Investigate before deciding
A team may need to know whether an external API supports a requirement, how hard old data will be to migrate, or which technical approach to choose. A spike sets aside a short period to investigate before the team commits.
The work might involve reading documentation, testing an integration, analysing data, or building a quick experiment. The result is evidence for a decision: findings, options, risks, and a recommendation.
Set a time limit
The team should state the question and time limit at the start. A spike should end with an answer or a clear account of what remains unknown, not quietly turn into implementation. Experimental code may not be ready for production; if the team keeps it, it must identify and finish the remaining work.