Charlie Hull
Category: Process
I’ve worked in search for twenty years now – originally at Muscat with Dr. Martin Porter of stemmer fame, building an early web search engine, then for 17 years at Flax working with clients all over the world from governments to startups, and now at OpenSource Connections. I’ve seen a fair few search projects over that time, and not all have gone well. I’ve seen everything from basic misunderstandings of what a search engine is for and can do, terrifying data quality issues, lack of a project plan, uncontrolled outsourcing, unrealistic timescales, uncaring management…the list goes on. I’ve also seen some great search projects with effective teams (I have fond memories of one for a major newspaper publisher which went like clockwork, and one for a recruitment company that took many months – this was delivered late, but only by a single day!).
As a search consultant one is usually brought into a project when the client has realised they don’t have the skills or experience in-house and need some help. If you’re really lucky, this happens before the project actually starts, but more often the project started months ago and isn’t going very well. There’s only a fraction of the data indexed, queries are terribly slow, the UI is very pretty but isn’t actually wired up and there’s already murmurings from upper management about how slow everything is and when will it be finished, because we’ve got customers waiting, don’t you know. So, the natural reaction is to bring in an expert, who will fix everything. Hopefully they can start on Monday and just dive in and start coding. People start saying things like:
“What’s your hourly rate? Can we get you a login to our system over the weekend? When will it be fixed?”
The trouble is that this is exactly the wrong approach. Simply adding more people to a already broken project won’t help (although unfortunately it’s a common tactic used by some of the larger consultancy outfits, after all the more billable hours the more profit, never mind if the project is never actually finished…). Sure, the expert can probably find a few problems, sort out your configuration a little, get some more data indexed…and then before you know it you’re several more months in and the same structural issues with the project remain, no-one really understands what the external consultant has built and you’ve burned up a lot more hours and money. No-one comes out of this looking good, including the consultant: he/she has just made you dependent on them, possibly for years, for endless tweaks and bugfixes. This probably won’t be a healthy relationship.
So, that’s not how we work at OSC. We have no interest in making you dependent on us, although we do want to help you fix things. We want you to own your search – this may not happen immediately but it’s a target, from the very beginning. To make this possible we’re going to need to take a step back, even if you’re already running late on the project.
We’re going to start by helping you to analyze your search maturity on several axes to understand where the gaps are in your capabilities and what are the most important ones to concentrate on to achieve your aims. We’re going to start with a Discovery, a process we’ve run many times before, using tools we’ve developed based on that long (and occasionally painful) experience I mentioned above. Only after we’ve done this (it takes around a week or two including preparing our assessment) can we move on to implementation, and we’re going to do that in manageable 90-day chunks of time – we call these Accelerators – and after each one we’ll all re-evaluate and check we’re making good progress on those aims.
We call this the OSC Proven Process and we try to stick to it – even if you really, really want us to start coding straight away. Trust us, you don’t want to do that. We all need to completely understand the why of your project as much as the how. If possible we also want to train your team so everyone (even the non-technical people) have a common language to use and a baseline of understanding. During the Accelerator itself we’ll constantly be working to pass on knowledge about what we’re doing for you, to make sure we don’t inadvertently build in that dependence – we want to be able to work with your team, not just for your team. Eventually we hope your search team will be able to stand on their own.
It’s hard to take this route when a search project is already delayed – it might seem that it could delay it further – but in the long run we think it’s the right approach. We think it significantly reduces the risk of a project going off the rails and becoming one of those painful anecdotes.
If you want to know more, or even if you disagree, do get in touch. I promise not to tell you too many horror stories!
Image from Vector Vectors by Vecteezy