How do we get faster?
Or keeping up with the the pack in the age of acceleration
Everyone wants to get ahead of their competitors. The metaphor “get ahead” implies speed, and the ability to move quickly is the key competitive advantage in the 21st Century. How well can we adapt to the latest geopolitical changes? How do we adopt the latest technologies faster? The businesses that can move faster are the ones that will be the most successful.
So, what do we mean by moving faster? When I was at T-Mobile in the early 2000s, we had an average project delivery time of 18 months. Our systems were generally business critical - no one wants their phone network to go down - and we generally worked with big projects that required a lot of planning.
Our projects were also relatively straightforward. Rolling out a mobile phone network is basically running an implementation against a globally agreed standard. We knew upfront what we were looking to deliver, and once we had delivered that, we knew we could declare victory.
However, most projects aren’t like that. At MarketCast, we recently launched our Brand Effect Resonance ratings as a new part of advertising currency. The high-level requirements are to provide a way to measure how the ad experience impacts the effectiveness of the advertising. Taking a concept like that and making it a reality, the challenge isn’t the operational issue of synchronizing 100s of people working together; it’s about working out in detail what you need to do.
In cellphone engineering terms, that’s the equivalent of designing the 3G standard rather than rolling it out.
So, how do we run fast when unsure of where we’re heading?
There are three things I’ve learned:
1. Make sure you have fast iterations
This is the mantra of agile software development, but you can apply it to pretty much any practice. Cut things into smaller manageable chunks, deliver them end-to-end, and then work out how you need to improve them.
If you’re building a data product, don’t focus on ironing out all the rough cases before you start putting data in front of people who use it. Get it in front of people early and get their feedback on what they think needs to be improved. It might not be the rough edges that you’re thinking of.
The biggest drag of time - and morale - is re-work, and early, continuous feedback is the only way to prevent that. Make sure the feedback is fast. Once you’ve got something, get everyone in the room and discuss how to improve it. Don’t wait for it to be finished before you test whether it’s the right thing. We once built a beautiful dashboard for clients to log in and review their results, only for the client to say that they just wanted raw files dropped into the AWS account. They already had internal tools that needed the data loaded into, and our pixel-perfect dashboards were never used.
How much faster could we have got if we’d iterated faster?
2. Automate what you can
It goes without saying that new tools are coming to market pretty much weekly, and the reason there’s such turnover in tooling because they are constantly getting better. ClickUp is probably the latest darling of work management tools, and it pretty much eliminates the need for status meetings. When you get teams together, you don’t want people to be discussing what they’ve been working on, or checking on progress, you want people to be collaborating on meaningful problems that are holding you back, and by automating much of the project management, that’s where you can get to.
Testing is another area where massive leaps in automation have arrived over the last 20 years. We used to have large teams of people manually testing things on various devices. Now, you can write automated test suites in tools like Cypress that will run through a full suite of regression tests any time anyone makes a feature change. Most teams don’t invest upfront into getting these things right because they think it will take too much time and slow them down.
They couldn’t be more wrong. Investing upfront in the automation of tasks you’re going to be doing a lot of downstream is exactly the way to make sure you’re faster than the competition.
3. AI really can do the boring bits
Do you need to match together a set of different identifiers from different datasets? Or write code to parse a complicated data structure and turn it into a flat file? Perhaps create terms of service for a new product? Even write a first draft of a press release for your new product.?
These are all tasks that you should now be using generative AI to assist you with, and it can really make you a lot faster. Look at how much time you spend in your day doing repetitive things that you could get an AI chatbot to do for you. I now constantly use GitHub’s copilot when writing code and ChatGPT for data matching or idea generation. I even occasionally use it for a bit of desk research, although it really does need a lot of fact-checking. DALLE makes all my images.
AI’s don’t look like they're going to replace much, but they will make us all a lot faster, and that’s what we need to keep ahead.