惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Recent Announcements
Recent Announcements
J
Java Code Geeks
U
Unit 42
GbyAI
GbyAI
大猫的无限游戏
大猫的无限游戏
L
LangChain Blog
D
Docker
F
Fortinet All Blogs
N
Netflix TechBlog - Medium
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
罗磊的独立博客
I
InfoQ
The Cloudflare Blog
小众软件
小众软件
V
Visual Studio Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
Engineering at Meta
Engineering at Meta
S
SegmentFault 最新的问题
爱范儿
爱范儿
Hugging Face - Blog
Hugging Face - Blog
P
Proofpoint News Feed
V
V2EX
月光博客
月光博客
Martin Fowler
Martin Fowler

jola.dev

Migrating your Bluesky account the hard way | jola.dev Cluster singleton pattern | jola.dev cove.town, atproto self-hosted self-hosting | jola.dev Speeding up a Phoenix LiveView web app with a CDN | jola.dev Self-hosting an atproto container registry | jola.dev Migrating to the new Tangled knot2 | jola.dev Self-hosting and Tangled | jola.dev Self-hosting your PDS | jola.dev Taking control of your atproto account | jola.dev No cost, no value | jola.dev Latch - an Elixir atproto OAuth library | jola.dev Limited output is a feature | jola.dev Distributed rate limiter with HRW in Elixir | jola.dev A computer can never be held accountable | jola.dev Elixir Cluster 101 | jola.dev How to stop Claude from saying load-bearing | jola.dev Let libraries be libraries | jola.dev CI workflows on Tangled for Elixir | jola.dev Automatically syncing your blog to atproto and standard.site | jola.dev Appreciation for the small web | jola.dev Treating LLMs as programming books Publishing your blog to standard.site in Elixir Generating OG images in Elixir The social contract of writing Highest Random Weight in Elixir bunnyx: a bunny.net Elixir client library Building for the joy of building Running local models on an M4 with 24GB memory How to hit your Claude weekly limit so you can go outside and touch grass Dropping Cloudflare for bunny.net
Estimates Are More Valuable Than You Think
Johanna Larsson · 2025-08-04 · via jola.dev

Disclaimer: This blog post was written in 2025 during a time when I experimented with LLM-assisted writing.

I know estimates have a bad reputation. Most engineers hear "estimate" and immediately think of micromanagement, unrealistic deadlines, and that manager who asks "is it done yet?" every few hours. I've seen teams reflexively pad their numbers by 3x just to avoid the inevitable disappointment when reality doesn't match the plan.

But after years of building products, I've come to see estimates differently. Good estimation isn't about control or tracking every minute of your day. It's about creating trust and transparency in your team.

Estimates as Conversations, Not Contracts

When I give an estimate, I'm not making a promise or signing a contract. I'm starting a conversation. Saying "this should take about two days" immediately opens up important discussions. Someone might point out a dependency on the auth service that adds another day. Another teammate might ask if we're including the database migration in that estimate. These conversations surface hidden complexity before it becomes a problem.

The real value comes when estimates become a shared language between you and your stakeholders. Instead of being rigid deadlines that everyone dreads, they become reference points for meaningful discussions about progress and priorities.

Creating Visibility Without Surveillance

I've watched teams completely transform once they got comfortable with estimation. Not because they suddenly became perfect at predicting the future (nobody is), but because they created visibility into their work.

Your PM knows when to check in without being intrusive. Your teammates know when you might need help without you having to ask. Your users have realistic expectations about when that feature they've been waiting for might actually ship. No more surprise delays that derail quarterly planning. No more vague "it'll be done when it's done" responses that frustrate everyone involved.

The Teams That Ship

Here's what I've noticed about teams that consistently deliver: they're not necessarily the ones who code faster or work longer hours. They're the ones who communicate better. They treat estimates as a tool for collaboration rather than a source of stress.

These teams understand that missing an estimate isn't failure if you've been transparent about why. They know that adjusting an estimate based on new information isn't admitting defeat, it's being professional. They've learned that honest communication about progress and blockers builds more trust than any perfectly hit deadline ever could.

Good estimation is really about creating a culture where everyone understands what's happening, why it's happening, and when things might change. It's about replacing anxiety and uncertainty with clarity and shared understanding.

What's been your experience with estimation? Have you found ways to make it work for your team, or is it still a source of tension?

Written by Johanna Larsson. Thoughts on this post? Find me on Bluesky at @jola.dev or why not give it a vote on Bubbles.