Developer listening
Where developers talk about your product in 2026
Stack Overflow receives fewer new questions, while developer discussions also appear on GitHub, Reddit and other communities. Compare channel-specific evidence before changing your monitoring coverage; the available studies do not measure a direct migration.
Takeaways
- Choose monitoring sources from product questions, failures and evaluations. Specify product and package identifiers, exclusions and a response owner; test the coverage against relevant posts.
- Repository activity, new-question counts and survey responses measure different behavior. They cannot establish where the same developers moved, and public listening cannot reveal private AI conversations.
- Stack Overflow question volume fell to levels last seen at its 2009 launch by May 2025.[1]
- GitHub passed 180 million developers in 2025 and merged an average of 43.2 million pull requests a month, up 23% year over year.[2]
- The 2025 Stack Overflow survey reports that about 35% of respondents visit Stack Overflow because of AI-related issues at least some of the time.[3]
- Reddit has data partnerships with Google and OpenAI. Separate citation studies also identify Reddit among frequently cited domains.[4]
What happened to Stack Overflow?
The cited analyses report a long decline in new Stack Overflow questions, with a sharper decline after ChatGPT launched. The timing alone does not establish how much of that change AI tools caused.
Using Stack Exchange's own Data Explorer, The Pragmatic Engineer showed question volume falling since about 2014, then dropping sharply after ChatGPT launched in November 2022.[1] By May 2025, monthly questions had slumped to levels last seen when the site launched in 2009. Data Explorer figures reported by DevClass and Futurism put December 2025 at about 3,600 questions, against roughly 100,000 a month at the start of 2023.[5]
New Stack Overflow questions per month
New-question counts do not measure total readership. Stack Overflow's 2025 survey describes continued use of the site, including visits to resolve problems with AI tools.[3] These figures do not establish whether individual visitors changed how often they asked questions.
Keep Stack Overflow in your monitoring coverage when it contains relevant questions. Do not use its declining question volume to estimate the share of your product discussions occurring elsewhere.
Where did the questions go?
The cited studies do not track the same developers across these channels. They show substantial use of AI tools alongside continued use of public developer communities.
OpenAI reported more than 900 million weekly ChatGPT users in February 2026.[6] In the 2025 Stack Overflow survey, 84% of respondents said they use or plan to use AI tools, and 51% of professional developers reported daily use.[3] Those conversations are private. You cannot listen to them directly.
Private AI conversations are not available to public social-listening tools. A public post may mention an incorrect AI answer, but do not assume that every issue or support question began in an AI chat.
Which channels carry the most developer signal?
Select channels based on product mentions, support needs and audience relevance. The examples below illustrate the kinds of discussion to inspect, rather than a universal coverage requirement.
| Channel | Scale signal | What you hear | Best use |
|---|---|---|---|
| GitHub issues and discussions | 180M+ developers reported in Octoverse 2025[2] | Bugs, feature requests, integration friction | Product feedback, support |
| Relevant subreddits vary by product category | Comparisons, frustrations, recommendations | Competitive intel, AI citation source | |
| Hacker News | One front page post drew ~18,000 daily site visitors in a 2021 case study[7] | Launch reactions, deep technical critique | Launch feedback, positioning |
| YouTube | Tutorials and creator channels vary by topic | Tutorials, reviews, walkthroughs | Education gaps, creator partners |
| Stack Overflow | New questions near 2009 levels in the cited analysis[3][1] | Long-tail how-to questions | Docs gaps |
| X and LinkedIn | Varies by audience | Hot takes, launches, hiring signals | Sentiment, buying signals |
GitHub issues and discussions
GitHub passed 180 million developers in 2025, adding more than 36 million in a year.[2] Merged pull requests averaged 43.2 million a month, up 23%. More than 1.1 million public repositories now import a large language model (LLM) software development kit (SDK), up 178%.
Monitor your own repositories and relevant external projects. Framework issues that name your SDK or competitor migration threads can reveal integration problems that never reach your support queue.
Reddit is both a channel and a citation source
Reddit licensed its data to Google in February 2024 and later to OpenAI.[4] Profound found Reddit was the most-cited domain in Google AI Overviews and Perplexity from August 2024 to June 2025.[4] A Reddit thread about your product is a conversation you can join and a source an answer engine may quote.
Hacker News launch discussions
Referral volume from a Hacker News submission varies. In a case study updated in January 2021, Nick Lafferty reported almost 18,000 daily website visitors after a front-page appearance, compared with a usual daily audience of about 50. That is an individual historical result, not a launch-traffic forecast.[7] Hacker News guidelines prohibit generated and AI-edited comments.[8] Write any response yourself.
Review source coverage and participation rules together. A channel that is useful for monitoring may restrict promotional replies or AI-assisted comments.
What should you listen for?
These categories help assign posts to the right team. Review exclusions periodically so a monitoring rule does not hide relevant feedback.
- Questions. Someone cannot get your product to work. Answer with a working example.
- Friction. Someone hit a bug, a limit or a confusing doc. Route it to the team that owns it.
- Buying signals. Someone asks for alternatives, compares pricing or evaluates options. Route it to sales with context.
- Praise. Someone shipped with your product and said so. Thank them and ask to feature it.
- Competitor mentions. Someone discusses a competitor's outage, pricing change or missing feature. Learn from it.
Developers distrust AI accuracy. In the 2025 survey, 46% distrusted AI tool output against 33% who trusted it.[3] When a public post explicitly describes an incorrect AI answer, investigate the claim against your documentation. The survey does not measure how frequently such errors become public complaints.
How do you set up developer listening?
Start with three sources, sort every post into a track, and route each track to an owner.
- Pick your keywords. Product name, common misspellings, SDK package names, command-line interface (CLI) commands, error strings and competitor names.
- Cover three channels first. GitHub, Reddit and Hacker News. Add X, LinkedIn, YouTube and community forums next.
- Sort into tracks. Questions, friction, buying signals, praise and muted. Mute with a reason so you can audit what you ignore.
- Route with context. Friction to a ticket. Buying signals to sales. Questions to whoever owns docs and replies.
- Reply as a person. Draft from your docs, have a human review it, disclose who you are, and post from the original thread. For the rules on each channel, read Replying to developers: community rules and disclosure.
- Compare monitored sources with AI citations. Check which community domains actually appear in answers for your prompt sample. For how those two halves connect, read Answer engine optimization and developer social listening.
For the basics of the discipline, read What developer social listening is.
Sources
- Stack Overflow is almost dead. The Pragmatic Engineer.
- Octoverse: A new developer joins GitHub every second as AI leads TypeScript to #1. GitHub Blog.
- 2025 Stack Overflow Developer Survey. Stack Overflow.
- Reddit Is Winning the AI Game. Columbia Journalism Review.
- AI has basically killed Stack Overflow. Futurism, citing DevClass and Stack Exchange Data Explorer.
- An update on our mental health-related work. OpenAI, February 27, 2026; reports more than 900 million weekly users.
- What happens when you're on the front page of Hacker News. Nick Lafferty.
- Hacker News Guidelines. Y Combinator.
Last verified October 22, 2026.