Over-Prescribing Scrum

Dave Gullo

I just finished reading my brother Daniel Gullo‘s book “Real World Agility” (http://www.realworldagility.com) and I wanted to share some of my thoughts about Scrum and Agile and their application to early stage companies.

The best analogy I had used to describe Scrum is that it serves as a “Business Process Interface” for software development. A hacker from one company can drop into another if they follow these principles, and each company does not have to build their development process from scratch. More officially, it is an iterative and incremental framework for managing product development.

Never miss fantastic posts such as this one.

Unfortunately, the term “Interface” has a loaded meaning for software engineers because an Interface is a contract, a promise to implement certain methods and contain certain members. Scrum contains a set of guidelines represented in Ceremonies and Artifacts. The problem is that not only did I drink the “Scrum as an Interface Koolaid”, but I also manufactured and distributed that Koolaid for a while.

The problem with this analogy is that Interfaces are a strict prescription, and require an exact implementation… Exactly implementing every aspect of Scrum is not “non-prescriptive”, and that’s exactly what we did in the early stage of VideoAmp.

The biggest challenge facing the earliest years at VideoAmp was that we could not do Lean Startup by taking tiny bites and see if customers like it. As a startup DSP, we had to deal with building a large scale platform that could automagically spend our clients’ money with zero bugs.


Incidentally, this is why you don’t find kids in their dorm rooms building ad-tech. The sheer risk, size, and cost of the infrastructure, along with the Herculean effort to get feature parity with our competitors meant a long and arduous march towards our first platform release.

What slowed that down was following Scrum to the “T”. BECAUSE we were not iterating on potentially shippable increments. Instead we should have locked ourselves in the cave in the early days, and treated every engineer’s time like a precious little baby seal.

During those early months of VideoAmp, we presumed that we could drop in the entire Scrum process like a old tranny in a new frame, and it would just drive us forward. Instead we ended up implementing every facet of the Scrum Interface, rather than form-fitting it.

Therefore, the analogy to an Interface is not ideal, because the purpose of agility is to be customizable and form-fit around a particular organization. You can pick and choose, and you can drive with your feet if you want to. Some call this “Scrum-But…”, but we call it something else entirely.

It wasn’t until about a year into our company that I had the chance to visit one of our past clients. One of the leading ride sharing companies which rhymes with “Whift”.  I realized over this brief yet poignant lunch with their founding CTO that they too had over-applied Scrum early on.  At that moment it hit me like a ton of bricks. If a company of at least 180 engineers is doing less process than ours, then we’re srsly doing it wrong.

This leads me to a funny phrase used over the years which I call “Playing Company”. It’s akin to children who are playing house, who are going through the motions of what they think adult behavior should be.

They model this behavior by what they observe, and by what sticks with them. I have seen this recur with smaller stage companies where they get mired by what other companies do, vs. what they feel should be done to be effective at their present stage. Taking this a step further, we were “Playing Scrum” without regards to its overall effectiveness.

Prior to starting VideoAmp, I ran a company (“CLI”) that provided technology to post Series-A-funded companies. I directly hacked on many projects, and also sub-contracted about a dozen of the best hackers in the world to emerging companies in San Francisco. The engagements were partially remote and partially on-site. This gave an interesting perspective as to how companies were using Scrum (and operating with agility), and also how they subscribed to it with a differing amounts of buy-in to this Interface.

One of the blindspots that I now see was that I attended many of these daily / weekly ceremonies, but was less present for others. Is a company really going to pay a contractor to sit through a sprint retrospective or sprint review? Nope, not usually.

At best, we were part of their daily stand-ups, but we usually not part of sprint planning, story point poker, and retrospectives; and on rare occasions, invited to celebrate in their sprint review. In general, the ceremonies I attended with CLI as an outside contractor were pretty much the ones we are left with today at VideoAmp.

For early stage startups, I can reliably state that there are risks of over-applying process. In simple terms you can “beard-stroke” your company into nonexistence if you focus too much on “sticker charts” and “playing company” vs. optimizing product development yields.

I have seen the opposite of this this with multiple client companies who had read The Lean Startup (all the rage back then), and ticked the boxes, said “yeah, yeah, yeah”, and did basically exactly the opposite by taking a “Build it with a year long waterfall process… and they will come…”. But instead their cash burned away and the only thing that came was late payment notices from their vendors.

So I have seen many millions of development efforts blown due to a total disregard of process… AND due to an over-application of it.  So what is the right size then? 


“Great, so you did it wrong. What would you say it is you do here then?”


Some of the right-sized process that has worked for us of late:


In “Real World Agility” my brother speaks in regards to Agility as a way to address what-is-possible in software development. I love how he uses the “add a single field to a web-page” example, as I independently use that all the time to describe a full stacker.  At VideoAmp, we are moving into cultivating engineers into full-full stackers who can do this across our complex architecture, and go all the way from an Apache Spark data warehouse, to an Angular front-end Material Design widget. This has become such a theme that I am already writing a future post on just this topic.


Extreme Programming

Yup, that’s right we code until our eyes and fingers bleed. Just kidding. We are not alone in being nimble-minded and adopting many of the tenants of XP.  This is something smaller companies can enjoy by starting with a clean slate. A type of “later mover” advantage you get over large corps which can be mired in antiquated and/or glacial processes. Unless you are keeping airplanes from falling out of the sky, or making sure nuclear reactors are not melting down, you probably don’t need a Six-Sigma process.

For us, XP means a tendency towards test-driven development, systems which are driven by Continuous Integration, and where possible Continuous Delivery. We do not silo folks into a single functional role like “Front-ender”.

Instead, we give a collective ownership of code to an entire team. Every hacker is encouraged to branch out of their current area of expertise by learning more parts of the tack. This is another up-coming blog post on how we do this…

Minimal Meetings

We have a fairly sizable office given our company’s size, yet we only use about ½ of it. We all work in one big room, and with all the meeting rooms and break-out areas, people always get a room when they have to have conversations which may disrupt the working environment. There are also very positive synergies you get from everyone in the same room, such as leakage of what people are working on, what is blocking them, and when people need help other Hackers are quick to help.

Since we have a very close “in the office” kind of culture, we are able to survive with much fewer prescribed or standing meetings. We do not have to use standing meetings as a means to “force” different teams to work together; they are already coupled together loosely in cross-functional teams, and when they need to, they just grab a room with a whiteboard.

Still Doing

Daily Standups: where we really do NOT sit down. It’s a 2-5 min meeting, and people now resist the urge to get into “Story-time” and decide to have breakout convos immediately afterwards.

Spint Planning: which is an artful and scientific dance of product owners and cross functional teams and their leads, kinesthetically moving mockups (from design sprints) around on the board, and using dry erase to ascribe sequencing and owners, and eventually Jira ticket numbers once it’s all journaled. We used to do this in tandem with Jira, but it’s way to fricking slow. So now, we do everything on a super-long wall with everyone standing around, and let the memorialization happen shortly after the muse is done.

Sprint reviews: happen every month, and for us it’s kinda a big deal. We fly in most of the company from our remote offices in SF and NYC, and it’s the duty of every person not to schedule time over this event.

This is the big reveal for all new facets of the platform, and this is how we’re able to keep our sellers selling exactly what’s on the back of the truck.

Strong emphasis on 1:1’s: this is where stretch goals, and interpersonal issues are managed, and where we try to focus the reflection. Retrospectives for every sprint used to burn alot of precious time, and we’ve consciously decided to invest more into 1:1s vs. a standing and mandatory reflection time. That being said, we still reflect as a team and as an entire division from time-to-time when issues trigger the need, but it’s not during prescribed intervals.

Instead of Scrum, we have moved to a Kanban approach which is much lighter weight, requires much less administrativa, and allows us to still have a great point-in-time view of what’s backlogged, in flight, and done.

Where are we Heading?

As we continue to grow, we’re outgrowing the rule of 7 +/- 2 of agile teams. While we are promoting full-full-stack, we still can not survive with an “N-to-N” level of interaction on our teams, so we are scaling our teams by building pods of cross-functional members. This goes all the way from the data team, through API and frontend engineering, all the way to our Client Success (aka customer support) teams.

As a parting thought, I can see us possibly readopting portions of these Scrum procedures that we had originally started with, as a means of coping with continuing growth. But we’ll just have to wait and see.

If you made it this far, I’m sure that I have crossed the streams with many of you folks. I’m just stating what has worked and not worked for us thus far. Happy to hear your anecdotes and stories of real-world application of these principles of agility.

And, P.S. don’t forget to check out Daniel’s “Real World Agility”.


Dave Gullo

Dave Gullo is a hacker and entrepreneur who has been building the web since 1994. His roles have included founder / entrepreneur, software and systems engineer, human computer interaction, and technology team leader.
Some of the interesting projects he has been involved with include: bootstrapping one of the web’s first online snowboard shops, building large-scale unified messaging systems, engineering print-to-web systems and free-form full text searching of careers and autos, optimizing the performance of in-flight entertainment systems, building rental management systems for ski resorts, building innovative shopping experiences for ecommerce like “size-based shopping” and “car-figurators”, “tracking the online trackers” by automating thousands of concurrent browsers and observing network activity and data collection, and silver-plattering technology and top-notch hackers to post series-A companies in the Silicon Valley.

Today, Dave is the Co-Founder at VideoAmp where he hacks and prototypes new technology and manages the technology team’s growth and culture.