Power-Automate-Discovery-Sprint: What Two Weeks Really Deliver

Many mid-market companies buy an automation pilot without knowing reliably which process is worth automating and how much effort the implementation actually costs. A two-week Power-Automate-Discovery-Sprint is meant to close exactly this gap before licenses are ordered and development hours are booked.

What such a sprint actually delivers often differs from expectations. It is a structured inventory with reliable effort estimation, not a finished automation and not a success guarantee. This article makes clear what realistically emerges in two weeks and where the limits lie.

What does a two-week Power-Automate-Discovery-Sprint deliver concretely?

A Power-Automate-Discovery-Sprint delivers three fixed results within two weeks: a documented current state process assessment, a prioritized list of Quick Wins, and an effort estimation in person-days for the subsequent implementation.

The current state process assessment is created from interviews with the specialist departments and a look at the systems actually in use, not from what is stated in the organizational chart. This regularly shows that the lived process deviates from the official description, for example because individual employees manually track exceptions in Excel.

From this assessment, the Quick Win list is created: sub-steps that can be automated with reasonable effort, evaluated by benefit and implementation effort. Not every bottleneck found lands at the top of the list; some processes are too rare, too variable, or too closely tied to a single person to be worth doing in the short term.

How is the effort estimation created, and which license costs flow in?

The effort estimation in the sprint covers two levels: person-days for development and ongoing license costs, which vary greatly depending on the type of automation. Power Automate Premium costs 15 US dollars per user and month for digital process automation, while a Power Automate Process license for unattended RPA bots costs 150 US dollars per bot and month (Source: Microsoft, Power Automate Licensing FAQ).

This difference often determines whether a Quick Win pays off. A process affecting ten users with Premium license costs significantly less per month than a single unattended bot with Process license, even if the bot seems like the more elegant solution at first glance. For us, this calculation is firmly part of the effort estimation, not just in the implementation phase.

For the prioritization of processes, we orient ourselves to Microsoft's own approach for Automation Center of Excellence teams, which consistently steers process selection via expected return on investment rather than gut feeling (Source: Microsoft, Holistic Enterprise Automation Techniques (HEAT)).

What does the sprint deliberately not deliver?

The Discovery Sprint delivers neither a finished implementation nor a guarantee of return on investment without the subsequent implementation phase. Both would not be done properly in two weeks either, from a craftsman's perspective: development, testing with real data, and error handling require time that is reserved for analysis in the sprint.

This boundary is not a sales argument but a real source of error if ignored. The consulting firm EY observed over years with customers that 30 to 50 percent of first RPA projects fail, often because companies automate a process that is more complex or variable than originally assumed (Source: EY, Get ready for robots).

The current state process assessment in the sprint is meant to reduce exactly this risk by making exceptions and special cases visible before development rather than only during testing. The risk cannot be eliminated entirely this way, because whether an automation holds up in everyday operation is only finally shown after the go-live of the implementation phase.

An honest result is also part of the sprint: If the assessment shows that a process is currently not worth it, for example because it runs too rarely or is currently being fundamentally rebuilt, we say so. More about our approach in the Power-Automate-Consulting can be found in the associated service description.

What does a sprint look like in practice? A process flow diagram from a NordFlux project

At a trade business with around 40 employees, the sprint ran in two clearly separated weeks. In week one, the first day had kickoff and system access, followed by three days of interviews with accounting, planning, and order intake, accompanied by a process assessment of invoice receipt and order processing directly at the systems in use.

In week two, the assessed processes were prioritized against each other, effort estimations in person-days for each Quick Win were created, and on the last day presented in a results meeting with management. The result was a list of four Quick Wins, led by the automated Invoice receipt, with effort estimation, recommended license form, and a clear sequence for implementation, but without a single line of productive flow.

The decision whether and in what order to implement remained deliberately with the customer. This is the core of the promises from our positioning: The customer retains control over scope, pace, and budget of the automation, and the sprint provides the factual basis for it.

Frequently Asked Questions

Does the Discovery Sprint replace the later implementation?

No. The sprint provides the analysis and prioritization; development, testing with real data, and productive rollout follow as a separate, separately commissioned implementation phase.

Do we already need Power-Automate licenses to start the sprint?

No. The sprint runs largely tool-independently via interviews and system review. Which license form is sensible in the end, Premium per user or Process per bot, is itself part of the result.

What happens if in the end no process is worth it?

Then that is a valid result. An honest no before the investment is cheaper than a failed automation project afterward, especially given the high failure rate of inadequately prepared RPA projects.

How quickly can implementation start after the sprint?

After the results meeting, a prioritized offer with effort estimation is available. The start of implementation then depends on the customer's decision; technically, a direct follow-up is possible.

Is there a guarantee for a specific ROI from the sprint?

No. The sprint provides an estimate based on the current state assessment, not a promise. The actual benefit depends on clean implementation, data quality, and process stability after go-live.