Markets are increasingly complex, international, and contain a technological component. Thanks to patent landscape analysis, more companies are gaining an overview of a technology sector, their competitors and the chronological development of technology areas. Organizations are able to reach a better assessment of the economic value of their patent portfolios, and their strategic position based on the information they derive from landscape analyses.
Due to the prevalence of patent data it is possible to retrieve a lot of information. Patent data provides not only information on technologies, and patent applicants but also important tips on interesting markets, companies and sectors.
It has been claimed that utilizing patent information to the best advantage could shorten R&D time by 60% and save 40% of the R&D costs.
Even today with structured retrieval of this data in different databases it is still not easy to create a reliable patent landscape.
What is a Patent Landscape?
Several definitions are available however the most general is:
Patent landscape reports provide a snap-shot of the patent situation of a specific technology or company, either within a given country or region, or globally.
How does one perform a patent landscape analysis?
Before you start you have to become familiar with the topic by reading some general articles or reviews. Than as a first step you have to define what type of landscape you would like to create.
Several different types of landscape reports can be identified:
- Company Portfolio
- Regional (mostly country wise)
For the next step you have to take into consideration several different issues:
- Patent families
- A lot has been said about which family structures should be used in patent landscapes. It’s hard to define whether or not INPADOC families are useful for a patent landscape. The choice is topic dependent, but we know that INPADOC has a broad definition of a patent family, and quite often mix up different inventions into one family. For that reason, databases with a more formal family definition seems to be more useful.
- Citation searching
- Sometimes citation analysis can give additional records, although the different citation policies (especially between Europe and the US) can generate a significant number of false positives. A manual check of these references is often needed to determine relevance.
- Patent assignee names
- This is still a draw back for creating landscapes. There is no standardization of company names. Manual corrections are necessary!
- The use of classes in addition to keywords is absolutely necessary. For this reason only global systems are useful like IPC and CPC.
- For CPC we have to be aware that not all countries are assigning CPC yet. Depending on the technical area 10-30% of the families do not have a CPC assigned. Records coming from Japan, China, Korea and Russia do not necessarily have CPCs assigned to them yet.
- Do you want to include data from all countries? Sometimes it’s hard to retrieve data from some countries especially those that publish in a different language.
- Topic (broad or small?)
- Broad topics are seldom useful for landscaping. Topics like Pharma industry, Green energy, Automotive industry, or Nanotechnology are so broad that they cover many different classes. Topics within a small set of classes are more reliable for landscaping.
From a practical perspective, you do not want to underestimate the time needed to plan, and execute a landscape!! To create a patent landscape the following steps should be taken:
- Knowledge of the topic
- Acquire knowledge of the topic by reading some background articles or reviews!! You don’t need to be an expert, but at least you need to understand which records are of importance, and which are not
- This is the most critical step in the process. Documents missed in the first search are not included in any subsequent step. Therefore, a broad approach is essential. The different issues mention above need to be taken into consideration, and need to be included in the search.
- Clean up
- Normally a clean set will have between 80-90% relevant answers in order to create a reliable landscape. The basic set of patent data to be analysed must also be as homogeneous as possible. Differences in the spelling of patent applicants need to be eliminated, and individual patent publications for the same invention have to be combined into patent families. It is an advantage to use data from specialised databases, such as the Derwent World Patents Index, which edits, and processes the patent data rather than using raw data from publically accessible patent databases.
- Checking the data set
- At this point it is important to check whether or not your set makes sense. A good way to check is to look if some key patents that you have discovered previously are in the set.
- Sometimes it is necessary to focus on specific parts of an area. An example can be the difference between processes, and products or tools. Many companies are only focussing on one specific area. In the pharmaceutical area for instance, an analyst can focus on compounds, crystals and salts, formulations, or usage. Different aspects of the same category can sometimes be mixed up in a single INPADOC family.
- As soon as a clean dataset has been created the analysis of this data can start. There are many tools available to create overviews. Even basic handling in Excel can do a lot.
- It is important to visualize only global trends, and not specific data. Reports that are too detailed will not be read.
- As normal in reports, the most important part is the summary. Managers will may normally only read this portion of the report. A one-page summary is very efficient, and can provide the information needed for making a decision.
Generating patent landscaping reports seems simple, but it isn’t. Many steps have to be taken before a reliable landscape can be created. It’s important at the start to already have in mind what kind of landscape report you are going to prepare, and choose a topic of interest, but preferable not one that is too broad. It’s also extremely important to have a clean (80-90% relevance) dataset that the landscape is based on; otherwise the outcome will be rubbish. It is also important to avoid talking about white spot analysis, that’s something like looking for black holes. Finally, do not use landscapes for questions that require a legal opinion (like Freedom-to-operate conclusions!!). Patent landscapes are not aimed to be as precise as other patent searches.
Since May 2014 he is working as an independent consultancy in Patent Information, and is founder of Patent Information Services B.V.