Knowledge data 2026 | Stack Overflow Developer Survey

5.1 Context Challenges

Time spent finding information

75% spend 5 or more hours/week searching for information; 11% spend 20+ hours

Question

In a typical week, how much time do you spend looking for information you need to do your work? Please provide a whole number representing hours per week, rounding up to the nearest whole number if needed.

	Respondents	Percent
1	316	2.6%
2	799	6.7%
3	747	6.2%
4	1,171	9.8%
5	1,935	16.1%
6	643	5.4%
7	216	1.8%
8	1,255	10.5%
9	47	0.4%
10	2,119	17.6%
11	5	0.0%
12	266	2.2%
13	10	0.1%
14	46	0.4%
15	832	6.9%
16	203	1.7%
17	6	0.0%
18	43	0.4%
19	1	0.0%
20	848	7.1%
21	10	0.1%
22	4	0.0%
23	5	0.0%
24	45	0.4%
25	119	1.0%
26	3	0.0%
27	1	0.0%
28	11	0.1%
29	1	0.0%
30	125	1.0%
32	11	0.1%
33	1	0.0%
35	16	0.1%
36	1	0.0%
38	7	0.1%
39	3	0.0%
40	46	0.4%
42	6	0.0%
43	1	0.0%
45	4	0.0%
48	6	0.0%
50+	72	0.6%

View and share

Important project context

Project goals (85%) and code/technical docs (73%) are the most needed context types.

Question

When working on a project or task, which types of context are most important for you to have? Select up to 5.

	Respondents	Percent
Project goals or requirements	10,806	85.2%
Code, repositories, or technical documentation	9,297	73.3%
Prior decisions and rationale	5,580	44.0%
Related tickets, tasks, or issues	4,679	36.9%
Business goals or product strategy	4,189	33.0%
Pull requests, commit history, or release notes	3,332	26.3%
Customer or user feedback	3,296	26.0%
Team norms, preferences, or ways of working	3,183	25.1%
Compliance, security, or policy requirements	2,498	19.7%
Current status or ownership	2,490	19.6%
Past discussions in chat, email, or meetings	2,000	15.8%
Data, dashboards, or reports	932	7.4%
I do not usually need additional context	266	2.1%
Other	79	0.6%

View and share

Barriers to finding context

Incomplete info (63%), knowledge in people's heads (61%), and outdated info (56%) are top barriers.

Question

Which of the following make it difficult to get the context you need for work? Select all that apply.

	Respondents	Percent
Information is incomplete	7,825	63.2%
Important context lives in people’s heads	7,541	60.9%
Information is outdated	6,989	56.5%
Information is spread across too many tools	6,603	53.4%
I do not know where to look	3,816	30.8%
Important decisions are hard to find later	3,226	26.1%
Different teams have different versions of the truth	3,144	25.4%
Search does not return useful results	2,881	23.3%
AI tools do not have enough context	2,471	20.0%
I do not know which source is authoritative	2,272	18.4%
AI tools use the wrong context	2,075	16.8%
Access or permissions prevent me from seeing what I need	1,961	15.8%
I usually have the context I need	1,193	9.6%
I intentionally limit context because of privacy or security concerns	753	6.1%
Not applicable	291	2.4%
Other	193	1.6%

View and share

Managing work context

Context management is highly manual — keeping separate notes (54%) and starting fresh AI sessions (45%) dominate.

Question

Which of the following do you do to manage or prepare context for your work? Select all that apply.

	Respondents	Percent
Keep notes or prompts in a separate document	6,495	53.9%
Maintain personal documentation or notes	6,482	53.8%
Start new AI chats or sessions to reset context	5,415	45.0%
Maintain team documentation or wikis	5,178	43.0%
Ask teammates where to find information	4,572	38.0%
Manually select files or documents for AI tools	4,174	34.7%
Search across multiple tools	3,993	33.2%
Copy and paste information between tools	3,939	32.7%
Reuse saved prompts or templates	1,892	15.7%
Limit the context I share because of privacy/security concerns	1,198	10.0%
I do not do anything specific to manage context	829	6.9%
Other	199	1.7%

View and share

Discovering context too late

79% discover important context only after starting a task, highlighting systemic context gaps.

Question

How often do you discover important work context only after you have already started or completed a task?

	Respondents	Percent
Very often	1,539	12.1%
Often	2,964	23.4%
Sometimes	5,465	43.1%
Rarely	2,136	16.8%
Never	130	1.0%
Not sure	253	2.0%
Not applicable	200	1.6%

View and share

5.2 Knowledge sources and trust

Sources for work-related answers

Coworkers (72%) and code repos (63%) are the go-to sources — internal beats external.

Question

When you need to answer a work-related question, which sources do you use most often? Select up to 5.

	Respondents	Percent
Coworkers or teammates	10,405	71.9%
Code repositories or code comments	9,065	62.7%
Internal documentation (i.e. Design, product, or planning docs), wiki, or knowledge base	8,687	60.1%
Internal chat tools, such as Slack or Teams	6,572	45.4%
Official docs	4,415	30.5%
Product, vendor, or API documentation	4,221	29.2%
Stack Overflow	3,472	24.0%
Manager or leadership	3,272	22.6%
Project management or ticketing tools	2,916	20.2%
Past decisions or architecture records	2,248	15.5%
Meeting notes or transcripts	1,536	10.6%
Other	1,008	7.0%
Vendor forum	411	2.8%
Customer, support, or sales information	362	2.5%
Not sure	196	1.4%
None of these	122	0.8%

View and share

Importance of information accuracy

Accuracy is the expectation for official docs, API documentation, and code repos; chat tools and Stack Overflow face lower accuracy expectations.

Question

For each source, how important is it that information you use to operate is accurate?

	Important	Least important	Not very important	Somewhat important	Very important
Code repositories or code comments	27.6%	0.1%	0.7%	6.3%	65.2%
Coworkers or teammates	34.1%	0.1%	0.7%	8.4%	56.7%
Customer, support, or sales information	28.9%	0.9%	2.6%	17.1%	50.6%
Internal chat tools, such as Slack or Teams	43.8%	0.3%	2.3%	18.6%	35.0%
Internal documentation (i.e. Design, product, or planning docs), wiki, or knowledge base	35.9%	0.2%	1.2%	11.9%	50.9%
Manager or leadership	30.8%	0.1%	0.8%	8.6%	59.6%
Meeting notes or transcripts	37.0%	0.9%	3.8%	23.8%	34.5%
None of these	14.3%	9.2%	5.1%	16.3%	55.1%
Not sure	17.4%	9.4%	4.7%	25.5%	43.0%
Official docs	21.7%	0.1%	0.7%	5.0%	72.5%
Past decisions or architecture records	37.4%	0.2%	3.7%	20.3%	38.4%
Product, vendor, or API documentation	25.3%	0.0%	0.8%	5.5%	68.4%
Project management or ticketing tools	41.2%	0.4%	2.0%	15.2%	41.2%
Stack Overflow	40.9%	0.7%	3.4%	25.0%	30.0%
Vendor forum	36.0%	0.5%	3.2%	22.1%	38.2%

View and share

Importance of current information

Recency/timestamps matter most for official docs and product/vendor docs; people tolerate stale human knowledge.

Question

For each source, how important is it that information you use to operate is recent or clearly timestamped?

	Important	Least important	Not very important	Somewhat important	Very important
Code repositories or code comments	27.1%	1.8%	6.0%	14.6%	50.5%
Coworkers or teammates	30.3%	4.4%	11.7%	23.4%	30.1%
Customer, support, or sales information	31.1%	1.2%	4.4%	14.4%	49.0%
Internal chat tools, such as Slack or Teams	34.1%	1.7%	7.3%	21.2%	35.8%
Internal documentation (i.e. Design, product, or planning docs), wiki, or knowledge base	33.2%	1.0%	5.2%	18.6%	41.9%
Manager or leadership	31.4%	3.3%	9.1%	20.2%	36.1%
Meeting notes or transcripts	33.1%	1.4%	6.1%	16.3%	43.1%
None of these	11.8%	11.8%	9.7%	30.1%	36.6%
Not sure	25.0%	10.4%	9.0%	27.8%	27.8%
Official docs	29.8%	1.0%	3.9%	13.2%	52.0%
Past decisions or architecture records	30.8%	1.9%	8.3%	23.6%	35.4%
Product, vendor, or API documentation	31.3%	0.9%	3.6%	12.9%	51.2%
Project management or ticketing tools	33.6%	1.0%	4.2%	16.8%	44.5%
Stack Overflow	33.3%	2.3%	8.5%	22.2%	33.7%
Vendor forum	29.5%	1.0%	6.5%	20.2%	42.8%

View and share

Importance of source traceability

Source traceability is most critical for code repos and customer support docs; coworker knowledge is trusted by reputation.

Question

For each source you selected, how important is it that you can quickly check where the information came from before relying on it for work? By “check where the information came from,” we mean being able to quickly open or review the original source, context, citation, document, message, ticket, dashboard, or person behind the information.

	Important	Least important	Not very important	Somewhat important	Very important
Code repositories or code comments	29.9%	1.7%	5.8%	14.4%	48.1%
Coworkers or teammates	29.4%	4.4%	14.9%	24.6%	26.8%
Customer, support, or sales information	31.4%	3.0%	6.7%	14.6%	44.2%
Internal chat tools, such as Slack or Teams	33.8%	2.3%	9.5%	22.9%	31.5%
Internal documentation (i.e. Design, product, or planning docs), wiki, or knowledge base	32.5%	1.8%	9.2%	19.7%	36.8%
Manager or leadership	28.4%	4.5%	14.0%	22.3%	30.8%
Meeting notes or transcripts	34.6%	3.1%	7.5%	20.5%	34.4%
None of these	13.5%	10.1%	12.4%	28.1%	36.0%
Not sure	30.0%	12.9%	4.3%	27.9%	25.0%
Official docs	26.8%	4.8%	10.0%	14.7%	43.7%
Past decisions or architecture records	33.7%	1.8%	6.2%	21.8%	36.5%
Product, vendor, or API documentation	27.7%	3.7%	9.1%	16.2%	43.4%
Project management or ticketing tools	36.7%	1.8%	5.4%	17.5%	38.5%
Stack Overflow	35.5%	2.4%	8.7%	20.7%	32.7%
Vendor forum	32.9%	3.1%	6.2%	20.5%	37.3%

View and share

Source attribution for AI answers

79% consider source attribution important or very important for trusting AI answers.

Question

How important is source attribution when deciding whether to trust an AI-generated technical answer?

	Respondents	Percent
Very important	7,269	52.1%
Important	3,701	26.5%
Somewhat important	1,956	14.0%
Not very important	753	5.4%
Least important	266	1.9%

View and share

Signals of trustworthy information

Recency (72.6%) and authority (60%) beat attribution and peer validation as trust signals.

Question

When deciding whether to rely on work-related information, which signals matter most to you? Select up to 5.

	Respondents	Percent
It is recent or clearly up to date	9,435	72.6%
It comes from an official or trusted source	7,799	60.0%
It is consistent with other trusted information	5,967	45.9%
It is clear, complete, and easy to understand	5,540	42.6%
I know who created or owns it	4,946	38.1%
It includes sources, citations, links, or supporting evidence	4,887	37.6%
It is marked final, approved, accepted, or official	3,960	30.5%
Other people on my team use or endorse it	3,212	24.7%
It was reviewed or validated by a subject matter expert	3,161	24.3%
It is not marked draft, outdated, deprecated, or superseded	2,281	17.6%
It respects access, privacy, or permission rules	1,169	9.0%
Not sure	411	3.2%
None of these	115	0.9%

View and share

5.3 AI adoption and cost

Tasks delegated to AI agents

Code tasks dominate AI use (73% write code, 62% debug); critical ops like deploying (10%) are rarely delegated.

Question

For each job function, which tasks have you delegated to an AI tool or agent in the last 30 days?

	Respondents	Percent
Writing, editing, or generating code	10,027	72.9%
Debugging or troubleshooting	8,552	62.2%
Writing or maintaining tests	6,960	50.6%
Writing or updating documentation	6,642	48.3%
Reviewing code, pull requests, or technical changes	6,182	44.9%
Making technical design or architecture decisions	3,629	26.4%
Investigating an incident, outage, or production issue	3,560	25.9%
Planning, scoping, or estimating work	3,461	25.2%
Answering questions from teammates, customers, users, or stakeholders	2,941	21.4%
Analyzing data, reports, or business metrics	2,773	20.2%
Onboarding to a new project, tool, system, or codebase	2,706	19.7%
Finding or fixing security, privacy, or compliance issues	2,676	19.5%
Changing production code, systems, or infrastructure	2,601	18.9%
Maintaining dependencies, packages, or development environments	2,547	18.5%
None of the above	2,255	16.4%
Monitoring systems, logs, metrics, or alerts	1,864	13.6%
Deploying or releasing software	1,346	9.8%
Coordinating work across teams or functions	774	5.6%

View and share

Control over information used by AI

64% actively curate what info AI uses; 17% have little or no visibility into AI inputs.

Question

When using AI tools or agents for work, how much control do you usually have over the work information they use?

	Respondents	Percent
I decide exactly what information to give the AI tool	5,928	44.8%
I choose from available sources, files, or systems	2,549	19.3%
I do not use AI tools or agents for work	1,751	13.2%
The AI tool chooses the information automatically	1,374	10.4%
I am not sure what information the AI tool uses	830	6.3%
Someone else manages what information the AI tool can access	811	6.1%

View and share

Factors in AI adoption

Output quality, security, and price drive adoption — organizations are pragmatic and risk-averse.

Question

When you or your organization decides whether to adopt, continue using, or expand use of an AI tool or model for your work, which factors matter most? Select up to 3.

	Respondents	Percent
The tool produces useful and accurate results	3,520	30.7%
The tool meets security, privacy requirements	3,290	28.7%
The price is acceptable for the value it provides	3,090	27.0%
The tool meets legal or compliance requirements	2,557	22.3%
Not sure	2,522	22.0%
Your company policy allows for usage	1,844	16.1%
Employees or teams are asking to use it	1,451	12.7%
There is evidence that the tool improves business outcomes	1,446	12.6%
The tool reduces time spent correcting, retrying, or reworking outputs	1,401	12.2%
The tool is worth paying more for because of better results	1,137	9.9%
Usage and spending are easy to monitor or limit	1,089	9.5%
The tool works well with how people already do their jobs	1,055	9.2%
The tool connects with systems people already use	767	6.7%
The organization can switch tools or models without major disruption	665	5.8%
Human oversight is manageable	663	5.8%
Other	368	3.2%
Review burden is low	320	2.8%
The tool has observability already integrated or built in	213	1.9%

View and share

Organizational AI usage practices

38% of orgs set AI spending limits; 18% have no formal governance.

Question

Which, if any, of the following has your organization done related to AI tool or model usage? Select all that apply.

	Respondents	Percent
Set limits on AI tool or model usage or spending	4,367	38.1%
Reviewed internal usage or spending data for AI tools or models	3,240	28.3%
Approved different AI tools or models for different types of work	3,122	27.2%
Set limits on who can use certain AI tools or models	2,901	25.3%
Increased spending on AI tools or models because results justified it	2,366	20.6%
Not sure	2,043	17.8%
None of the above	2,030	17.7%
Compared buying an AI tool with building something internally	1,290	11.2%
Stopped using or reduced spending on other software tools while increasing investment in AI tools or models	992	8.7%
Negotiated shorter commitments or the ability to switch AI tools or models over time	364	3.2%

View and share

AI cost and usage limits

33% self-monitor AI usage; cost pressure changes model selection and task selectivity.

Question

Has cost or usage limits ever changed how you personally use AI tools at work?

	Respondents	Percent
Yes, I personally monitor my own use	337	32.6%
Yes, I choose cheaper models when possible	261	25.3%
Yes, usage is metered to manage costs	246	23.8%
No, cost/limits have not affected my usage	231	22.4%
I have not encountered cost or usage limits	213	20.6%
Yes, I am more selective about tasks	171	16.6%
My company/team has increased budget for AI	132	12.8%
Not sure	117	11.3%
Yes, my company/team has switched models/vendors that are more affordable	71	6.9%
Yes, I choose faster models even when they cost more	51	4.9%

View and share

Cost-conscious prompting

75% changed prompts for quality, 38% for cost — quality optimization outpaces cost-cutting 2:1.

Question

Have you changed how you write prompts or structure requests to reduce cost or improve output quality?

	Respondents	Percent
Yes, to improve output quality	422	75.4%
Yes, to reduce cost	215	38.4%
No	88	15.7%

View and share

See the Knowledge highlights

Get the big picture

Knowledge Data

5.1 Context Challenges

Time spent finding information

Question

Important project context

Question

Barriers to finding context

Question

Managing work context

Question

Discovering context too late

Question

5.2 Knowledge sources and trust

Sources for work-related answers

Question

Importance of information accuracy

Question

Importance of current information

Question

Importance of source traceability

Question

Source attribution for AI answers

Question

Signals of trustworthy information

Question

5.3 AI adoption and cost

Tasks delegated to AI agents

Question

Control over information used by AI

Question

Factors in AI adoption

Question

Organizational AI usage practices

Question

AI cost and usage limits

Question

Cost-conscious prompting

Question

What’s next?