In part one of this series, we broke down how to set up our estimation process and tooling so we could easily measure how our assumptions stacked up against reality.
In part two, we’ll explore the next two steps in the Agency Profitability Flywheel
Installing Quantitative Feedback Loops (Meetings & Discussions)
Investing in Process Improvements
These steps are designed to empower our team to make data-driven process improvements and make the things we’re trying to predict more predictable.
Step Three: Installing Qualitative Feedback Loops (Meetings & Discussions)
The next step in our journey to creating accurate estimates is where the magic happens. Unfortunately, many agencies that start going down a data-driven path end up relying too heavily on the data for their decision making. The truth is, while data is certainly useful for decision making, it’s often not a complete solution.
The primary purpose of these data feedback loops is to inform and guide great conversations to uncover the nuances behind why and how gaps emerge between assumptions and reality.
The key to this step is determining a cadence that is sustainable for your organization, and teams, to review and discuss reports detailing the estimated and actual results of client work. During these discussions, the goal is to collaboratively work towards “operationalizing excellence” as we like to say. This is a fancy way to describe creating and improving processes to avoid things we want less of and repeat things we want more of.
The two most common ways of structuring these meetings are through project performance cadences and retrospectives.
The only real difference between these two flavors of meetings is that retrospectives typically occur at the end of a specific project, while project performance cadences are set meetings that occur on a fixed cadence (weekly, bi-weekly, monthly, etc.).
These conversations are meant to accomplish a few important things:
Review a report of estimates vs actuals for completed and/or ongoing projects
Ask questions about why things did/are or didn’t/aren’t going as planned
Surface insights and opportunities from the team
Build a backlog of process improvements based on that feedback
Course correct in-progress projects if the opportunity exists
Successfully facilitating these conversations comes down to three key things:
Focus on the process, not people
A common mistake in these meetings is focussing on the numbers in the report and using it to grill the team about going over budget. This is a one-way track to completely killing all buy-in and engagement in these meetings.Weaponizing the data should be avoided at all costs. In some cases, it may even make sense not to share the data with the team to avoid introducing bias into the conversation. Remember to use language like “us” -- avoid singling people out, and always refer to the process when discussing where things went wrong, and the reasons why things didn’t go according to plan.
Facilitate, don’t dictate
Another common mistake is coming to the conversation with biases or opinions on how to solve the problem. It’s extremely important to withhold those biases until the team gets stuck and requests input.The objective is to invite the team to engage in the process of ideating ways to work more efficiently and consistently, biasing that conversation can prevent team members from speaking up, or rob the group of the opportunity to surface ideas that are out-of-the-box.
Protect the time
Whatever format you chose, do everything possible to avoid letting client work push these meetings out of the calendar.Consistency is key to building momentum in this process and signaling to the team that the process and their input in that process is important and valuable.
Step Four: Defining Iteration Cycles (Process Improvements)
In the first two parts of the flywheel, we set up the infrastructure to place data points on our relationship line, to enable us to quickly understand how the inputs we collect during discovery with a client maps to effort.
This last step will enable us to improve the way we work, to ensure that we can deliver consistent quality in our work, and systematically place “rails” on the variability of those data points, which will make our relationship line more accurate and reliable over time.
This final step is meant is to take all the gold we’re extracting from our tools and team and turn it into repeatable processes and models that will make our agency more predictable over time.
This means defining iteration frameworks for two things:
The delivery processes
The estimation processes
Just like the process for our qualitative feedback loops, the key to this step is defining a cadence that can withstand the forces of client work and can happen consistently as things get busy.
Our favorite way to do this has generally been to treat process improvement cadences like product development sprints. We do this with bi-weekly meetings wherein the team prioritizes a backlog of process improvements surfaced from our bi-weekly performance meetings, assesses their complexity and effort, assigns ownership, and is afforded time to execute on those tasks.
Finally, we define when we will sit down to re-visit our estimation frameworks. During this time, we use the data we’re collecting from our time and cost-tracking tools to map those against our scoping metrics to develop the relationship line we discussed earlier in the post, and build out statistically reliable relationships between the inputs and outputs of our scoping and sales process.
Using this data, we can now adjust our estimation process to account for this new and more accurate information. This might mean building a library of projects grouped by type and sorted by size/scope, or developing graphs that map the line between our scoping metrics and effort based on the types of projects we do. It may mean updating drivers in a spreadsheet, or simply just jotting down notes to help us capture more of the context we’re extracting from our feedback cadences into our thought process during estimation.
By combining these data-driven relationships, our reporting and feedback cadences and the increased predictability of our ever-improving processes, we should see the gap between our estimates and our real-world result close over time.
Ultimately, this should lead to more consistent and higher profits. Meaning, better, more consistent work for clients, and better, more accurate resource plans plus fewer evenings and weekends working on overruns.
Final Thoughts
Whether it's protecting margins, making sure our team gets home to their family for dinner or predicting when we’ll need to hire our next staff member, when we can accurately predict what it will take to deliver a promised outcome to a client, everything in our agency gets easier. This fundamental skill is the bedrock of our operating system as a service business.
When done right, this process can eventually be managed by the team without much oversight from ownership or even executive management, effectively empowering the team to be more independent, while continuously improving the quality and accuracy of our agency’s work.
And that my friends, is what makes operations nerds like me shed tears of joy.
This is a guest article from a member of the Bureau of Digital community. We're always looking for good tips and lessons, if you're interested in contributing please email smith@bureauofdigital.com.