Lesson 24 of 25

The Treasure of Specificity

A vague request gets an average answer, because the model has to pick a middle of the road you did not describe. Every concrete detail you add - a word count, a named place, a numbered structure, who is reading it - removes a guess it would otherwise make for you. Specificity is not about writing a longer prompt; it is about replacing the decisions you left open with the ones you actually want.

Reach for it whenWhen the first answer came back technically correct and completely generic - the kind of output you could have got for any company, any city, any reader.

Specificity works differently here, and more powerfully than people expect: on an answer engine a specific question returns better sources, not just a better-shaped answer. "Travel in India" returns listicles; "current ticket price for Golconda Fort for Indian nationals" returns the page that actually says so. The adaptation for creative chapters is the usual one - research the specifics here, write elsewhere - but note that it is doing real work, because the specifics most likely to be wrong in a detailed post are exactly the perishable ones.

Worth knowingIt will not write the 1200-word post, and it has no image generation at all, so the third example becomes a research task about what a stock-style image should contain rather than a generated image. On the data analysis, do not paste your own sales figures here: this tool is pointed at the open web, it is not where private business data belongs, and it cannot compute reliably over a pasted table anyway. Analyse in Gemini or ChatGPT; use this for the external benchmarks that analysis is missing.
  • A specific question returns better sources, not just better prose.
  • Research perishable specifics here; write them in elsewhere.
  • Do not paste private sales data - use it for benchmarks.
  • No image generation; research the brief instead.
  • Ask for each source's date, not just its link.
Example 1: The Vague Wish vs. The Specific Blueprint (Content Creation)

Instead of

Write a blog post about travel in India.

Perplexity version

**Write the post elsewhere; get its facts here.** Perplexity will not produce 1200 words of enthusiastic travel prose - it answers with citations rather than writing with voice. But the specifics that make a detailed itinerary useful are exactly the ones that go stale, and a confident wrong ticket price is how a well-written post loses a reader's trust.

Research this first, then take the answers into ChatGPT, Claude or Gemini to write.

I am writing a 3-day Hyderabad itinerary for first-time visitors. I need current, sourced facts.

1. Current opening days and hours for Charminar, Chowmahalla Palace, Salar Jung Museum and Golconda Fort. Cite the official site for each and give the date the page was last updated.
2. Current ticket prices at each, for Indian nationals and for foreign visitors, plus any separate camera or shoe-storage charge. Official sources only.
3. Is the Golconda Fort light and sound show currently running? Timings, languages, which evening in which language, and ticket price. This is the single detail most often wrong in these posts - it has been suspended before.
4. Which Hyderabad Metro lines and stations serve each of these places, and which of them the metro does not usefully reach? Cite the official metro map and route information.
5. What is the documented month-by-month temperature range in Hyderabad, and what do official or tourism sources say about the best months to visit? I want numbers, not "the cooler season".
6. Any current visitor restrictions, closure days, construction or conservation work at these monuments? Check recent news as well as official pages.
7. For Laad Bazaar and the Old City: what do recent sources say about market days, timings and when it is least crowded?

For every answer: the source link and the page's date. Where the only source is a travel blog rather than an official page, say so - a blogger's 2021 ticket price is not a current ticket price. Where you find nothing current, say so explicitly so I can write "check before you go" rather than invent a figure.
Open Perplexity 2,048 characters
Example 2: The Blurry Request vs. The Sharp Command (Business Analysis)

Instead of

Analyze my sales data.

Perplexity version

**Do not paste your sales data here.** Two reasons, both real. This tool is pointed at the open web and private business figures do not belong in it; and it is not built to compute reliably over a pasted table - it will produce plausible arithmetic you cannot check. Run the analysis in Gemini with the file attached, or in ChatGPT.

What this tool supplies is the thing a strict analysis of your own data deliberately excludes: the outside world. Your analysis correctly says "no industry benchmarks, nothing from outside the file" - so get those separately, sourced, and compare afterwards.

I run a [specific business - say, a mid-sized consumer products company selling in Telangana and AP]. I have just analysed one quarter of my own sales. I need external context, cited.

1. What published figures exist for my sector's quarterly growth in India over the last two years? Cite the reports, their publisher and their date, and say whether each is independent research or industry-association material.
2. What is documented about seasonality in this category in India - which quarters are strong, and how much of it is festival-driven? Cite sources. My own data is one quarter with no prior period, so seasonality is precisely what it cannot tell me, and this is the gap worth closing from outside.
3. What are the reported typical margins and revenue mixes for businesses of my size in this sector? Cite sources, and be plain about how thin the public data is for small and mid-sized Indian firms.
4. What published benchmarks exist for the metrics I am tracking - average order value, repeat purchase rate - for Indian consumer businesses? Flag which numbers come from vendor marketing rather than research.
5. What is reported about my sector's outlook for the coming two quarters, and what are the named sources of uncertainty?
6. Any GST, labelling or regulatory change coming that would affect my pricing or product mix? Link the official notifications.

Cite everything with dates. Where sources disagree on a market figure, show the disagreement instead of averaging it - market sizing estimates vary enormously and the range is more honest than a midpoint.

Then take this back to your analysis and ask one question there: where does my own data contradict these benchmarks? That is the interesting finding, and it needs both halves.
Open Perplexity 2,347 characters
Example 3: The Rough Sketch vs. The Detailed Commission (Image Generation)

Instead of

Create an image of a woman working.

Perplexity version

**No image generation here at all.** Perplexity cannot make the picture, and there is no reframing that changes that - generate it in Gemini, ChatGPT or a dedicated image tool.

There is one honest research role, though, and it is not a consolation prize: before you commission an image of a specific place and a specific kind of worker, it is worth knowing what that place and that work actually look like. A generated image of "a modern Hyderabad office" built from a model's averaged impression is how stock photography of India ends up looking like stock photography of anywhere.

1. What do Hyderabad's Hi-Tech City and Madhapur office buildings actually look like? Point me at photographs and architectural coverage with sources - the recognisable buildings, the skyline, what is visible from an upper floor. I want reference images to attach to the generation prompt.
2. What is documented about how Indian tech office interiors are actually laid out - open plan, desk density, what is on the walls? Cite coverage with photographs rather than a description.
3. What is written about accuracy and stereotype in stock imagery of Indian professionals - which visual clichés recur, and what do Indian designers and photographers say about them? Cite the commentary. This is the most useful answer on the list: knowing the cliché is how I keep it out of my prompt.
4. What do current sources say about usage rights for AI-generated images in India, particularly for commercial use? Link the platform terms and any Indian legal commentary. If this image goes on a company blog, that is a question I need answered before I generate it, not after.
5. Are there Indian photographers or image libraries documenting Indian workplaces whose work I could licence instead? Cite them with links. Sometimes the right answer to an image brief is to pay a photographer, and no generation tool will ever suggest that.

Cite everything with dates, and say where you could not find a reliable source.

Then take the reference images and the list of clichés to avoid into Gemini or ChatGPT, and write the detailed prompt there - subject, setting, clothing, camera - with your reference photographs attached. The research makes the specificity real rather than invented, which is the whole point of this chapter.
Open Perplexity 2,296 characters