{"id":53384,"date":"2025-08-21T20:54:32","date_gmt":"2025-08-21T20:54:32","guid":{"rendered":"https:\/\/dhunterdigital.com\/?p=53384"},"modified":"2025-08-21T20:58:20","modified_gmt":"2025-08-21T20:58:20","slug":"how-long-does-a-chatgpt-prompt-really-need-to-be","status":"publish","type":"post","link":"https:\/\/dhunterdigital.com\/index.php\/2025\/08\/21\/how-long-does-a-chatgpt-prompt-really-need-to-be\/","title":{"rendered":"How Long Does a ChatGPT Prompt Really Need to be?"},"content":{"rendered":"<p style=\"font-weight: 400;\">I asked ChatGPT the same question twice \u2013 first with a 19 word \u201cShort\u201d prompt and then with a 354 word \u201cFull\u201d prompt that followed best practices.<\/p>\n<p style=\"font-weight: 400;\">Did it make a difference?\u00a0 In this case, it made a huge difference: bigger is better when it comes to GenAI. But\u2026 there should be some structure to your long prompt.<\/p>\n<h2 style=\"font-weight: 400;\"><strong>The Prompts<\/strong><\/h2>\n<p style=\"font-weight: 400;\">The Short prompt was the type of simple request I would give an experienced and autonomous Senior Analyst.\u00a0 The Full prompt represented the structured guidance that I would give an Intern.<\/p>\n<p style=\"font-weight: 400;\">The Short prompt was simply: \u201cPlease compare the risk factors in Lattice Semiconductor&#8217;s 2024 and 2023 10-K SEC filings and note what has changed.\u201d<\/p>\n<p style=\"font-weight: 400;\">The \u201cTask\u201d part of the Full prompt was not that much longer:<\/p>\n<ul>\n<li>\u201cYour task is to compare the risk factors that the company\u2019s management listed in their SEC 10-K filings for the last two years. Please summarize:\n<ul>\n<li>What are the major risk factors?<\/li>\n<li>What risk factors are new or have become more urgent in the most recent year?<\/li>\n<li>What risk factors from last year have been dropped or downgraded?\u201d<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p style=\"font-weight: 400;\">What made the Full prompt 354 words vs. the Short prompt\u2019s 19 words were the additional sections of the prompt beyond the task that I added:<\/p>\n<ol>\n<li>Your Role \u2013 <u>how<\/u> I want ChatGPT to act<\/li>\n<li>Task Goal &amp; Intent \u2013 <u>why<\/u> I asked this question<\/li>\n<li>Needed Analysis \u2013 <u>what<\/u> the question is<\/li>\n<li>Sources &amp; Context \u2013 <u>where<\/u> to get the information and <u>what<\/u> ChatGPT needs to know<\/li>\n<li>Audience \u2013 <u>who<\/u> is going to review the answer<\/li>\n<li>Output \u2013 <u>how<\/u> to present the answer and document ChatGPT\u2019s process ChatGPT<\/li>\n<li>Guardrails \u2013 <u>controls<\/u> to help make sure that ChatGPT answers the right question from the right sources.<\/li>\n<\/ol>\n<p style=\"font-weight: 400;\">Note: you can read the Full prompt at the end of this post.<\/p>\n<h2 style=\"font-weight: 400;\"><strong>The Results<\/strong><\/h2>\n<p style=\"font-weight: 400;\">First, some metrics:<\/p>\n<table style=\"font-weight: 400;\">\n<tbody>\n<tr>\n<td width=\"138\"><\/td>\n<td width=\"126\">Short Prompt<\/td>\n<td width=\"108\">Full Prompt<\/td>\n<\/tr>\n<tr>\n<td width=\"138\">Prompt Length<\/td>\n<td width=\"126\">19 words<\/td>\n<td width=\"108\">354 words<\/td>\n<\/tr>\n<tr>\n<td width=\"138\">Run Time<\/td>\n<td width=\"126\">1m 7s<\/td>\n<td width=\"108\">1m 36s<\/td>\n<\/tr>\n<tr>\n<td width=\"138\">Response Length<\/td>\n<td width=\"126\">432 words<\/td>\n<td width=\"108\">901 words<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p style=\"font-weight: 400;\">There was not much difference in run time between the two prompts, but response length doubled with the Full prompt. The real difference, as we will see below, is in the quality of results, and the resulting confidence in the results.<\/p>\n<p style=\"font-weight: 400;\">The answers from the Short and Full prompts were similar in that they both included an overall summary and noted the same changes. The big difference was in the depth of analysis and quality of presentation.<\/p>\n<p style=\"font-weight: 400;\">While the answer from the Short prompt was three simple bulleted lists, the Full prompt delivered much more depth across nine distinct sections:<\/p>\n<ol>\n<li>Executive Summary<\/li>\n<li>Major Risk Factors \u2013 Comparison Table<\/li>\n<li>What\u2019s New or More Urgent in the Most Recent Year<\/li>\n<li>What Was Dropped or Downgraded<\/li>\n<li>Side-by-Side Risk-Factor Mapping<\/li>\n<li>Implications and Watch List<\/li>\n<li>Short \u201cChange Log\u201d<\/li>\n<li>Sources<\/li>\n<li>Required Administrative Details<\/li>\n<\/ol>\n<p style=\"font-weight: 400;\">Reading through the whole answer gave me confidence that ChatGPT had executed what I asked for. Also, because of the way the prompt was structured, I had output that was ready to be pasted into an executive presentation.<\/p>\n<p style=\"font-weight: 400;\">Now for the caveats\u2026<\/p>\n<ul>\n<li>The content of the answer looked correct and easily understandable. However, given where we are in the evolution of GenAI it would be smart for me to skim the source documents.<\/li>\n<li>Also, while exhaustive, the report from the Full prompt was not perfect, e.g. the summary\u2019s title was \u201cExecutive Summary (\u2264 200 words).\u201d<\/li>\n<\/ul>\n<p style=\"font-weight: 400;\">So, ChatGPT rose to the challenge of the Full prompt, but for me there is no direct copy-paste into an email for the boss \u2013 it\u2019s work, but less work than doing the whole thing myself.<\/p>\n<h2 style=\"font-weight: 400;\"><strong>Conclusion<\/strong><\/h2>\n<p style=\"font-weight: 400;\">Your prompt needs to be long enough to get the quality of results that you want.<\/p>\n<ul>\n<li>Short prompts are great for exploratory research and brainstorming<\/li>\n<li>Full prompts are best for detailed research and client facing output<\/li>\n<\/ul>\n<p style=\"font-weight: 400;\">If you are asking ChatGPT to \u201cfind the best burger joint in Portland\u201d then you can use a short prompt as the stakes are low.\u00a0 If you want ChatGPT to \u201cfind an acquisition target under $1B\u201d then you better use a Full prompt, and likely several of them.<\/p>\n<p style=\"font-weight: 400;\">One practical note\u2026 you need to evaluate if you can do the work faster than ChatGPT. If you\u2019re writing a 300 word prompt to get a 500 word blog post which you then need to edit, then you might want to just write the blog yourself.\u00a0 But, if you are running competitive scans weekly, then it makes sense to develop a Full prompt for ChatGPT and keep it in a document for re-use.<\/p>\n<h2 style=\"font-weight: 400;\"><strong>AI Content Statement<\/strong><\/h2>\n<p style=\"font-weight: 400;\">All text in this post is human generated, written by me.\u00a0The image was generated by ChatGPT-5 with DALL-E<\/p>\n<h2 style=\"font-weight: 400;\"><strong>The \u201cFull\u201d Prompt<\/strong><\/h2>\n<p style=\"font-weight: 400;\">Technical Note: if you are going to try this prompt I suggest that you first <a href=\"https:\/\/dhunterdigital.com\/index.php\/2025\/08\/21\/how-to-use-nested-bullets-in-chatgpt-prompts\/\">convert it to Markdown<\/a>.<\/p>\n<ul>\n<li>[Your Role] You are working as a senior consultant at The Boston Consulting Group (BCG).<\/li>\n<li>[Task Goal &amp; Intent] You are building a comprehensive analysis of Lattice Semiconductor (NASDAQ: LSCC) and as part of that we need to understand the business risks that that company is facing.\n<ul>\n<li>Note that here after \u201cLattice Semiconductor\u201d is referred to as \u201cthe company.\u201d<\/li>\n<\/ul>\n<\/li>\n<li>[Needed Analysis] Your task is to compare the risk factors that the company\u2019s management listed in their SEC 10-K filings for the last two years. Please summarize:\n<ul>\n<li>What are the major risk factors?<\/li>\n<li>What risk factors are new or have become more urgent in the most recent year?<\/li>\n<li>What risk factors from last year have been dropped or downgraded?<\/li>\n<\/ul>\n<\/li>\n<li>[Sources &amp; Context]\n<ul>\n<li>Get your information only from the company\u2019s fiscal year 2023 and fiscal year 2024 SEC 10-K filings.<\/li>\n<li>Use the 10-K files attached to this prompt titles \u201cxxx.pdf\u201d and \u201cyyy.pdf.\u201d<\/li>\n<li>For context use the attached file titled \u201czzz.pdf.\u201d<\/li>\n<\/ul>\n<\/li>\n<li>[Audience] Your report will be reviewed by the senior partners at our BCG office. Please write appropriately for that level.<\/li>\n<li>[Output]\n<ul>\n<li>Present your findings in a structured manner:\n<ul>\n<li>1) use text and tables as needed,<\/li>\n<li>2) format the findings such that they are suitable for pasting in a MS Word Document or MS PowerPoint slides, and<\/li>\n<li>The findings should be no longer than 2,500 words.<\/li>\n<\/ul>\n<\/li>\n<li>In addition to the findings, please list the following information:\n<ul>\n<li>1) all sources that you used. For each source note where you found it, the URL, and the content creation date<\/li>\n<li>2) the LLM or agent name and version that was used to fulfill this request.<\/li>\n<li>3) the date and time that this request was fulfilled.<\/li>\n<li>4) the requesting prompt.<\/li>\n<\/ul>\n<\/li>\n<li>[Guardrails]\n<ul>\n<li>Ask me questions, one at a time, if you need any clarification on this request.<\/li>\n<li>Use only reliable sources. For example:\n<ul>\n<li>1) do not pull information from Reddit or social media, and<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">2) do pull information either from the SEC EDGAR database or the company\u2019s website <\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>I asked ChatGPT the same question twice \u2013 first with a 19 word \u201cShort\u201d prompt [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":53387,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[45,9],"tags":[40,46,50,47,51],"class_list":["post-53384","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai","category-technology","tag-ai","tag-chatgpt","tag-competitor-analysis","tag-prompting","tag-strategic-planning"],"_links":{"self":[{"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/posts\/53384","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/comments?post=53384"}],"version-history":[{"count":2,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/posts\/53384\/revisions"}],"predecessor-version":[{"id":53386,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/posts\/53384\/revisions\/53386"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/media\/53387"}],"wp:attachment":[{"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/media?parent=53384"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/categories?post=53384"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/dhunterdigital.com\/index.php\/wp-json\/wp\/v2\/tags?post=53384"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}