This material has been reproduced and communicated to you by or on behalf of Kaplan Business School pursuant to Part VB of the Copyright Act 1968 (the Act).
The material in this communication may be subject to copyright under the Act. Any further reproduction or communication of this material by you may be the subject of copyright protection under the Act.
Do not remove this notice.
In keeping with today's topic — the appropriate use of GenAI — parts of this content used GenAI to scaffold concepts, draft examples, and design activities that build critical evaluation skills.
The deck used OpenAI's ChatGPT and Perplexity to generate initial ideas, which were then refined and supported with peer-reviewed sources. This slide models the practice you will apply in your own work: use the tool, then verify, refine, and acknowledge it.
Weeks 1–6 built the measurement skills. This week adds a tool that helps you produce and analyse content at scale — which feeds directly into visualisation (Week 8) and strategy (Week 9).
| LO1 | Critically evaluate the role of generative AI in marketing and social media contexts. |
|---|---|
| LO2 | Analyse and optimise AI-generated content using prompt engineering and performance metrics. |
| LO3 | Design AI-enabled strategies using ethical frameworks. |
We follow one brand — BrewLab, a Melbourne specialty coffee retailer — through every step: from a marketing idea, to a written prompt, to sentiment analysis, to synthetic research, to a responsible-use decision.
Each section ends with a short knowledge check. Two hands-on activities run inside Section 3.
The size of the opportunity, and the question it raises about creativity. LO1
The numbers point to a clear shift. GenAI moves the marketer's role away from producing every draft by hand, and toward directing, judging, and refining what the tool produces.
A 5–15% productivity lift is an average from studies. Your result depends on how well you brief the tool and how carefully you check its output. The skill you build today decides whether GenAI helps or harms your work.
Coca-Cola's holiday activation invited the public to make their own branded artwork using OpenAI's tools and DALL-E.
Where does human creativity end and machine creativity begin in digital marketing?
Hold this question. It returns when we discuss appropriate use and academic integrity later today.
BrewLab is a Melbourne specialty coffee retailer. It runs cafés, sells beans and home brewing gear, ships online across Australia, and runs a loyalty app.
Every activity today uses BrewLab. This keeps the examples consistent with Weeks 1–6 and lets you focus on the method rather than a new scenario each time.
The prompt techniques you learn apply to any brand. BrewLab is simply the case we practise on.
Four application areas for social media and marketing analytics. LO1
GenAI can produce text, images, and video quickly and consistently. This lets a small team keep a steady posting schedule across several platforms.
BrewLab launches a summer cold brew. GenAI drafts an Instagram caption, a LinkedIn note for wholesale partners, and three subject lines for the customer newsletter — all in one sitting, all in BrewLab's voice.
GenAI can tailor messages to what a person has done or shown interest in. Matching the message to the audience raises engagement and conversion.
Adjust wording, offers, and timing based on user behaviour. A returning customer and a first-time visitor see different messages.
Answer questions in real time and guide customers through the store or app, day or night.
A BrewLab app chatbot recommends a lighter roast to a customer who usually buys light beans, and offers a wholesale sample pack to a café owner browsing bulk options.
Read the tone of reviews and comments to track how people feel about the brand, in real time.
Forecast trends and estimate how a campaign is likely to perform before you spend on it.
Compare two versions of a post or ad and shift toward the stronger one, faster.
You will study predictive modelling in Week 11 and A/B testing in later weeks. Today we focus on sentiment analysis, which you can run with GenAI right now.
BrewLab feeds a week of Google reviews into GenAI and learns that most complaints share one theme: slow online order dispatch. That single insight guides the next fix.
GenAI can generate synthetic data — realistic but invented responses and datasets. This gives teams quick, low-cost material for early research.
Synthetic data is invented. It is useful for prototyping and teaching, but it is not evidence about real customers. Section 5 covers exactly what it can and cannot be used for.
| Application area | Example uses | Main benefit |
|---|---|---|
| Content creation | Posts, images, videos, ad copy | Speed, consistency, creativity |
| Personalisation | Tailored messages, recommendations | Higher engagement and loyalty |
| Analytics and insights | Sentiment analysis, trend forecasting* | Deeper, actionable insight |
| Market research | Synthetic survey data, demographic insight | Faster, lower-cost research |
| Automation | Chatbots, A/B testing*, scheduling | Greater efficiency and scale |
*Covered in later weeks.
Talking to GenAI so it gives you output you can use. LO2
A prompt is the instruction you give a GenAI tool. Prompt engineering is the skill of writing that instruction clearly enough to get useful, on-brand output.
The quality of the output depends on the quality of the prompt. Compare these two:
Vague. The tool has to guess the platform, audience, tone, length, and goal — so it returns something generic.
Specific. The tool knows exactly what to produce. Next slide breaks this prompt into its parts.
You will rarely need all six every time. Start with task, audience, and format, then add role, tone, and goal when the output needs sharpening.
The same tool, the same brand — the only thing that changed is the instruction.
BrewLab is launching its first summer product line. In small groups, act as the marketing team.
value price, energy, authenticity
trust-driven, value quality and consistency
detail-oriented, value supply and margin
Notice that improving the output almost always means improving the prompt. That editing loop is prompt engineering.
Product: BrewLab's new bottled cold brew, sold in supermarkets.
Sentiment analysis with GenAI — and where it gets tone wrong. LO2
Sentiment analysis sorts text into Positive, Neutral, or Negative. GenAI can do this and explain its reasoning.
Classify three real-sounding BrewLab reviews. Then ask: does the result match how a customer would actually feel? Does it fit the audience the review is about?
Some text uses positive words to mean the opposite. Word-spotting alone fails here, which is why we ask GenAI to reason, not just label.
Punctuation, capitals, and context carry meaning. A model that only counts positive words will miss sarcasm — and misread how customers really feel. Always check tricky cases by hand.
Generating research material with GenAI — and reading it critically. LO2
Good customer data is hard to get. Surveys are slow and costly, and response rates are low. GenAI offers a shortcut — synthetic data — with a trade-off attached.
Some studies report AI-generated responses matching human data closely — up to around 90% correlation — under specific conditions. Read that as a possibility, not a promise.
Scenario. A consulting team advises BrewLab on entering a new city. You want quick qualitative material from target segments to shape better questions.
Identify recurring words and sentiments — trust, sustainability, taste, price. Group the responses using a simple coding method (thematic analysis). Use the themes to design your real interview questions.
Scenario. BrewLab is testing product-market fit for a monthly coffee-bean subscription aimed at home brewers.
Paste the table into a spreadsheet and inspect the spread of answers.
Use Power BI or Tableau to visualise the data with bar and pie charts — a bridge to Week 8.
Scenario. BrewLab's analytics team is launching an app feature for Gen Z coffee drinkers and needs draft personas to guide the campaign.
A persona is only a starting point. The next slide shows how each part of a persona turns into a marketing decision.
| Persona element | How you use it | BrewLab example |
|---|---|---|
| Motivations | Shape ad messaging and content themes | Highlight fast, affordable coffee for busy students |
| Pain points | Find gaps for fixes or how-to content | Explain the app's reorder feature clearly |
| Digital behaviour | Choose channels and posting times | Post TikTok clips at night, when Gen Z is active |
| Platform preference | Split budget across the right platforms | Persona A on Instagram, Persona B on Reddit |
| Buying habits | Set promotions and pricing | Offer a student discount to price-sensitive users |
| Language and tone | Match copy to each segment | Playful for one persona, plain for another |
Synthetic data reflects patterns in the model's training data, not BrewLab's actual customers. It can smooth over the niche segments and outliers that matter most, and it can carry hidden bias. Presenting it as real research is inappropriate use — a point we return to in Section 7.
Six risks, a set of best practices, and a framework for using GenAI well. LO3
GenAI needs large-scale data and often lacks clear consent.
Get explicit permission, be transparent, protect data.
Models can reinforce existing societal biases.
Use diverse data, de-biasing methods, and regular audits.
AI decision-making can be hard to see into.
Disclose data sources and decision logic; set oversight.
GenAI can spread false or harmful content.
Fact-check strictly and avoid deceptive practice.
Generated content can infringe others' rights.
Use licensed data and track content provenance.
Targeting can exclude or stereotype groups.
Promote inclusive strategies; avoid invasive profiling.
These risks are live for BrewLab too. Personalisation that excludes a group, or an AI image that copies another brand's work, creates real harm and real liability.
The common thread: a human stays responsible at every step. GenAI drafts and suggests; a person consents, checks, and decides.
A simple loop for using GenAI well: Problem → AI → Interaction → Reflection, then back to the problem.
Define the problem before seeking a solution. Focus on who you are solving for and what success looks like. Explore root causes before jumping to an answer.
Choose the right tool for the task. Different tools suit different jobs. Match the tool to the purpose rather than forcing a fit.
Evaluate output for accuracy, relevance, and clarity. Check sources, flow, and audience fit. Refine through feedback loops, and keep human oversight — verify, document, reflect.
Tools, appropriate use, and how to reference GenAI in your work. LO1 · LO3
| Tool | Use in marketing / social media | Strengths | Limitations |
|---|---|---|---|
| ChatGPT | Copy, tone refinement, ideation | Versatile, strong natural language | Can generate inaccurate content |
| Jasper AI | Copywriting, branding, strategy | Marketing templates, tone control | High cost, limited transparency |
| Claude | Long-form synthesis, summaries, evaluation | Long context, explainable reasoning | Fewer plug-in integrations |
| Perplexity AI | Fact-based content, research summary | Includes citations, concise | Limited creative generation |
| Canva Magic Write | Social content with design, captions | Visual and text together | Text output less flexible |
| Copy.ai | Product copy, email, captions | Fast ideation, easy to use | Generic tone, little customisation |
Match the tool to the job. Research summaries suit Perplexity; long synthesis suits Claude; quick copy suits ChatGPT or Copy.ai.
| Appropriate use | Inappropriate use |
|---|---|
| Brainstorming ideas, outlines, images, or draft structures | Submitting AI-generated text as your own work |
| Summarising papers or concepts to aid comprehension | Using AI to avoid reading assigned texts |
| Refining or proofreading your own writing | Using AI to complete whole assessments without synthesis |
| Exploring different perspectives in discussion | Using AI as a ghostwriter for reflective or critical tasks |
| Generating synthetic data to test models, if permitted | Presenting AI-generated data as empirical research |
This is the Coca-Cola creativity question again — applied to your own work.
No generative AI allowed.
You may use generative AI for research and content generation that is appropriately referenced.
You must use generative AI to complete the assessment.
Always check which level applies to each assessment before you start.
If you use GenAI to brainstorm or generate examples, acknowledge it in your methodology, introduction, or notes.
"This report used OpenAI's ChatGPT (2025) to generate initial campaign ideas, which were then refined and supported using peer-reviewed sources and real market data."
OpenAI, 2025. ChatGPT [AI language model]. Available at: https://chat.openai.com [Accessed 30 May 2025].
Include your prompts and the AI's responses in an appendix. Full referencing guidance: library.kaplan.edu.au → referencing other sources → generative AI. For support, contact the Academic Success Centre.
The theme across all seven sections: use the tool, then verify, refine, and acknowledge.
Data Visualisation and Storytelling for Stakeholders. The survey data you generated today becomes a chart that tells a clear story.
Support: Academic Success Centre · elearning.kbs.edu.au/course/view.php?id=1481 · Referencing: library.kaplan.edu.au
Press T or Escape to close