Launching a new product rarely feels steady. You are juggling half-formed assumptions, investor questions you cannot quite answer, and the pressure to get it right with limited resources. Public data can cut through some of the noise. Perplexity can pull fresh surveys, reports, and competitor benchmarks. ChatGPT can shape those findings into outputs you can actually use, like customer personas, SWOT tables, forecasts, or campaign ideas.
Used together, they turn open data into something you can act on, from refining your audience to validating what messages actually land.
With so much research now freely available online, startups do not have to rely on guesswork or pricey subscriptions. The challenge is keeping the process grounded because ChatGPT can drift into making things up if you let it. To keep your research accurate, set guardrails and keep a tight loop between what Perplexity finds and what ChatGPT produces.
Here are five ways to do it.
1. Getting Specific About Your Customers 🧠
Categories like “Gen Z” or “busy professionals” are vague. They are labels, not audiences. To know who you are building for, you need detail on values, habits, and trade-offs.
Perplexity Prompt:
"Find recent publicly available surveys or reports on UK consumers’ attitudes to sustainability in fashion, food, and household goods. Prioritise behavioural data on purchase frequency, willingness to pay, and barriers to adoption. Provide the report in PDF if available."
ChatGPT Prompt:
"From this report, build three customer segments. For each segment, include:
– A short name that captures their identity in plain language
– Key psychographics (values, motivations, contradictions)
– Buying behaviours with real percentages from the report
– Trade-offs they are willing or unwilling to make (price, convenience, ethics)
– One or two direct quotes or figures from the report to evidence each insight
– A confidence rating (High, Medium, Low) for each finding based on how explicit it is in the data After all segments, include a section titled ‘Source Integrity’ outlining:
– Data gaps or unclear metrics in the report
– Any contradictions between data points or interpretations Use only the information from the source. Mark missing content as ‘Not in source.’ Cite the full report name at the end."Personas built this way guide what you prioritise, who you launch with first, and how you speak to them.
2. Spotting Market Shifts 🔭
When people ask where the market is heading, vague answers are not enough. You need evidence of what might change in one year, three years, or five years.
Perplexity Prompt:
"Find recent publicly available forecasts on the eco-conscious product market in the UK. Focus on data that projects adoption curves, market size changes, or emerging technologies such as biodegradable packaging or resale platforms. Provide the most recent report available."ChatGPT Prompt:
"Break down the report into a forecast timeline:
– Next 12 months: visible or immediate shifts
– Next 2 to 3 years: predicted behavioural or structural changes
– Next 4 to 5 years: potential disruptions from technology, regulation, or culture For each time period, include:
– A short summary of key changes supported by a direct quote or data point
– A confidence rating (High, Medium, Low) based on how directly the report supports the claim
– Any contradictions or uncertainty noted between different parts of the report At the end, write a ‘Source Integrity’ summary covering:
– Overall reliability of the data
– Missing information or weakly supported forecasts
– Citations for all referenced reports."This stops you pitching an idea as “the future” when it is already in the past.
3. Understanding Competitors 🗃️
Saying “we are better than X” is not strategy. To compete, you need to know exactly where others are strong and where they fall short.
Perplexity Prompt:
“Find recent publicly available competitor comparisons of streaming platforms (Netflix, Amazon Prime, Disney+). Include pricing models, customer churn rates, feature usage data, and satisfaction scores.”
ChatGPT Prompt:
"Turn this into a comparative matrix with:
– Columns for each competitor
– Rows for price tiers, satisfaction levels, feature adoption, churn rates, and brand perception For each data point, include:
– A short quote or figure from the report as evidence
– A confidence rating (High, Medium, Low) depending on how explicit the data is Then write a short section called ‘White Space Analysis’ highlighting:
– Two areas where competitors overlap or underperform
– Opportunities or gaps that could be targeted by a new entrant End with a ‘Source Integrity’ note covering missing data, unclear comparisons, and citation details. Use only what is in the source."This shows you where to focus time and resources instead of chasing what everyone else is already doing.
4. Refining Campaigns with Real Data 🧾
Campaigns built on hunches waste money. If you have a small budget, you need proof of what has worked before.
Perplexity Prompt:
"Find recent case studies of eco-friendly product campaigns in Europe. Prioritise those with reported metrics such as conversion rates, cost per acquisition, and audience breakdowns. Provide links or PDFs where available."ChatGPT Prompt:
"For each campaign, create a structured summary with:
– Brand and product type
– Core message and positioning
– Target audience (demographics and psychographics)
– Channels used and approximate spend if mentioned
– Reported outcomes (engagement, conversions, sales)
– Creative tone or distinctive messaging styles
– Direct quotes or metrics from the case study as evidence
– A confidence rating (High, Medium, Low) for each key takeaway After summarising all campaigns, synthesise patterns in messaging, audience targeting, tone, and channel mix. Finish with a ‘Source Integrity’ section listing:
– Data gaps or unclear success measures
– Potential bias in reported results
– Full source citations."Patterns from past campaigns give you a shortcut to ideas that already work.
5. Making Feedback Useful 📋
After launch, feedback is messy. Reviews, survey data, and complaints can feel overwhelming. You need a way to see patterns without drowning in noise.
Perplexity Prompt:
"Find recent reviews, survey data, or reports on consumer satisfaction with health-conscious beverages in the UK. Focus on pain points related to packaging, taste, pricing, and sustainability. Provide the most recent available dataset."ChatGPT Prompt:
"Code this feedback into a table with four themes: Packaging, Taste, Price, Sustainability. For each theme, include:
– Recurring complaints or positives
– Exact numbers or percentages where available
– One or two verbatim quotes showing tone or sentiment
– Suggested product changes based only on this data
– A confidence rating (High, Medium, Low) based on how well the source supports each point Finish with a ‘Source Integrity’ section outlining:
– Data limitations, gaps, or vague areas
– Any contradictions or outliers in sentiment
– Full citation of the original report."This turns scattered opinions into structured evidence you can act on.
Keeping ChatGPT Grounded 🧮
When you pass Perplexity output into ChatGPT, it can drift. A few habits keep it tied to the source.
- Tell it to use only the content provided
- Ask for evidence and confidence ratings
- Include a “Source Integrity” summary in every output
- Keep inputs small, one report at a time
- Push for structured outputs like tables
If you want to take it further, set up a simple workflow that runs Perplexity searches on a schedule so competitor views or trend timelines stay fresh without manual effort. You can also save your preferred prompt structures inside a custom GPT or as a reusable template file, keeping your research system consistent over time.
Safe Prompt Template 🧬
Copy and paste this whenever you put Perplexity data into ChatGPT. It reduces hallucinations and keeps outputs grounded.
“Use only the information provided in the text below. Do not add external knowledge. For each insight, include an evidence anchor (quote or figure) and a confidence rating (High, Medium, Low). Mark missing or uncertain details as ‘Not in source.’ Present findings in a structured format such as tables or bullet points. At the end, include a short ‘Source Integrity’ section summarising data gaps, contradictions, and the report name.”
Final Musings 🌿
Perplexity gives you raw evidence. ChatGPT shapes it into something you can work with. Used together, they make public data practical, helping you find clarity in the noise. In the blur of startup advice, grounded research remains the quiet advantage that wins.
