A marketer's take on Chat GPT

A marketer's take on Chat GPT

Like many others, I have been playing around with large language models (mainly ChatGPT) over the last month. I think it’s going to be a game-changer for many enterprise functions, specifically marketing and sales. As Yogi Berra said, “It's tough to make predictions, especially about the future,” but I believe in 5-10 years, the vast majority of marketing and sales jobs that involve pure copywriting and design (for creating content such as marketing brochures, white papers, website copy, blogs and social media posts, sales/BDR emails) will be automated away (please don’t shoot the messenger!)

Amongst many other tests, here’re a few I did with ChatGPT to generate content:

Emails:

  • input a product webpage and summarize the benefits for an outbound email in 200, 100, and 40 words (the last one being important for BDR teams!)
  • adjust the tone of the email to friendly, formal, or funny
  • adjust the audience to executive-level or individual contributor

Blogs:

  • input a blog link and generate 3 other ideas related to that blog
  • identify keywords for optimizing those blogs
  • write a 500—1000 word blog for each related idea, using the previously-identified keywords
  • for each blog, generate ideas for titles, social posts, and email copy for an SDR
  • adjust the title (perhaps, optimize for click-bait!)

Other:

  • add a product description and generate a press release
  • input a blog link, generate 5 social posts that highlight different aspects. Optimize for sharing vs. commenting.

Depending on the test, the results ranged from OK to amazing, but in all cases, they were useable with minor tweaks.

What this means for marketing and sales tech: I think we’ll see a lot of integrations in the next 2-3 years to:

  • Auto-generate content that combines text and design (use case: input a Word draft for a case study or white paper, and get a cleaned up, designed doc in seconds)
  • Leverage email platforms (use case: define the product/solution, audience, and tone to auto-generate emails ready for hitting ‘send’. Another one: auto-generate a newsletter based on new blog posts or content in the last month)
  • Create the spectrum of promotional content for an asset (use case: create emails, banners, landing pages, social posts, all promoting a single white paper)

I’m not sure about where all of this is leading to in the mid/long term, but we will probably see a web flooded with AI-generated content where search results start looking eerily similar to each other. That would be depressing, but most companies would be willing to make the tradeoff between a lot of inexpensively created, “good enough” content vs. “hand-crafted” content, especially when it’d be hard to notice the difference (from a recent post in Nature: “An artificial-intelligence (AI) chatbot can write such convincing fake research-paper abstracts that scientists are often unable to spot them”).

A bunch of folks are looking at edge cases and concluding that “ChatGPT isn’t perfect.” That argument is a nonstarter. For many marketing/sales use cases, the results are more or less acceptable today and will get scarily better as the models iterate.

In a subsequent post, I will cover other topics such as video/audio content and demand gen.

Marketing/sales folks: what’s your reaction to Chat GPT?

Dan Matics

Senior Media Strategist & Account Executive, Otter PR

2 天前

Great share, Ashwath!

回复
Nathan Schulz

Salesforce Development & Architectural Consulting

1 年

It's possible that over time, a large quantity of the input into the language models will itself be model output - so there's a non-trivial risk of some kind of generational decay in quality, like when you copy the same physical medium too many times.

Great examples. I 100% agree with you, but I also think that Marketing & Sales professionals can up their skill sets by harnessing these tools coupled with their own expertise to produce even better results.

Nho Lam

Digital Marketing | GTM Stategy

1 年

Great insight Ash. As you've outlined, clear implications across many marketing functions here. Often times the biggest hurdle is outlining and writing that first draft—there's value in quickly whipping up an initial AI-generated draft and then passing that around for human feedback. With ChatGPT's (current) capabilities however, it will still need very detailed outlines and prompts to create quality content. Because of the technology's built-in biases (trending toward averages of current/past data, avoiding negativity, etc), it will be hard to generate cutting edge stuff without human input. The market may indeed be flooded with high volume, low quality AI generated content. However, I also see this technology as making it easier to generate high quality, AI assisted content. Maybe your company's SME/thought leader that's always too busy to write a blog or update your core messaging docs now can 3x his/her copy output. Or your favorite copy writer suddenly frees up time by leveraging AI generated content—able now to cherry pick high priority projects that require heavier human touch, while still maintaining content volume on lower priority items. Bottom line...there's going to be, like, so much more stuff out there now!

Ritwik Roy

Senior Equity Research Associate - Jefferies

1 年

Good read, hope you’ve been well!

要查看或添加评论,请登录

社区洞察

其他会员也浏览了