I was working in an investment bank. We were connecting seven systems, which by today’s count is really nothing. We had 12 teams looking at everything, data for every application, someone doing connectors and adapters. The project took almost two years. We spent 30 million euros.
When we went live, basically, we just we’d been pushing messages through to the point where we had a fairly low error rate and then we deemed it successful. I just remember thinking, wow this is painful, like almost two years to connect seven systems.
I remember thinking if this is success, then we actually could do a lot better.
Have We Learnt How To Handle Data Any Better?
I think the one challenge all companies still have is that data is extremely siloed.
The problem is with the volume of data and the lack of meta information and context around that data; means that indexing across two parts of the business to get a combined dataset that makes sense to someone is still really difficult. I still think people are struggling with that.
With things like GDPR it was the first time many non-regulated industries were suddenly having to adhere to a regulation around how they treated data. I think it was a bit of a wake-up call for many organizations.
[These companies stated that] We actually don’t know where our customer information is, we don’t know how relevant is, we don’t know who stores what where.
I think probably many companies are still struggling with really knowing where their information is being stored, but everything new that gets built is regulated the right way, which is obviously very positive.
The data is oil thing is nonsense. It’s nonsense because you can’t just plug data into a machine and power something. It actually doesn’t work that way. And so just collecting data in the way you would collect oil in drums makes no sense.
You shouldn’t let go of the data I admit, but I think you have to have a top down approach. What are the insights that we believe as a company we should be getting and work backwards from that.
I think everyone keeps starting from the bottom, which is let’s collect data and then put some data science on it to figure stuff out. I think you should be doing that of course because you have to figure out mechanisms of managing that data. But actually, it’s really up to the business leaders and the front of the business to say; what do we think we should know about our world, about our customer. [Then] architect it, not with what data you have, but what do you think you should have, or should be doing, and then work backwards.
What we’re seeing is the companies that do that tend to figure out more products more quickly.
Mulesoft v Salesforce?
With any acquisition there’s always going to be cultural differences, and the one thing I’ll say is if you look at acquisitions historically, a lot of them don’t go very well. They don’t go well because the human aspect wasn’t managed that well.
I think one thing Salesforce did really well with us was really gave us a lot of space to be ourselves, but also time to learn how they did things because ultimately we had to lean into the process versus lean back.
Of course, there’s been some hiccups, the Salesforce cultures’ a bit different from our culture. At some point we’ve had to adopt their way of doing things, like hiring, where we had a very specific hiring methodology that we had to let go of.
We had quite a heavy hiring process to find the right people for our organization, but Salesforce hires like 6,000 people a year. So that process is a filtration process which has put everyone in the pipe and filter out at every stage as many people as possible, and the people you have left are roughly going to fit in the organization. Then you have well-defined roles and the ones that do really well light up and the ones don’t either move to another role or move out of the organization.
But you realize if you understand how Salesforce operates, that there’s just no other way of doing it. So we gave up our hiring process and now salesforce runs it, turns out it works just as well. It’s just that we hadn’t got the, I guess, the feedback loop running hiring at that scale.
Company Founded, Built up and Then Bought, Why not kick back?
I took a few months off. I went skiing a lot this last season.
What’s interesting about where we are is most software companies when they go IPO they’ve got a product and now they are just scaling it. When we went IPO, we were actually really starting the next generation of product to run application networks, which is taking APIs and integration, but how do you organize that as a communication layer across the enterprise?
So we haven’t really run out, there’s still a lot more runway for us. It’s still very early in the enterprise for APIs, even though most companies have an API strategy. Most people are only two years down, what we find is they build a lot in the first couple of years and then they realize, actually we need less but better ones, and you know it evolves.
That’s very motivating, because it means there’s a lot we need to do. There’s a lot more innovation left to happen.