How To Write Engineering Policy Documents

How I Write an SOP That Actually Helps as a Program Manager at Amazon Most SOPs gather dust. Too long. Too vague. Too disconnected from the real work. At Amazon, a good SOP doesn’t just document a process. It makes the next person’s job easier…immediately. Here’s how I write SOPs that people actually use: 1/ I write it like a checklist, not a policy doc ↳ Clear steps ↳ Clear triggers ↳ No corporate speak Example: I once rewrote a 5-page doc into a 1-pager titled “How to Launch a New Data Feed.” Each step was 1 sentence, each had an owner. Adoption went up overnight. 2/ I start with the “when” and “why,” not just the “how” ↳ Why does this SOP exist? ↳ When should someone follow it? Example: I added a top section: “Use this when onboarding a new team to the dashboard. Purpose: prevent access issues and missed metrics.” That framing reduced questions by half. 3/ I link directly to the tools and templates ↳ No “search the wiki” ↳ Just: click → fill → done Example: Instead of “Use the onboarding tracker,” I write “Fill out this tracker → [link].” That one link removes 3 minutes of confusion. 4/ I include edge cases and common mistakes ↳ “If X happens, do Y” ↳ “Avoid this—it’s where people get stuck” Example: I once added a tip: “If permissions fail at Step 3, ping analytics-infra in Slack.” That one line prevented dozens of Slack threads. 5/ I test it with someone new ↳ If they’re confused, the SOP isn’t done ↳ Feedback closes the loop Example: I had a peer follow my SOP step-by-step, cold. Their questions helped me rewrite 4 sections before publishing. A great SOP doesn’t just live in Confluence. It lives in your team’s day-to-day execution. What’s your #1 tip for writing SOPs that actually get used?

If I had a dollar for every organization I've worked with where the SOPs were good, I wouldn't have a dollar. From my work with companies such as GSK, Novartis, and Pfizer, I hold that: 📋 SOPs must be functional above all else. Their purpose is to help people complete tasks successfully and safely, on time, with expected outcomes. ❌ But most SOPs fail because of: 1. Too Much Information • Every task 20+ steps • Information not concise or focused • Steps containing rationales (belongs in policy docs) • Poor titles that don't indicate task purpose Example of what NOT to do: "Please take a moment to review the testing documentation below." (It's not a favor—just write "Review the testing documentation") 2. Format & Language Issues ⚠️ • Walls of text without reading cues • No white space for visual breaks • Complex words where simple ones work ("utilize" vs "use") • Multiple actions crammed into single steps Real example of what NOT to do: "Remove one packet from the pouch and carefully add all contents to the water sample, swirl the sample until all the reagent dissolves into the solution." (That's 3 separate steps crammed into one!) 3. Structure Problems 🔍 • Steps not chronological • Sections bleeding into each other • Missing process mapping (critical for understanding flow) • Key information (like definitions) buried at the back ✅ The solution starts with three key principles: 1. Map Before Writing 🗺️Process mapping isn't optional; it's the foundation for any usable SOP (like your clinical trials, start with a protocol, not a prayer). 2. Write for Real Use ✍️One action per step, simple language (save the fluff for your cotton swabs). 3. Structure for Success 🎯Put key information where readers need it (hint: definitions belong up front, like your safety goggles). 💡 As I tell my pharma clients: "Will incorporating these concepts make your SOPs longer? Yes, sorry. Will it make them more usable? Yes, not sorry." ⚠️ Because in pharma, unusable SOPs aren't just inefficient—they're a compliance risk (or worse, accident) waiting to happen. Questions? AMA in the comments ⤵︎

Hey CTOs: Your 23-page brilliant technical manifesto is worthless if no one understands it. Harsh? Maybe. But I've watched countless great ideas die because technical leaders couldn't bridge the communication gap between the server room and the board room. As a CTO, you're not just fighting technical challenges—you're fighting a war of translation. Every day, you're trying to: ➝ Explain complex technical concepts to non-technical executives ➝ Rally your engineering team behind big-picture changes ➝ Get buy-in from people who speak a completely different language I spent years watching CTOs struggle with this (hell, I was one of them). They'd write 27-page technical manifestos nobody read, or try to wing it with vague "vision statements" that put everyone to sleep. But there's a better way. Most CTOs try to solve everything with one massive document. Instead, you need 2 documents: the vision doc and strategy doc. It's not sexy, it's not complicated, but it works. Here they are: 1. The Vision Doc This isn't your typical fluffy vision statement. It's a one-pager (yes, ONE page) that paints a crystal-clear picture of where you're headed. It covers:   ➝ Value proposition ➝ Capabilities needed ➝ Solved constraints ➝ Future challenges Most importantly, it speaks to both business and tech. 2. The Strategy Doc This is your practical roadmap built on 3 pillars: ➝ Diagnosis (the real problem) ➝ Policies (rules to keep you on track) ➝ Action items (your next concrete steps) No 5-year plans here—just clear, executable steps that get you moving. Think of it like this: Your Vision Doc is the destination, your Strategy Doc is the GPS. You need both. Look, the tech landscape is too complex and moves too fast for unclear communication. Having the right technical vision is pointless if everyone else can't understand it. I've got breakdowns on these docs coming in future content because I believe this is the biggest unlock for technical leaders right now. The days of hoping people "just get it" are over. Time to take real steps and level up your communication game.