Welcome back to Search Optimization 101! Last week, our friend Charlie Hull from OpenSource Connections – the experts when it comes to Search Optimization – shared some helpful tips to help you determine if your search is broken (if you missed it that list – check it out here).
Today we are tackling the next problem – what does it even mean if my search is ‘broken’?
I’m Charlie Hull of OpenSource Connections and I’ve been helping companies build better search engines for twenty years. I’m going to help you find out if your search engine is broken – and perhaps help you fix it!
In my previous blog I gave you a list of things to check that would help you identify if your website’s search engine is broken. Hopefully you didn’t find too many problems – but for those you did find, let’s dive in a little deeper.
What do we mean by ‘broken’?
The job of a search engine is to take queries (what your users type in to the search box) and present a list of results that are the most relevant to those queries – and to do that job in a very short time. It seems so simple – but it’s actually quite hard to do well.
Remember that a broken search will almost certainly be causing frustration for your users. Trained by Google or Bing, they expect a search engine to understand what they mean, even if they can’t express that need very well. Unhappy users will find another way to answer their questions, or go to your competitor, or complain to you directly.
So what might be causing our search to be broken and what can we do about it?
It’s important to think about search optimisation as an ongoing process – we can’t fix everything in one go, and new problems may appear in the future as your data (and your users’ behaviour) changes. But let’s make a start!
Start at the source
The root cause of many search engine issues is the source data – your list of products or perhaps the documents you’re presenting to your users. This is after all what the search engine will have to search through. Ask yourself what level of control you have of this data:
- Was it created by your organisation and if so do you have editorial or content guidelines?
If different authors are using different terms to describe similar things (a classic example is the US language ‘pants’ versus the UK term ‘trousers’) or aren’t creating data in a consistent way this can be a problem.
- If it was created outside your organisation do you have any way of controlling, editing or annotating the data?
For example, do you classify products, or add a publication title to every incoming article, or clean up the data in any way? How consistent is this process?
No matter what some commercial vendors promise, even the most advanced search engine can’t easily make sense of bad data. This is a classic case of ‘garbage in, garbage out!’. So, data quality should be one thing you focus on.
Talking about search
It’s common to assume that search engines are solely the responsibility of the IT department. After all, they installed the software and keep it running!
The trouble is, unless everyone else understands at least a little about how it works and what it needs to be supplied with, your search engine may generate terrible results – even if it never actually shows any actual software errors.
The IT department probably aren’t experts in news, groceries, gardening (or whatever subject area it is that your business covers) and therefore may not know what a “good” result is.
Unless you communicate regularly about search – and how to make it better – then your poor IT department will probably get blamed for poor results, even though they need the help of other parts of your organization to fix things.
Search can touch many parts of an organization and should be seen as a shared responsibility across departments. You should create a ‘search team’ that regularly meets to work on improving search.
Measure and repeat
How do you know your search is broken?
If you read my previous blog you’ll probably have a list of problems – but this won’t be a complete list. Make sure you don’t just focus on fixing your (or someone else’s) own personal bugbears with your search engine – perhaps they’re not the most important things to fix anyway.
If you have a lot of data to search, or lots of users, then you will have a lot of queries. Some will be common (e.g. “shoes” or “romance novel”) and some more specific (“red shoes for evening”, “romance novel set in 18th century france”).
You should make sure to record all your user queries with as much other information as possible:
- What results were presented to them, if any?
- What else did the user click on?
- Have you seen this particular user before?
- How did they interact with the results?
It’s also important to think about what are the most important searches to your business:
- If you sell items via your website, what kind of results make you the most profit?
- Are some of your users more valuable to your business than others?
- Is your search engine helping you fulfil your mission or objectives?
This exercise will help you focus on which search optimisations are likely to make a big impact on your business.
Assess your control
Depending on the search engine you are using, you’ll have various levels of control available to you. It might be part of another software system that limits what you can do, or be provided by a third party as a hosted solution.
- What dashboards do you have available?
- Can you easily change and test new settings?
- Is it the latest version of the software?
Your IT department will be able to help with this assessment – but you should also consider what skills and experience you have available to you in-house. If you don’t have full control (and the knowledge of how to use this) then the improvements you can make will be limited.
We’re now beginning to understand some of the root causes of broken search; we’ve started to assess our source data, got our search team talking, begun to measure and quantify the problem and worked out how much we can change if we need to do so.
Check out the next post that discusses some practical steps you can take to optimize search!
This piece was contributed by Charlie Hull from OpenSource Connections.