1. Where is test strategy with an
agile team?
In search of long-term ideas that guide test design
Maaret Pyhäjärvi
Twitter: @maaretp
2.
3. Atypical Agile Team
• Product Development
• Team of 9 (1 testing specialist) + PO
• Release in production when done (includes
tested)
– Continuous Deployment with little test automation
– Kanban with conversational WIP limits
– NoEstimates, focus on identifying slices of value
– NoProjects
– Testing = Checking + Exploring
4. Ideas that Guide All Testing
• Knowing the product (by asking around)
– Purpose of existence
– Functionality, Performance
– Browsers, .NET MVC
– Choosing the right features into development pipeline
(lean startup)
• Delivering professionally
– Done means done – value in use delivered
– Production monitoring is an option for getting information
– Reporting on product (Lead time; Net Promoter Score) not
on testing
• Actionable information first
– Awareness of reporting time
Strategy should be more
specific to the product at
hand?
5. Supporting Documents
• Quality Target
– “Awareness thing”
– Outlining rough types of testing with split to
roles
• System Testing Support List
– “Feature Breakdown”
– Connections in the system between the
features
• Elisabeth Hendrickson’s cheat sheet
6. What Really Happens
• Split Jira item into smallest possible testable
chunk; talk about value and design (dev-test-po)
• Implement & Pair-test to introduce to testing
specialist
• Explore sympathetically and extending as long as
needed
– Feature in isolation, split to browsers starting from
most likely to break
– Feature in combination with other features
– …
– Monitoring in production
• Get better (scope of test automation; refactoring;
pairing and group work; individuals’ skills)
7. Key Observations from the
Experience
• CONTINUOUS DEPLOYMENT IS TESTING GAME-CHANGER
– Continuous deployment allows for applying indefinite time on selected tactics, so the
prioritizing of the next tactic to use is on selecting the next move that reveals
information that is immediately useful
• TACTICS OVER STRATEGY FOR TESTING
– There’s so much commonality on things to do in testing of different projects that
strategic ideas seem almost invisible, and the focus on is applying the right tactics to
reveal the right information more efficiently and timely.
• TEMPORARY AND TACTICAL AIDS FOR AGREEING AND
REMEMBERING
– Two documents have seemed relevant in the lifecycle of this product: a quality target
agreeing who does what kinds of testing activities and a feature breakdown to remind
of connections in the system between the features
• STRATEGY IS ENABLING TESTING IN PRODUCT DEVELOPMENT
– Most of strategic effort goes into software development in general: shortening the
feedback cycle to enable flexibility and relevance of testing and building in testing from
the inception of a feature idea, driving through feature splitting to the ideas of tactics
we’d apply on a particular feature.
Editor's Notes
Specific context – while I personally see that different past contexts have had very different choices on writing a strategy document to communicate high-level decisions on approach on what to focus on
Strategy: ideas guiding test design, specific, long-term
Sw dev as collaborative game, order of moves / tactics does not mostly have impact on their availability
Strategy for one is a tactic for another
The aspects of testing concerned with evaluation of where we are now and setting of goals and long-term plans for future play.
Compare to tactics which is move by move, whereas strategy is long-term
In chess, one needs to become ‘master’ before strategy knowledge becomes determining factor in game outcome over tactics
Many chess coaches emphasize study of tactics as the most efficient way to improve one’s results
Strategic goals are achieved by the means of tactics while tactical opportunities are based on the previous strategy of play
“Strategy requires thought; tactics require observation” – Max Euwe
Need to communicate ideas that guide test design within the team with non-testers
Continuous learning about the growing product
In retrospect things appear clearer than they are when all things are uncertain