least a dozen different searches for the bank of information.
Remember, you want to go into that meeting armed and fully prepared. You can use these models for a debrief of all the topline information (and then some) about their specific company.
Another neat way to maximize AI in your onboarding process is to use it to draft the basis of an epic onboarding email campaign. This could include welcome emails, check-in emails after a certain time period, and follow-up emails for non-responsive customers.
If you’re clever with your prompts, you can present your company information to the AI model, and get it to segment your customer base. You will need to provide the following information:
- Your company name,
- Your industry,
- The product you sell, and
- How it helps solve customers’ pain points.
You can even use these LLM models to generate a short, introduction video script to welcome the new customer aboard.
Create customer success playbooks
While ChatGPT is known for its songwriting, poem-crafting, and high-school essay-plagiarizing capabilities, its creativity can be applied to more strategic and action-led creations, too.
LLMs like ChatGPT and Gemini can generate suggestions and draft full documents – perfect if you’re strapped for time and need a new customer success playbook from scratch. You can use these models to generate a step-by-step process to ensure customer satisfaction.
Your playbook should provide ways to make your customers' lives easier. You can do that by identifying an issue, such as a high churn rate, and creating a series of actions your customer-facing team can follow to improve the situation for your users and business.
By feeding these AI models your customer’s profile, it can generate a robust playbook following a rough version of these steps:
- Set a goal for your playbook
- Work out where your playbook fits into your customer journey
- Identify the customers to prioritize
- Build out the assets you'll need
- Track and report your performance
Customer engagement and upselling
CSMs can use AI models to analyze customer data, find out their preferences, and consider their past interactions to generate tailored recommendations, cross-sell or upsell opportunities, and proactive notifications. What’s good about that? You can harvest that all-important intel to create targeted outreach strategies and foster long-term customer relationships.
With a canny use of prompts, you can understand why a customer might be ready to upgrade their subscription. This can be acted upon by asking ChatGPT or Gemini to write an email introducing a customer to a higher-tiered product.
A major sticky point for all CSMs is making sure the right resources are readily available to customers. CSMs can employ ChatGPT and Gemini to generate handy “how-to” landing pages or blogs to assist product education and create written FAQs based on common queries.
Conduct market and trend analysis
Like any good CSM worth their salt, staying ahead of the curve and implementing new trends that hit the customer success space is vital. But it’s hard juggling all your clients’ needs, your BAU tasks, plus market research. Realistically, there aren’t enough hours in the day.
By utilizing Gemini periodically (let’s say you pencil in some time to do this once at the beginning, or end, of the week), you can use its web-scraping capabilities to discover what real-time trends are floating around in the customer success community.
Here’s one we tried earlier:
CSC: Can you list any new trends emerging in customer success that I (a CSM) should adopt to deliver a better customer experience, increase product adoption, drive value and increase revenue?
Chat-GPT: Sure, customer success trends are constantly evolving and updating to reflect new technology, customer expectations, and business environments. As of mid-2023, some emerging trends that could impact the role of Customer Success Managers (CSMs) are…
In our example, ChatGPT then proceeded to list trends such as:
- Artificial intelligence and machine learning
- Customer success operations
- Customer health scoring
- Outcomes-based customer success
- Emphasis on customer education
- Product-led growth
- Expansion of the CSM role
- Increased use of self-service portals and communities
Problem-solving and troubleshooting
A genius way to harness the power of ChatGPT and Gemini is to analyze customer sentiment and feedback to help better understand customer satisfaction, identify potential areas for improvement, and track overall customer sentiment over time.
Both models can assist CSMs in providing effective customer support by generating responses to common customer queries. If you copy and paste a customer query into one of these tools, they can interpret the sentimental crux of the text and find the best way to provide targeted support, drive product enhancements, and resolve the issue in a timely and accurate manner.
Like most tidbits in this article, it’s worth clarifying that while ChatGPT and Gemini can provide valuable assistance to CSMs, they are fundamentally artificial intelligence. As augmentations of human consciousness, they should be treated with caution and never used without sense-checking. These tools shouldn’t be used with the goal of removing the human from the equation entirely.
… But which one is better for customer success teams?
If you’re using an AI model like ChatGPT or Gemini to produce creative, lengthy text, then ChatGPT is your best friend in this LLM war. Gemini has the advantage of scraping Google for every published insight, but currently, ChatGPT is better for creative content.
If you’re looking for shorter, factually accurate answers or advice, Gemini is your best bet; ChatGPT has a limit to its learning as it doesn’t scrape the entire internet like Gemini.
Currently, Gemini is in its “beta testing” phase, working out its kinks and becoming a smarter, more powerful tool. It’s hard to tell at this stage which LLM will come out on top, but for the time being, ChatGPT is the clear winner for creatively aiding customer success professionals.
Let’s wrap this up
Remember, ChatGPT or Gemini are machine learning models – they can never fully replicate the warmth and humanity of you, a CSM.
The choice between the two models ultimately depends on the specific requirements and use cases of the CSMs, and experimentation with both models can help determine which one better suits their needs.
The world of customer relationships and business as a whole is shifting, and the future of customer success will inevitably be linked to the progression of AI. Which AI model will be the preferred favorite by CSMs in a year’s time? At this point, who can say?
We will say that, as a Google product and tapped into the biggest search engine in the world, the potential for Gemii is limitless. Our parting words? Watch. This. Space.
It's an exciting time for AI in customer success. Be a part of the conversation.
Has this article got you thinking, "Actually, how is customer success going to change?"
If you're beginning to wonder about the impact of artificial intelligence, then add to the conversation by taking part in our AI survey. We're creating a brand-new report all about AI in CS to set the record straight and find out what are the ramifications for customer success.
Taking part in this multiple-choice survey takes less than 60 seconds. We'd love to hear your thoughts.