Corporate Entrepreneurship in 8 Steps
An enterprise-friendly framework for rapid experiments
Enterprises need a platform for running rapid experiments that’s compatible with existing talent and operations.
1. Build the team.
Assemble experimentation teams equipped to deploy tests autonomously.
Enterprise experimentation is about getting a strong team of people to effectively collaborate on solving a wide variety of challenges over an extended period of time. Getting this right from the beginning is foundational to success. Good ideas are secondary. Winning organizations establish a core team to make key decisions, build project plans, manage day-to-day execution, and maintain relationships with an extended resource network.
Decision-maker: A product (or project) manager mandated to make decisions and connect with executives for support as necessary
Connectors: Networkers who provide access to key partners, cut lead times with social capital and smooth miscommunications. They have wide organizational networks and long company tenure
Executors: Do-ers who have the skills to mock up a financial models, create user scenarios, and deploy marketing to customers
Extended Resource Network
Sponsors: Allies to pool resources, build political support and justify investments with shared use cases. They have wide organizational networks and long company tenure.
Partners: Other working groups in the organization that provide access to key assets, like Product, Dev, Marketing, Sales, FP&A, Legal, etc.
Vendors: 3rd parties that fill skill or production gaps; can go around organizational bottlenecks.
- Choose team members passionate about solving this problem
- Build the smallest possible teams that are still able to manage and deploy tests on their own
- Pick all-rounders over specialists
- Prioritize critical soft skills like team fit and collaboration when selecting members
- Attract team members with perks like non-traditional operational rules or a fast track for promotion
- Hire outside the team only for highly specialized skills, like on-demand development services, to streamline direction
2. Set the culture.
Grant permission for your team to be scrappy, work outside the system, and trade personal favors.
The primary goal of experimentation is sustainable speed, but it’s hard to move fast. Organizations are built to resist change and discourage this type of behavior. Therefore, managers and executives need to create space for their teams of entrepreneurs who are always on the lookout for creative shortcuts and non-traditional solutions. These teams can’t always play by the rules
Celebrate hacks and workarounds.
Remind your team that resourcefulness and scrappiness are a virtue by celebrating creative, efficient, and non-traditional solutions. Many times these opportunities exist outside the scope of the system.
Ex: Prior testing indicates that users strongly prefer videochat to VoIP for this service, but there’s a legal hold up. You test a rough prototype of videochat the next day using an old personal email address as a contact address.
Recognize agility with flexibility.
Agile teams are more resilient than agile individuals. Your team will encounter many difficult situations for the first time. Encourage and authorize them to make make the tradeoffs required to keep the ball moving. Recognize that this experience can be an emotional grind, and help keep their spirits up through the ambiguity. You’ll help them to avoid emotional burnout.
Ex: The core team needs to revise its dummy landing page for another round of testing in 1 week, but web dev says it’ll take 6 weeks. The team quickly enrolls a trusted pre-approved vendor instead of doing it on their own.
Make experimentation teams sought after experts.
Teams can’t experiment in a vacuum. Set up your team to act as an integral resource to your business partners to reap rewards that come with that role. Encourage them to stay accessible, and helpful to others. Operational exceptions tend to be made for people that give back.
Ex: An extended team member liked the format of your kickoff session and needs help building a similar agenda. You send templates, and connect them to the right folks in the team for last minute help with planning.
- Incentivize action with recognition for movements and learnings, not just successes
- Grant the team ownership of project decisions by authorizing them to click send on their own
- Establish norms that make team members feel safe to share ideas — even contrarian ones
- Grant additional freedom in their work schedule to reduce the potential for member burnout that can arise from fast-paced work
3. Hook up the collaboration.
Install an operating system that’s capable of supporting hourly test iterations.
Modern workflows require new capabilities. The iterative process of experimentation requires an operating system that can support decentralized collaboration. The system should function as a virtual project room, since it’s unlikely that your team will be camped out together all day. Legacy tools are often unable to support these new requirements.
Task-Tracking: Basecamp, Trello replace email, post-its, to-do board
Real-Time Communication: Slack, Google for Work replace text messages, emails on Outlook, conference calls
Content Collaboration: Quip, Dropbox replace static word/ppt files, formal status updates, project room drive-bys
Design & Production: UXPin and Adobe CC replace powerpoint and paper and pen
Analysis: Domo, Google Analytics replace excel and data requests
- One size doesn’t fit all — find what works for your team, but encourage them to experiment with new solutions
- Secure blanket IT approval for experiment team tools
- Use existing privacy and security documentation to push tool use through legal channels
- Look for teams that already secured approval for useful tools and use those tools too
- Use resources like ProductHunt.com and Medium for inspiration and reviews on new products
4. Select the metrics.
Measure your experimentation process separately from the product you’re building.
Experimentation is an operational framework that can be applied to all kinds of innovation challenges within your organization. It’s important to make it a scalable practice. Experimentation KPIs can help you measure how well your team and company are functioning. These system metrics should be evaluated separately from the innovation currently being run through it.
Example process metrics:
Average test cycle time in days:
Number of VP meetings required per test:
Person hours required per test:
Example product metrics (based on key milestones):
Percentage of variables solved:
Time since last check-in:
Total budget used:
- Look for multiple, small improvements that each improve efficiency 10%
- Create a regular structure for weekly, cross-team reflection and feedback
- Establish a regular testing rhythm to set expectations as to when tests take place
5. Segment the problem.
Test the components of a business separately, but revise them at the same time.
Experimentation is much more effective when teams break big problems down into discrete solution components. Use this strategy to make big projects more manageable. Product, operations, marketing, and finance are common learning areas. Revise your offering once you have enough information across all areas.
Product: New moms prefer healthy to low fat
Operations: Stores can order new signage within 2 days
Marketing: We can reach new moms through our existing in-store channel
Finance: New moms are willing to pay $25, a 20% raise in price
- Deconstruct big problems into smaller, independently solvable puzzles and solve them
- Isolate variables so that testing yields conclusive data
- Look to test modular components that are capable of standing alone, that integrate into the final product
6. Prioritize the unknowns.
Rank business assumptions that are most important and uncertain.
Rapid idea validation. Once you have broken your problem down into learning areas, you should identify all of your assumptions and rank them according to importance and uncertainty. Since resources are almost always limited in an enterprise, successful experimentation requires a good understanding of tradeoffs and prioritization to decide what tests are worth running.
- Always prove customer demand first — nothing else matters if people don’t want what you are building
- Build a learning plan to design experiments that get you the information you need in a logical order
- Set clear objectives, learning goals, and success metrics before each test
- Always keep a record of your decision-making process and the logic of your tradeoffs so you can always go back if you encounter problems down the road
7. Develop a plan.
Build an experimentation roadmap to validate your ideas with multiple small tests — not just one big one.
A winning experimentation approach de-risks critical elements of the business at key development milestones. Good management achieves this by setting an action plan for getting to believable conclusions in the shortest amount of time. Winning requires designing small experiment variations, setting clear measurement goals, adhering to a challenging pace, and testing isolated variables. Small, controlled tests are the platform that powers the building of better products.
Strip out extra variables to make learning easier.
You should always aim for data that’s simple to interpret. This will prevent you from building an overly complex test that yields convoluted results. Testing a complex offering isn’t just harder on the execution side, it also makes learning from the results more difficult. Transform multivariable tests into simpler discrete tests.
Ex: Separate testing for pricing and value proposition into two A/B email campaigns to create a clean dataset of results.
Plan tests & measurements together.
Define your testable variables, your hypothesis, and the metrics for success together as a set before the test is launched.
Ex: or a click-thru rate test predefined outcomes that represent kill, change, or expand results so you eliminate confirmation bias.
Set a rapid, challenging cadence of test cycles.
Choose a testing cadence that inspires the team to do it faster. A larger number of cycles of simpler tests out-learns a couple of longer, more complex tests. Strive to be nimble.
Ex: Build two one-week tests rather than one quarter long test and have time to implement multiple rounds of learning.
- Look for ways to short-circuit your own hypotheses before investing in formal tests
- Quantity yields quality — Train the team’s ability by running many small experiments early
- Don’t get attached to specific solutions — document member’s “pet” ideas…then move on
- Launch related tests in parallel so you don’t rely in any one test and reduce the chance of confirmation bias
- Develop the habit of asking “what will I learn from this” to avoid aching over the wrong details
- Look for natural experiments that are already happening out in the world
- Adhere to Occam’s Razor — the simpler the solution, the better
8. Create reusable materials.
Assemble a toolbox of reusable assets and methods to accelerate future rounds of testing.
Experimenters aren’t just managers or strategists. They’re builders. During the course of launching a single experiment, the team will need to build new assets and create new processes, tools, and content. Document everything in detail and be sure to save documents templates. They’ll be useful next time around.
Planning: testing calendar, key variables list, decision framework, strategy sessions, assumptions list
Teamwork: stakeholder kickoff session, IRC or Slack channels, daily scrum, test learning debrief, post-mortems
Development: user stories, product requirements, touchpoint designs, one-line financial sketches, customer journey maps
Test Materials: test protocol, employee playbook, test notebook, test scorecards, interview recording equipment
Test Methods: survey, landing page, user interview, AdWords test, static non-functional mockups, working UI prototypes, co-creation session
Reporting: assessment dashboard, results spreadsheet, test summary posters, financial and web analytics models, investment guides
- Build templates at the same time you make assets so they’re ready to be re-used immediately
- Only invent new processes and tools when existing ones get you no closer than 80% of what the test requires
- Develop the habit of routinizing processes or tools to reduce time required to invent new ones
Get started experimenting today.
Innovation inside an enterprise organization is hard — much harder than inside a Silicon Valley startup. Lean startup offers a promising framework for how to bootstrap your new business when you can afford to be laser-focused on product/market fit. But what about when you also have to optimize for organizational fit, too? By following this 8 step agile development framework, you can enable agile experimentation to occur as an integral and parallel process to enterprise execution — by design, not by exception or heroics.
Thanks for reading! You’re now well on your way to bootstrapping your first experimentation platform. This list is by no means complete, and I will continue to refine my craft, but for now I’ve found it to be a good starting point. Hope it helps.
Pick a small problem. Talk to people. Do research. Ideate a few solutions. Select one. Get going. Now.
I’m a product management consultant at Jump Associates in San Mateo, CA. I have experience solving for the extra complexities of corporate entrepreneurship for some of the world’s most valuable brands.
This post was coauthored by Mike Smith, Eric Schreiber, Jay Newman, and Udaya Patnaik.