Patent Search

Novelty Search Concepts and Strategy using Google Patents

Several patent search strategies and concepts can improve a novelty search. Below are several that are useful when using Google Patents as the primary search tool.

The Novelty Search

First, what is a novelty search? A novelty search is also known as a prior art search and a patentability search. The novelty search is the search that an inventor would use to determine if an invention was new. These types of searches seek to understand what the prior art teaches.

The Novelty Search is Not:
  • A State-of-the-Art Search: A state-of-the-art search is a search designed to bring a person to date on what is going on in the field. However, state-of-the-art searches do not reflect the absolute latest on inventions because patent publications generally lag application filings by 18 months.
  • A Freedom to Operate Search: A freedom to operate search is a type of search focused on determining whether someone is free to do something. For example, a person might want to know whether they are free to build a particular device or practice a particular method. These types of searches focus more on what the claims of a patent cover rather than the what the documents teach.
  • An Invalidity Search: An invalidity search is a search that focuses on finding prior art that would invalidate a particular patent or group of patents.

Search results separate mere ideas from patentable inventions.

Why search in the first place?

Prior art is the most important factor in whether your invention is patentable. Search results separate mere ideas from patentable inventions.

When you have an invention and you don’t know whether that invention is new, search.

What are you looking for? You want to find the piece of prior art out there that is the best possible match to your invention.

Nothing in this blog should be construed as legal advice or as creating an attorney-client relationship.

What is Prior Art?

Prior Art is … Prior

Prior art is a common term in the patent search world. Generally, it means documents that can be used to show a patent or patent application is not new. For example, a patent granted in 2017 is prior art against a patent application filed in 2019. In most places, a patent application filing date provides the reference point for what is “prior.”

Art? … Think of ‘State of the Art’

The term “art” in the phrase “prior art” is used in the same sense as the phrase “state-of-the-art.” In simple terms, you cannot get a patent on something that was already known when you filed your patent application. Prior art is the collection of documents and other things that make up what was known. How broad is prior art? Prior art can be in any language, in nearly any format, and go as far back as the earliest recorded history. Some of the more common forms of prior art are patents, journals, papers, websites, products, and manuals. There are also other things, like uses, that can be prior art, so it is generally a good idea to contact a patent attorney to determine if events not in the public record can lead to patentability issues.

Goals for a Prior Art Search

Find killer prior art

If it is out there, find the document that matches your invention. A great match is called killer prior art.

Killer prior art is killer because it kills your chances of getting a patent grant.

SMILE! … You just saved thousands of dollars.

When an invention dies it hurts. But it hurts twice as much when it dies after years of work and thousands of dollars.

When inventors do a lousy search … examiners find what inventors miss. Use the techniques below and you will be far less likely to fall into this trap.

find the closest prior art

Finding close prior art seems like it would help the patent office reject your application. This may be true, but it is often a misleading notion. Examiners typically conduct a fairly detailed search. That search commonly finds the prior art that the applicant found and more. Also, if the Examiner never becomes aware of the most important piece of prior art, any patent that issues will likely be a very dubious value.

You would like to find the most relevant prior art because that will allow your patent attorney to write the strongest patent application. One of the fastest ways to jeopardize the quality or validity of a patent application is to miss the most relevant prior art. Essentially, if there is a big obstacle in the road to patent protection, your attorney needs to know about it.

What to Search

Keyword searching versus class searching

Keyword searching is great because it narrows results to specific concepts and combinations of concepts. Keyword searching is conceptually simple and can yield valuable results with very little effort. However, there are downsides to keyword searching. First, similar keyword searches can yield large batches of overwhelmingly similar results. For example, three separate searches each yielding 2000 individual results may all have 1000 of those results in common. Searching through repeat results over and over again may be an inefficient way to find the result you’re looking for. Second, terminology can vary greatly from application to application, so relying on a seemingly relevant search term may inadvertently exclude the result you’re looking for.

Class searching is a little bit more difficult but has some distinct advantages. First, language limitations tend to be less problematic because the patent documents are grouped by their relevant features. Second, classifications are distinct enough that a complete search of a classification means that the classification can be easily eliminated from future searches. That avoids the needless review of duplicate search results.

Use ‘Close Results’ to Guide Your Searching

How do you know if a result set is a valuable result set? One way to evaluate whether a result set is valuable is to estimate the density of close results. A close result is a result that is similar enough to what you are looking for that it indicates you might be on the right track.

Close results include any of the best references that you might find but are much more numerous. It is likely that you will disregard most close results as not being the results you are looking for. However, even when you cast aside these close results, that does not mean they are not valuable. These close results found in a result set are your best clue to the density of quality results in a result set. This is a valuable enough indicator that you may want to count the number of close results you find in the first 20 documents reviewed. Because 20 documents is likely a meaningful sample size you will then have an idea if your search string has you on the right track. If you see five close results in the first 20 documents reviewed, then there is a good density of close references.

The documents you are looking for are most likely to be found where there is a high density of close results. So, counting close results with tally marks is not a bad idea. When you enter a search string, review the first 20 results giving a tally mark for each result that is a close result. If your last three searches had between 1/20, 3/20 and 2/20 close results and this search string gave a 5/20 close result density, then you probably want to go ahead and review the next 20 results. If you make it to a 10/40 density of close results you may then proceed to the next 20 results. Because Google patents relevancy ranks results, it is natural for the density of close results to fall off as you dig deeper into the result set.

Change your search strings based on close results

When you find a higher than normal density of close results that is a positive indication that can be leveraged into even better search results. If you know which search strings are giving you better results, then go back and look at the search string. Compare the search string to previous search strings that did not give result sets of similar quality. What is the same? What is different? You are likely to discovery search concept that may have improved your search results. Use the next few search strings that you generate to test which search concepts seem to give a better density of close results. You may want to highlight or make other notes about the search results that yields the highest density of close results. Continue refining your search strings to include combinations of better search concepts.

using AND and OR and Not (-)

Strategic and flexible use of AND and OR operators can allow you to radically change the size of result sets. A standard use of “pencil” and “pen” in a search would be (pencil OR pen). But if those terms are both search concepts that individually give a high density of close results it may be worth trying (pencil AND pen). If the density of close results is even higher, then you may want to spend even more time with that result set. Also, a search string with the unconventional AND along with a few other search concepts may yield a result set that you can completely review in a reasonable time. ANDing together synonyms or other highly related words tends to be most useful when those search concepts are very close to the core of the invention.

Title searching and abstract searching

Searching by title and abstract is a quick way to drastically narrow a group of search results. As with any drastic narrowing, it is very important to be aware of what is being excluded. If you keep in mind that a highly relevant result may be unintentionally excluded, it is a great way to get a result list down to a manageable set of results to review. In other words, if you want to completely review a list, a title or abstract limitation can quickly narrow a 10,000-result list down to 100 or fewer results.

Dividing the pie

Every group of results divides the mass of all possible results into a more manageable slice of results that can be reviewed one by one. Some slices of results are independent of one another like independent classifications. However, most slices of results generally overlap other slices of results. Because of this, consecutive searches commonly feature many repeat results. While inefficient, this is not all bad. Because Google patents highlights different things in different searches, the second search may draw your attention to the feature that makes a document relevant during the second pass of document reviews. However, it is important to avoid turning major sections of the review process into wasteful duplication.

How to De-Duplicate your Result Sets

One way to avoid turning major sections of the review process into wasteful duplication is to use the synonyms feature. By using the synonyms feature, similar terms are grouped together into a single search so that a single concept is not searched over in multiple passes. Classification searches have an inherent ability to limit duplication by searching distinct categories separately.

After deciding which strings are appropriate to search, it is helpful to consider how those search strings relate to one another. Namely, are the slices of results from the search strings likely to have significant overlap? When your search strings look like they will be creating duplicate results, you can use the NOT operator (-) to remove results from a results set. For example, first run a search with an abstract limitation then in a subsequent search NOT (-) out that abstract limitation. The same technique can be used for classification (CPC) limitations. If you keep in mind how search strings overlap each other and monitor the density of close results you will see a more close results and you searches will be more productive.

How to Judge your progress

choosing how much to narrow

It is often a good idea to try a mixture of search strings that create very narrow result sets and search strings that create rather broad result sets. The very narrow result sets allow a searcher to evaluate every document in a result set. There is a clear advantage to actually observing details relating to every result in a result set and that is especially true when there is a high density of valuable search results.

Automated Relevance Ranking

Anytime you don’t review all 1042 documents in a result set you are relying on the Google Patents automated relevance ranking to bring the most relevant results to the top of the result set.

The Google Patents automated relevance ranking is a tremendous timesaver when it is doing a good job to bring the most relevant documents to the top of the search results.

Relying on the Ranking Algorithm

When using a search engine like Google patents to evaluate search strings, there is significant value in allowing the ranking algorithm to bring the more relevant results to the top of the result set. In essence, for a search string with a 1000 document result set Google may have sorted the best result into the first 50 documents. The ranking algorithm is an amazing feature when you would not have reviewed the complete result anyway. But the ranking algorithm should not be trusted blindly. When a search string produces a 65 document result set there is a good chance that reviewing the entire result set would be valuable.

It is probably unreasonable to expect consistent quality sorting by Google Patents. The document you are looking for may just not have the attributes that would bring it to the top of a Google Patents search results list. When that is the case, you will need to rely on deeper level document review and usually narrowly focused searching is typically the answer.

It is a good practice to tailor your search strings so that some result sets are small and some result sets are large allowing you to do a full review of the smaller result sets and rely on the Google Patents automated relevance ranking to aid in evaluating the larger result sets.

How many combinations of narrowing?

If you have done a good job of identifying search concepts, then it is likely that the number of combinations of those search concepts is beyond what you would reasonably try in search strings. This is where judgment and attention to close results help you estimate which search strings will be most beneficial.

Look for 5 Great References

As you go through the search process it is important to quickly evaluate each reference that you see as either relevant or not relevant. While you may find a larger or smaller number of references, it is helpful to think about finding the five most relevant references. Each time you see a reference, ask ‘is this one of the top five references?’

The Scariest Reference is the Best Reference

Knowing how to deal with the scariest reference is exactly where you want to be when you file a patent application.

The Scariest Document

Sometimes when you find a reference you will think, “oh no that is really close to my invention.”
Write down the document number so you don’t lose that reference and then start digging into the reference. How many of the features of your invention are found in the reference you found? How is your invention better than the reference you found? Is there anything in the reference that you found that would make your invention better than it already is? Does your invention have some commercial value over and above the teachings from the reference you found? What is bad about the reference you found? Are there features of your invention that are more apparent now that you have something close to compare it to? The answers to these questions will help you both improve your invention and understand what is genuinely new about your invention.

If you have found the absolute closest piece of prior art, consider yourself lucky. First, having that piece of prior art will allow you to make the best upfront decision on whether to pursue patent protection. Second, having the best piece of prior art will allow your patent attorney to write an application that clearly shows the differences between your invention and the best prior art. Rest assured, that these differences are the differences that matter most.

What to Record when you find a reference?

When you find a reference that appears to be close prior art, right down the document number

US 7,654,321
US 20110315084A1

You may also want to write a small reminder about why that result was relevant.

Should you add a professional patent search?

There is no such thing as too much patent searching, but running out of money is a real thing. That is the essential balance that drives professional search decisions. In other words, if I spend the money for a professional search will I find that reference that will save me the cost of a patent application? Or if I spend the money for a professional search will I find that reference that really adds value to the patent application? The answers to these questions are a pure cost-benefit analysis. It can be difficult to make that decision but searching on your own can help you reap some of the benefits of a professional search without running up your search cost.

A good result set is concise

What makes up a good set of search results is largely subjective. A concise set of results is often more useful than a sprawling list of references. For example, if a searcher finds 15 seemingly relevant results, but cannot identify why half of those results are relevant results, many of those results are likely irrelevant. Often in a search, results will seem relevant until better search results show up. That is a quick way to get a list of 20 documents where half of the list is irrelevant.

Dive Deep into the Top Results

Is there a dominant inventor or two?

Whenever one or two inventors dominates a particular search area you are investigating, it may be a good opportunity to investigate the group of inventions associated with that inventor.

Is there a dominant company or two in a particular field?

Whenever a few companies dominate a particular search area that may be another opportunity to investigate the group of inventions associated with that company.

Skill Level?

Skills and knowledge matter in searching. If an invention is in an area that the searcher is not familiar with then industry or scientific terms may elude the searcher. These terms may be learned, but doing so may involve a significant degree of independent research. It is probably worth the extra effort and time for the searcher to spend time becoming familiar with these terms. Search skill is not particularly hard to obtain and primarily involves a series of tips, tricks, and a basic understanding of how a few systems work.

How many embodiments does your invention have?

An embodiment is a particular version of an invention. Commonly, a patent will have a set of drawings dedicated to one version of the invention and then another set of drawings dedicated to another version of the invention. Other times, embodiments are smaller variations like substituting a hinged door with a sliding door. These concepts can be particularly useful in controlling the scope and breadth of a search. A search focused on a particular embodiment can result in a very narrow result set. Conversely, a search that attempts to encompass all of the embodiments of an invention can be quite broad and give a very large result set. Using search concepts that come from various embodiments is a good way to manipulate and improve the sets of results the searcher reviews.

Web Alternates

Sometimes the best place to look for an invention that is a consumer product is the websites that are full of consumer products. If that is the case you may also want to do some similar searching on, and

How to Search Methodically?

keep track of the search concepts you use

As you go you will want a running list of search concepts and a running list of the searches that you have already tried. Google Patents does not have a search history, so it may be a good idea to keep a record of the search strings that you have executed.

Keep track of search strings that you use

You don’t want to cover the same ground twice.

Keep track of which search strings yield better result sets

This will help you make decisions about what search strings to use next.

Keep track of your top five results

When you get to the end of your searching what you really want to know is which references were the most relevant so don’t loose them in the shuffle of searching.

The power of Class Searching

Classifications are hierarchical, distinct from one another and designed not to overlap each other. This has special search significance in that classification search result sets from different classes in the same classification level will tend to have far less overlap than text search result sets. Certain patent documents are classified multiple times so those documents will show up in more than one of those distinct classifications.

Google CPC classification does not always match the CPC classifications of the patent office that classified the document. Multiple CPC classifications may apply to any given document. Google appears to be using machine classification as part of an attempt to enhance the classification of documents.

CPC classification definitions are available at and in Google Patents. The google patents version is quicker to get to, but the version is more useful and user friendly.

Using NOT (-) so that you do not re-cover the same ground.

You can use the NOT (-) to make sure that you do not cover the same ground over and over again. For example, a search of AB=windshield AND AB=wiper could be followed by a search of AB=windshield AND AB=-wiper. This would be an appropriate search sequence if the greatest interest was in the search concept ‘windshield’ and there was secondary interest in the search concept ‘wiper.’ In some respects, this search sequence is better than the simple search AB=windshield because it breaks the entire AB=windshield result set into two separate groups which can separately be evaluated for the density of close results.

How Much Searching is Appropriate?

A single search is an imperfect search

In most cases, the applicant will do some searching, either on their own or through a patent attorney. After the application is filed the patent examiner conducts another search. Even when the applicant gets a professional search, it is common for the Examiner to find highly relevant art not discovered by the professional searcher. This happens because examiners are skilled searchers that intensely focus on the claims. Those claims are usually drafted after the applicant’s initial search. The following sequence is common:

  1. applicant describes the invention to an attorney
  2. attorney has the invention searched
  3. attorney determines what is new about the invention
  4. attorney writes claims based on what is new about the invention (i.e. Something that wasn’t in the original search results)
  5. attorney files the application
  6. examiner searches what attorney described as new and finds a new and highly relevant result because examiner’s search was narrowly tailored to features that were not the focus of the original search.

The sequence is common and reasonable even with highly skilled professional search firms and patent attorneys.

Should you out search the patent office?

In an ideal world where money was not an object, each applicant would conduct a massive search for each invention and patent examination would offer very few surprises. Because professional searches are costly, most applicants get a limited professional search or no professional search at all.

How to out-search the Patent Office

Is it important to try to out-search the patent office before you file an application? Every case is different, but for most inventors and applicants my answer is no. I prefer for applicants to evaluate their personal knowledge of the field and then search until they feel they have made a reasonable effort to uncover the best prior art. Sometimes this involves a search by a patent attorney. Sometimes it involves a professional third-party search.

The Optimal Level of Search

The likelihood of the examiner finding new and highly relevant results can be reduced by iteratively searching and working on claims. However, that can be a very expensive proposition. Because repeated rounds of searching and claiming are usually cost prohibitive, most people do not do this.

Don’t skip Free & Easy

Whatever paid options are elected, I always encourage people to begin with some free searching using Google patents. Some people just want to pay for a search and not be bothered. That is fine, but it forgoes one of the highest value opportunities in the patent process. Essentially, if a free 10-minute search can change your entire outlook on an invention, then it makes sense for most people to take that opportunity.

Maneuvering the results

The first time you hit the results page after entering a search string notice the total number of results associated with that search string. Are there large number of results or small? Is it greater or fewer than you expected? The answers to these questions will help you decide how deeply you want to dig into the search results. You want to evaluate how effective the narrowing of the results was. If there is a relatively small number of results and those results appear to contain significant prior art than the narrowing was very effective. If the number of results is very large and there is a very low density of relevant results, then the narrowing may have been ineffective. Keep recombining search concepts into new search strings. Then pay attention to whether the results are better or worse. Each search concept in the search string is either contributing to quality results or not. Once you have evaluated a few of the search strings it will start to become apparent which concepts are boosting the quality of your results.

Reviewing Results

Using each of the views

There are six views in Google Patents that can help you review results. Depending on what features you are looking for you should choose the view that helps you sort through the results the most efficiently.

The Three Main Views:
  1. Results View – The simplest view. It resembles a result set from a regular google search except that drawings are included with most of the results.
  2. Document View – The document focused view that highlights the basic information like abstracts and thumbnails of drawings.
  3. Drawing View – A drawings focused view that is useful for document review when the drawings are the easiest way to determine if you have the right result.

Full Window Views are the views that dedicate the entire window to one of the above three views.

Side-By-Side Views are combination views incorporating two of the above views into a single window. The three Side-By-Side Views are:

  • Results-Document Side-By-Side View
  • Results-Drawings Side-By-Side View
  • Document-Drawings Side-By-Side View

Duty to disclose

Applicants, attorneys and others closely associated with the patent application process have a duty to disclose relevant information to the Patent Office during the application process. This responsibility does not require a perfect memory, but it is a good idea for patent applicants to identify and record the most relevant pieces of prior art to meet that duty of disclosure.

Why Patent Protection?

right to exclude

If you get a patent granted in the United States, you gain the right to exclude others from making, using, selling, offering to sell, or importing the invention. (35 USC 271)

Limiting competition beyond the right to exclude

Give your competition the ‘easy’ way out. Competitors frequently look at a patent and decide that it is easiest to go in another direction, i.e. not knock-off the patented product. Why? It is often easier for your competitor to walk away from a product they have not started developing rather than spend the time and energy needed to seek appropriate advice from their own patent attorney. In that case walking away is both simple and cheap.

An often overlooked advantage of having a patent is that sometimes competitors simply choose not to compete with you because they never spent the money to find out how broad your patent is.

increasing cost and complexity for your competition

When you have an exclusive right in a particular market, competition will often attempt to work around that exclusive right to the best of their ability. Sometimes those workarounds can be costly, complex, and make their products or services less desirable.

reputational benefits

Patents also offer reputational benefits. Because people recognize that getting a patent is not exactly an easy process many recognize a patent as a sign of prestige and industry leadership.

People also recognize that products bearing patent numbers and patent pending indications tend to be unique and differentiated products. Essentially, there is a general assumption that if a product is patented, competitors’ products will not have the same set of features.

Searches can help answer questions relating to value

Searches are an important way to determine how valuable an invention is. A groundbreaking patent in an important business area can have tremendous value. Other patents are so narrow that their primary value relates to marketing. Patent searches are the best way to generate a preliminary evaluation of the potential breadth of the patent. However, a real determination of value requires considering the search results in light of potential business opportunities.

Is the original technology patented?

When an invention is an improvement over an existing product or process, keep in mind that the original product or process may have been patented. This can greatly aid your search because finding the original patent or application can open the door to finding other related prior art.

Am I infringing?

When the original base technology is patented, it can be very important to find out if the patent on the base technology is currently in force. If the patent on the base technology is in force, then it is important to determine if the invention is likely to infringe on the original patent. When this becomes a concern, patent attorneys can assist in evaluating infringement.

Is it alive, is it dead?

Just because a patent is in Google Patents that doesn’t mean it is currently in force. Google Patents provides status indicators regarding whether a patent is currently in force. However, if your business will rely on whether a patent is in force, it is probably better to seek the assistance of a patent attorney rather than rely on the automated and potentially out of date Google patents status indicator.

The status of a patent is irrelevant to whether it is prior art. For example, a patent that expired may still be used as prior art against the new application. However, the status of the patent (alive or dead) does matter when you are trying to determine whether infringement of the patent is a concern.

Also, just because a patent is indicated as dead, that doesn’t mean that it cannot come back to life. A patent may expire because of a maintenance fee that was not timely paid but then come back to life with the payment of the maintenance fee along with a late fee.