Go BackGarbage In, Garbage Out: Your Outputs Are as Smart as Your Data
Hey fellow marketer! Ever find yourself winding down by casually scrolling through LinkedIn or Instagram? Not to disconnect, but to check out thought leaders, marketing pages, or what other agencies are up to?
Same here.
I like to think I’m not wasting time watching brain rot reels, because I feel I am learning something even if I am relaxing.
Yesterday, as I was scrolling through my feed, I came across yet another post about how brands had exploited the latest Milan Fashion Week in a sustainable way. You know the type—breaking down past activations, and other things. I don’t remember exactly what it said, it was one of those forgettable, regurgitated takes, that the social media manager had to post.
I open the comments looking for something smart rather than the usual “Genius!” (this is probably the most common comment out there on LinkedIn, turning the genius into mediocre everyday).
- Comment 1: “Great insights! Totally agree that brands need to embrace sustainability in a meaningful way!”
- ➝ Reply 1.1: “Exactly! Sustainability is everything in today’s landscape!”
- ➝ Reply 1.2: “Couldn’t agree more! Brands that don’t take sustainability seriously will fall behind.”
- Comment 2: “So true! This is the future of marketing!”
- ➝ Reply 2.1: “100%! culture-driven marketing is where everything is heading!”
- ➝ Reply 2.2: “This is the way forward. Excited to see how brands adapt!”
- Comment 3: “Brands should be more innovative! We’ll see what happens next year!”
- ➝ Reply 3.1: “Absolutely! Looking forward to seeing who stands out!”
- ➝ Reply 3.2: “Innovation is key! Can’t wait to see future campaigns!”
They are not wrong, but at the same time, they are not insightful; they all have the same tone of voice, and none of them said anything actually useful or stimulating. As I kept reading, I felt the need to scroll down—like I had to dig deeper to find something good. It was the same feeling you get when you’re stuck waiting for a skippable ad to end just so you can get to the content you want.
This is what happens when you are using AI with bad inputs, and here is why this is a problem
Decent output is not enough anymore
Table of Contents
Before generative AI, producing consistent, decent-quality content was enough. Being present across multiple channels mattered more than crafting something truly great.
But now?
Decent is the standard. Decent is what is at the basis of a content/comment dump in social media.
And it is not only the social media! AI-generated content is flooding social media, reports, slides, and even ideas themselves. It’s technically correct but instantly forgettable.
To stand out you have to take two paths.
- You stop using AI for your interactions/productions (I am joking, no one is going to stop using it)
- You find a way to get your output better than the vast majority of the dump.
Good input for AI is the only way to break through.
As you already know, good prompt leads to better results. I tried to improve one of the comments I saw, so with some prompt engineering the:
“Exactly! Sustainability is everything in today’s landscape!”
can become:
“Sustainability is crucial, but many brands treat it as a marketing checkbox. Consumers see through vague eco-friendly claims—real sustainability demands measurable actions, not just greenwashed slogans.”
I asked it to go deeper and not be always agreeable as the AI tends to always agree. This is still a lot generic though, it went a bit deeper, but still it is blending among one of the many comments you see.
Thus, I tried adding a layer of data
“Sustainability matters, but reliability matters more. GWI data shows 41% of consumers—and 67% of Fashionistas—want more sustainable brands. Yet, 52% (rising to 72% for Fashionistas) prioritize reliability. Brands that don’t deliver on their claims risk more backlash than benefit.”
Please notice that also the last comment is AI generated, but I gave to the artificial intelligence not only the original content, but I also provided it with this screenshot:
Source: GWI 2025
You can do the same with any piece of data that you have, or you can retrieve it from the internet if you trust the source.
Human Centric good input for AI to generate campaign ideas
I took into consideration a LinkedIn comment scenario to make it easier, but you can use the same approach to generate any kind of content or ideas, if you have good inputs. This is why, when we segment a database using insights from GWI, Kantar Media, or our own CRM studies, we ensure that each segment is deeply defined, backed by multiple data points, and enriched with behavioral insights.
A well-structured segment provides a comprehensive view of motivations, habits, and decision drivers. We then translate this data into detailed descriptions, allowing you to understand each segment deeply. When it’s time to engage a specific audience, you can feed these insights into AI to generate more targeted, strategic, and creative ideas.
For example, in the Italian segmentation, there exist 2 very different segments, with similar demographics (young, high socioeconomic level compared to other young segments, present in the most important cities) but very different lifestyles.
Achievers:
- Ambitious & Status-Driven
- Active & healthy
- Trend-Conscious & Stylish
- Risk-Takers & Competitive
- Premium Consumers & Selective
Liberal Savvys:
- Socially Conscious & Ethical
- Curious & Knowledge-Seeking
- Pragmatic & Price-Sensitive
- Creative & DIY-Oriented
- Culturally Engaged & Thoughtful
Cover to Liberals and Achievers long segment descriptions, PDF documents that take into consideration hundreds of variables for each segment.
I uploaded the 2 20-page PDF documents to ChatGPT and asked it to come out with execution/content ideas relative to the Fashion Weeks, able to resonate with one of the two targets.
Here are the results
Ideas for Achievers:
- For Achievers: “Runway to Rooftop – Fashion Week VIP Lounge”
- Pop-up rooftop lounge near Fashion Week venues, offering exclusive beer tastings & networking with industry insiders.
- Collaborate with fashion editors & luxury stylists to host mini styling sessions while serving premium, limited-edition beers.
Ideas for Liberal Savvys
- For Liberal Savvys: “Wear & Sip – Upcycled Fashion & Craft Beer”
- Partner with upcycled fashion designers to host a live workshop, where guests customize thrifted fashion pieces while enjoying sustainable craft beer.
- Create a zero-waste fashion & beer pop-up, selling eco-conscious clothing & limited-run beers, with a portion of proceeds donated to sustainable fashion initiatives.
Good Inputs Produce Better Contents and Ideas
What can we notice from these quick AI-generated ideas?
- It generated ideas related to sustainability only if the target really cares about sustainability.
- The ideas are very tangible and even creative
How to improve even more? Uploading, for example, a description of the brand values and brand strategy, the brand assets and territories, the hard points of the brand. This extra step would ensure more accurate content for the brands. Give good input for AI and take your improved outputs.
Conclusion: Give Good Input for AI
AI is not the enemy of creativity; it’s a powerful tool. But like any tool, its value depends on how you use it. If you feed it vague, generic inputs, you’ll get forgettable results. If you provide structured, data-driven, and human-centric inputs, AI can help you generate content, strategies, and ideas that are actually insightful and relevant.
So, use good input for AI to streamline content creation and idea generation, making your work more efficient and ensure you’re crafting diamonds and not content overflow.
community@humancg.com 
+44 0203 693 4480 
London | Milano | Napoli | Krakow | Budapest