By Parijat Oak (Associate Director, Intellectual Property Services) and Ed White (Director, Patent Analytics)
Patents are a valuable source of technical, market and competitive information. But how do you make sense of the vast and growing volume, with 120 million published patent documents in total and 6.3 million published last year alone?
Patent analytics provides an essential and practical key to unlocking the value of patents information. Through patent analysis, we can begin to understand the strengths, weaknesses and opportunities of patent portfolios, both of our own organization and those of our competitors, patenting trends around the world, technology landscapes and where white spaces may exist, and so much more.
But patent analytics is not for the faint-hearted. It requires a thorough understanding of the underlying data, how and for what that can be used and, perhaps more importantly, the limits of what can be achieved. The Derwent Patent Analytics Services team have a wealth of experience and expertise in conducting patent analysis. Laid out below are our top 10 tips for getting the most out of patent analysis.
#1 – Know Your Lens
There are many different approaches to working with and analysing patent data, but it is important to know which approach to use and when, and in what order. Analytics has many analogies to photography and imaging – resolution, the power and trade-off of different optics, digital versus optical zoom.
For example, just as you wouldn’t use a microscope to look at the moon, you can’t use the methodologies of freedom to operate searching to review a whole technology sector. Equally, a Patent Landscape is extremely tough to “zoom” into to support a patentability legal opinion. A landscape requires the use of analytical approaches aimed at creating a picture of several thousand data points. Zooming down to just three or four data points and forming an insight is the equivalent of zooming a photograph down to a few pixels- it will be blurry.
A better practice is to order your tools and approaches into a process. You can use the methodologies of landscaping to identify regions of the picture which you can then re-image using the methods of patentability or FTO. This is the equivalent of telescopically imaging the moon to select where you will send the robotic spacecraft to image the surface directly. Sending thousands of robots with cameras on them is possible but somewhat impractical. Both methods are equally valuable, but have their place depending on the question. When working with patent data, it is important to consider the available approaches and when to use them.
#2 – Know what you are measuring
Patent datasets have a lot of complexity in them, not least the fact that the same idea or invention can produce many individual patents and patent applications. Many analyses in the public domain refer to “patents” but without reference to what that means. The differences can be huge. You need to be aware of whether you are counting patents, or patent families or inventions. If you don’t take this into account, your conclusions will be off. As a simple rule, for accurate patent analyses you will almost always need to “expand” your dataset to include all family members and then reduce it back to counting families. This is easier using Derwent World Patents Index data, as the structure of the database surrounds sets of claims as they appear at the different patent offices of the world; thus, giving you the ability to analyse ideas and inventions, instead of “patents” – which for many (arguably most) analytical use cases is not ideal.
#3 – Dive deep and let the data speak
There is almost always a need to clean patent data, even well-curated data. The reason is that all patent databases must cater to a wide audience of potential users. Your analysis on the other hand, to be as useful as possible, needs to speak to a much smaller readership and user base. That means that context needs to be added.
Cleaning assignee information (essentially, the organization that owns the patent), indexing the technologies covered, the benefits of the inventions, industrial uses and the types of patent applicants all add this context, making the data specific to your company or organization. All the data visualization that takes place after this point then becomes easy to use, easy to read and easier to act upon.
Depending on the mission or the question (for big data sets, there are data science alternatives), there is value in the deep manual dive. It gives you detective rights and you can tease out the underlying stories – changes in ownership, hidden mergers, country expansions, technical development and the simple fact that you as the analyst have eye-balled every record. Knowledge of the data at this level provides the source for the narrative the data is trying to tell you – let it speak.
Read the full whitepaper now to see the top ten hacks for essential keys to unlocking the value of patent analytics.