March ReportMonday, 09 April 2018

Report: “Operations”

So much for slow beginnings!

My first month overseeing Invisible’s operations has left me dizzy and exhausted, but with a sense of triumph. March allowed reality and the team to get a sense of one another and take each others’ measure. In the beginning of the month, reality had us on our heels as we struggled to get a grip on the day to day operations of the company — all the client requests and agents commingled, with new managers on the field trying to figure out how the technology worked. It was a glorious mess.

March Progress: Fast forward several weeks to the end of March and you have a different story. Grappling with reality and its seemingly intractable problems every day has brought us up the curve quickly by forcing us to get on the field. The byproduct of this is that we have a much more battle-hardened legion of managers with a firmer grip on their areas than we did at the beginning of March. Now we will build on that. As such, the potential for a shatteringly impactful April is there — and every ounce of our collective energy is going to making that potential a reality. More on this below.

March wins:

• Overall, we targeted and made a substantial improvement in timeliness — March’s improvement shot. Better tech, daily escalations for at-risk clients, better paranoia around the client experience and what we might be missing, RLMs combing look for issues before the client notices. Multiple initiatives to fix this existential threat in the immediate.
• Got our new management team on the field! Jay in Building, Steven in Operations, Melissa in Support, me across Operations and Support.
• Built and deployed a piece of technology that allows us to monitor in and out communications to the DAL (Project Eru).
• Operations: Expanded agent base (between 50% and 100%).
• Operations: Rolled out the “Client Commandments” to Operators to improve their client sensitivity and begin accountability push.
• Routing: Changed managers from Tes to Skyler. Established #rtt-help channel to quickly surface issues in Routing and prevent them from causing downstream chaos.
• Building: Burned down the Client Building backlog — step change in management of Building team by Jay.
• Building: Deployed Hubstaff on Building Team to enforce accountability and more accurate time-tracking. Piloted a number of other management ideas that Steven will be rolling out to the Operating Team this week
• Quality: Built out the Quality Team, established its daily machines, armed Jake and Denisa with Eru — allowing us to maintain a daily level of minimum timeliness and pinpoint where our systems are breaking down.
• Support: Getting Melissa in the seat was a huge win, effectively taking us from minimal customer support (Michelle someone getting it all done), to brining in Melissa and beginning to train an army of Relationship Managers.
• Support: Huge improvement in the speed and reliability of client onboarding. Onboarding becoming more frictionless.
• Training: Hired Tiffany as our head of training. She’s experienced in building and shaping curriculums and will allow us to begin identifying and molding high-potential agents as soon as they begin training.
• Sentrying: Set up a Sentry Team to improve the speed at which we process credentials approvals.

In the below I’m going to break down from a high level the picture for April and our strategic and tactical plans to deliver our April goals.

About Me: I joined Invisible Technologies at the beginning of March, directly after leaving my job as a Portfolio & Reporting Analyst at Bridgewater Associates, LP (a quirky hedge fund in Connecticut). I loved Bridgewater but I ultimately left because I wanted the space to shape my own enterprise. And because the world’s a lot bigger than Connecticut.

The two firms actually share a surprising number of characteristics. Radical transparency. Real-time feedback. A high bar for excellence. One of the reasons that I joined Invisible was because these are my values too — and when I left Bridgewater, I was afraid that I wouldn’t find another place that was similar enough for me to thrive.

So here I am as a result. I guess we’ll see if I learned anything.

The Goal & The Bottom Line

It’s always worth restating the goal, lest one forget it. My goal is to develop an Operations unit that will deliver a frictionless customer experience for every request.

From a timeliness, quality, and cost perspective what that looks like is a every client receiving acknowledgment of their request (or the work itself) within 3 hours, high quality engagement between the bot and the client (clear, low number of back-and-forthes, no grammar mistakes), high quality work (the literal thing(s) asked for are done correctly with no additional followup needed on either side), at a reasonable cost (one within expectations).

In April, the goal is A MINIMUM VIABLE PRODUCT!

To achieve this, we need to get a notch higher in timeliness (from a 5 to a 6), and to make meaningful progress in terms of Quality (from a 4/4.5 to a 6), paired with a significant jump in efficiency (3 to a 5).

The Problems

The highest level problem is that this doesn’t happen all of the time. My estimation is that timeliness and quality occur in unison roughly 50% of the time. Regardless of the literal number, the bottom line is that it doesn’t happen nearly enough — we’re well short of the bar on all these dimensions:

Timeliness: 5/10, with 7 being acceptable. Too many dropped balls and missed SLAs.

Quality: 4.5/10, broken in to 2 categories:

Quality of the engagement: 4/10. Interactions with clients too churny and confusing. Poor grammar in a limited number of cases. Bot not thinking at the goal level and cutting through it.

Quality of the work: 5/10. Work product is pretty good, but inconsistent between agents, shifts, and teams. Careless mistakes too frequent.

Efficiency: 3/10. Too much agent time spent on non-client work. Some agents slow and unreliable. Not being managed effectively.

The Diagnosis

My diagnosis is these problems are all solvable, but need to be attacked at multiple levels simultaneously.

1. The DAL as it currently exists is too complex. Too many pipes with too many connections, too many manual steps. Even a highly-trained agent with a high sense of ownership would make too many mistakes. Mistakes at the front of the funnel (say Routing, or fuzzy delegations from the client) crash through the DAL. It mostly doesn’t have the ability to self-correct when these issues are set in motion.
2. Visibility in to what our agents are working on (and what they’re not working on), and where they are in back-and-forthes with the client is bad. This allows dropped balls and free-riding by agents. This is a major contributor to the timeliness problem.
3. We as managers have not done a good job mapping out how the DAL works. The principles for operation at various stages are ill-defined and not widely shared. This contributes both to timeliness and quality issues. For example, agents not knowing that when they have a credentials access issue to post in rtr-credentials. Some agents flag to an RLM, some pull in their shift or team manager, and some do nothing. Few move quickly enough even if they follow the machine to get it solved. Most need to be reminded of the correct process and what good looks like.
4. The DAL makes it difficult to understand the context across various threads. This is because emails are difficult to read in Slack and threads with the client can be long. It can take significant time and effort simply to understand what’s going on. Additionally, because there can be multiple instances for what’s being asked for in a thread, it’s easy to miss someting. The end result is a garbled, confusing message to the client. This is a big issue for quality and efficiency.
5. Managers are not yet managing their teams well enough: establishing and enforcing accountability and bubbling problems to the surface where they can be dealt with. The first issue is a bigger deal right now than the second. I don’t believe that agents sufficiently feel the pain of the client experience when it’s below the bar and are not sufficiently incentivized. Everyone is just basically assumed to know what good looks like and checks aren’t in place to make sure that they’re actually producing these outcomes. Additionally, agents don’t escalate (ask for help) nearly enough. This leads to them making mistakes that they could easily be coached through beforehand.
6. A meta-level one: the management team is not solving problems as quickly as it needs to to make a revolutionary turn in the development of the product. I don’t believe that a new, relatively inexperienced, separated management team can manage a global, remote workforce with nascent technology, growing quickly. It just won’t work. The size and scope of the issues is too big to solve with our current approach of everyone apart.

The Solutions & The Progress — April: Get to Minimum Viable Product

1. DAL Complexity: There are a number of initiatives that Skyler has mapped out that will reduce complexity across the board. I have confidence that his “DAL 2.0” strategy, if implemented, will go a long way toward reducing a lot of the pain. He and Gunar (head of Product) started interacting on them in the second half of March at an accelerating rate. The immediate goal is to rip off at least one win with some piece of DAL 2.0 in the second week of April. By end of April my goal is to have it 75%-100% instituted. Key thrusts: positioning RLMs at the front of the DAL to metabolize bad or complex delegations, limiting the number of internal handoffs, and creating an instance channel for every thread. March progress on this was minimal, with the month devoted to gathering information to synthesize what the biggest issues are.
2. Visibility in to what our agents are working on: This is going to require tech engagement to solve. Skyler will take that ball and work it through with Gunar. In the meantime, my solution is better checks from Steven (head of the Operating Team), the Prime Managers, and increased diligence from agents to narrow the gap. It will reduce the surface area that tech needs to solve and get us some immediate outcomes in this space. I’m not banking on a tech solution in April — our improvement is going to have to come from refining our existing process at the management and operator level. I want the cavalry to arrive no later than May.
3. Mapping out the DAL: Each manager needs to thoroughly map out his or her area, playbook it with a set of decision-making principles, and push it to their team. From an overall perspective, Skyler and I need to make sure that there pieces fit and that where there are risky area (read: “joints” between teams that require manual operation) that those are well-designed, clear, and guardrailed. Otherwise our sub-assembly lines will not feed our master assembly line.
4. DAL Context: This will be solved through the “DAL 2.0” push — Product’s highest priority for April.
5. Agent Accountability: This is mostly a training and management push. Tiffany, our new head of Training, is going to make sure that accountability starts on Day 1 with Invisible. Steven, who manages the literal agents in Operations, is going to conduct daily checks with respect to timetracking and output, calling balls and strikes with agents and providing real-time coaching. Over the course of the month, I expect that agents will calibrate to a much better level of self-regulation.
6. Management Diaspora: Enter the Utah lock-in for the management team! We are assembling the entire Operations management team, with Melissa (head of Support) in Utah for a month to co-work, co-live, and co-kick ass. The goal is to make 3 months of progress in one and hone our team to be able to operate effectively together even when apart. It is our problem-solving secret weapon and I am confident that if done well we will achieve massive impact across the 5 problems I’ve described above.

The Implementation

As I write this, I’m stepping off the plane in Salt Lake City, Utah. In two hours’ time I will be at the Lock-In location (a beautiful home in Bountiful, Utah), setting it up for the rest of the Operations Team as they arrive today.

When I left Bridgewater, I told Ray that one of the main reasons was that I wanted the opportunity to shape and experiment. The chance to try out my hand against reality.

I’m about to find out a) whether that’s what I wanted and b) whether I’m any good at it. As such, Melissa, Michelle, Steven, Skyler, Kamron, Tiffany, Jay, and I intend to make April a stunning success for Invisible. Onwards!

Marshall

Vice President of Operations