How and why LinkedIn learned to love patents

In 2012 LinkedIn found itself a potential target for corporate patent asserters. It had revenue reaching nearly $1 billion, with growth of 86%, yet owned only 22 patents. However, this changed fundamentally from 2012 to mid-2016, when LinkedIn grew its organic portfolio from 36 to over 1,000 patent assets and purchased more than 900, dramatically reducing its risk profile.

Corporate-to-corporate patent assertion remains a significant threat for companies, in spite of patent reform and a series of US Supreme Court decisions which have been negative for patent holders. For the past 10 years, non-practising entities (NPEs) – sometimes referred to derisively as patent trolls – have gained media attention because of the cost of and legitimate moral outrage over some of their practices. Corporate patent assertions tend to be more dangerous and more expensive than NPE activity and involve a broader category of business solutions. Instances of corporate patent assertions abound: examples include those involving IBM, Qualcomm, British Telecom, Alcatel-Lucent and more recently BlackBerry, all of which have patent licensing teams. A critical element to successfully defending yourself against such assertions is to have a viable counter-assertion patent portfolio. However, this is something that many fast-growing start-ups lack.

Start-ups are the driving force behind many technological advancements and, understandably, tend to focus on building a product and validating a business model. Because of this, patent filing rates for start-ups in the high-tech market (eg cloud computing, semiconductors, mobile and networking) are seldom commensurate with their inventions and future needs, especially when compared to those of large established companies – typically, start-ups file a few patent applications in core technology areas only. But what happens when a start-up becomes successful? Combine a high-growth start-up with a successful initial public offering and a small patent portfolio can make it a target for patent assertions from other companies. Established companies can face similar problems when entering an area in which they have not traditionally filed patents or expanding into international markets (eg, a Chinese company entering the US and European markets).

This article focuses primarily on high-tech companies which are vulnerable to patent assertions from operating companies, as opposed to NPEs. LinkedIn provides an example of how a company can proactively meet this challenge following its creation of a strategy which combines organic patent filings and patent acquisitions, and which has enabled it to quickly grow its patent portfolio and reduce much of the risk of assertions from existing companies.

Figure 1. Patent ecosystem

Corporate patent assertion risk and counter-assertion defence

The risk of assertion to LinkedIn comes from operating companies in its direct market ecosystem, those in the extended ecosystem and large corporate patent asserters with large patent portfolios. The market ecosystem includes direct competitors, as well as all of the companies with which LinkedIn works. Each of these operates in a technology space similar to that of LinkedIn and therefore may have patents which could affect its revenue. Because of close business relationships, the vast majority of these companies pose little patent risk – disputes rarely arise because both companies have vested business relationships with one another and with other members of the ecosystem. LinkedIn also benefits from an open source culture and participation in that community – patent disputes tend to occur far less frequently in open source projects.

However, each of these companies has its own ecosystem of competitors and other companies with which it works, thereby extending the ecosystem. The further removed companies are from LinkedIn, the less technology overlap there is and the lower the risk that a company will have patents which could be asserted against LinkedIn. The exception to this rule is large corporate asserters. These can be found almost anywhere in an ecosystem and have large patent portfolios, as well as a history of assertion (for more detailed analysis on the patent ecosystem and determining the risk from each company, see “The strategic counter-assertion model for patent portfolio RoI”, IAM issue 72).

An operating company (Company A) may choose to assert a patent against LinkedIn with one or more of the following goals:

• Revenue generation – obtain licence fees from LinkedIn for using Company A’s patented technology.
• Create freedom to operate – obtain licence rights to use LinkedIn’s current or future patented technology.
• Strategic goals or business interference – interfere with LinkedIn’s ability to offer products and services using Company A’s patent (eg, force a partnership, gain application programming interface or technology access, obtain injunctions, delay market entry or effect product redesign).

In a typical high-tech assertion between Company A and LinkedIn, Company A would be asserting primarily to achieve some combination of the first and second goals above, although it might also seek to achieve the third goal as a possible added benefit. To accomplish this, it would allege that LinkedIn’s products infringed on its patents, showing the affected revenue (ie, how much money LinkedIn was making from the allegedly infringing products).

How to deter patent asserters through counter-assertion

When it comes to patent assertions, the best defence is a good offence. Most high-tech patent disputes are handled under a Cold War-style détente strategy: “Sue me for patent infringement and I will sue you back.”

While there are some high-profile exceptions, such as cases between Apple Inc and Samsung Electronics Co Ltd, usually the prospect of escalating litigation over cross-infringement complaints creates a stable environment in which companies tend to leave each other alone. For example, Juniper Networks, Inc and Cisco Systems, Inc have never sued one another for patent infringement despite operating in technology areas so similar that they have been co-defendants in patent assertions (eg, TXED 6-09-cv-00324 and TXED 2-08-cv-00304). A robust patent portfolio is critical to mitigating corporate assertion risk. Like a Cold War arms race, simply having a large portfolio can be sufficient deterrent against assertion.

In addition, LinkedIn’s culture – as well as that of many Silicon Valley companies – tends to focus competition on product success, rather than mere patents. However, there are exceptions. For example, when companies begin to fall behind their competitors or a single company in a new technology space decides to aggressively enforce its patents, the détente begins to fail.

Figure 2. Counter-assertion negotiation

Returning to the Company A example, LinkedIn would deal with its allegations by counter-asserting (see Figure 2), highlighting which of Company A’s products and services infringed its own patents. Where Company A’s infringing products generate large revenue, LinkedIn would be in a stronger negotiating position.

The challenge for LinkedIn is to have a patent portfolio that is sufficiently large and diverse to ensure that it can find patents that Company A – and any other company that poses a high risk of asserting – might be infringing. Just as a large portfolio acts as a deterrent, a small portfolio demonstrates vulnerability. At the end of 2012, LinkedIn needed not only to start filing at an increased rate, but also to address its patent deficit through acquisitions – a practice known as ‘backfilling’.

LinkedIn background and patent strategy

LinkedIn launched the first professional social network in May 2003. Over time, it has become highly successful: it increased its revenue in 2006 by 723% to $9.8 million and then by an average of 128% each year from 2007 to 2011 (LinkedIn Securities and Exchange Commission (SEC) Form S-1 and 2013 SEC Form 10-K). The company went public in May 2011 and by 2012 was making around $1 billion in annual revenue and continuing to grow rapidly. Revenue had increased 86% over the previous year, R&D spend had risen by 95%, and headcount was up 63% (LinkedIn 2013 SEC Form 10-K). However, at the same time, LinkedIn had only one issued US patent and 35 pending applications. Company executives recognised the exposure to corporate patent assertions, but had held off dealing with the challenge until the business had proven itself. In 2012, LinkedIn hired a small internal patent team and began working with Richardson Oliver Law Group to come up with a mitigation plan.

The opportunities for risk mitigation can be divided into two categories: increasing organic filings to address future assertion risk and patent acquisition to address present and near future risk.

Future organic filings

To determine a target patent filing rate, LinkedIn reviewed the goals and filing practices of other high-tech companies. A more sophisticated return on investment (ROI) analysis was not carried out at the time because it already knew it was behind. The patent strategy team first looked at about 25 high-tech companies, analysing their filing rates and R&D spend. R&D spend is a good proxy for how much innovation is taking place, as well as for future revenue expectations – which corporate patent asserters will target. We analysed the companies to determine an imputed patent strategy and divided them into different bands based on the patent programme’s goals (Figure 3):

• no defined strategy;
• product coverage to deter copying;
• obtaining freedom to operate; and
• outbound licensing.

Figure 3. Target patent filings

We set our goal as obtaining freedom to operate; we wanted to be able to continue to deliver products and services to our growing customer base by relying on our core technologies, while minimising patent risk from larger patent holders. We thus set our target filing rate in the middle of that band – 0.25 patents per $1 million in R&D spend. As illustrated in Figure 4, LinkedIn’s filing rate had been below its target rate before 2001. Had it remained unchanged, the projected organic filings in 2016 would have missed the target for freedom to operate by almost 300 patents. Arguably, LinkedIn could have used a higher filing rate for the earlier years, as those years tend to produce some of the most fundamental patents. However, even without this, it needed to catch up.

Figure 4. Actual versus target portfolio growth projection

How LinkedIn increased patent filings

LinkedIn increased its organic patent filings to 0.42 filings per $1 million R&D in 2015. While this was above the 0.25 target, these additional filings helped to close the gap created by previous low rates. In order to increase the filing rate, LinkedIn needed to fundamentally shift its patent culture. Timetables and goals were discussed and set, and several targeted projects were launched.

First, invention harvesting sessions were held on a regular basis. Minimising the impact on inventors’ time and attention was critical to the success of this programme, so we gathered small groups of engineers together and reviewed their current work, captured the ideas they put forward and selected those that met the filing criteria.

Further, we benchmarked inventor incentives to better align with other social media companies.

We added a highly visible recognition component to our incentive programme, which included inventor appreciation events (eg, private movie premieres were very successful), programme-branded clothing and patent cube awards – all of which were designed to reward inventors and promote the programme.

In addition, we carried out education as to the size of the patent deficit problem and the plan, obtaining buy-in from the general counsel, the engineering and product teams, the finance department and the CEO.

With the increased filing rate came a significantly increased budget. A well-defined plan helped to clear the budget request.

Figure 5 shows the results of the efforts on filings. From 2012 to 2016, our filings increased from 20, 80, 120, 200, 300 and 330, to match the target filing rate and to compensate for LinkedIn’s high revenue and R&D growth.

Figure 5. Filings per $million of R&D spending

LinkedIn also needed to build out its internal patent team. It started by leveraging outside firms, but faced the problem of a cold start. A successful patent programme needs to be integrated into a company’s culture and business, and it needed to invest. Therefore, LinkedIn added its first full-time patent attorney, Pierre Keeley, in 2012, followed by an experienced patent paralegal, Grace Forker. As soon as additional headcount was approved, we added another key patent attorney, Puneet Sarna, from Dolby Labs – and supplemented the team with the secondment of a junior patent attorney. LinkedIn also needed to add to its outside counsel strength, so we brought in an IP strategy firm (Richardson Oliver Law Group) and another patent prosecution firm. Currently, our broad patent team includes our vice president of intellectual property, product and privacy, Sara Harrington; two patent attorneys; one patent paralegal; and a patent litigator. To handle the large influx of invention disclosures, we upgraded our tracking system for invention capture, patent application tracking and tracking metadata associated with our overall portfolio.

Having the executive buy-in and operating capacity to handle the additional inventions did not mean that we would get the additional invention disclosures. As mentioned above, the invention harvesting sessions proved critical to our success. We adopted a white-glove service approach to minimise the impact on inventors. We identified the team members needed for a harvesting session and then lowered the preparation time for inventors to close to zero – the main thing was that they show up to discuss their projects.

Typically, between five and 10 people from a single project would gather and discuss their project for an hour, resulting in between five and 15 invention disclosures. Projects would typically run for between two and six months before a session was held.

Inventors were a little nervous before the invention harvesting sessions, but ultimately the sessions were seen as successful. The patent team recognised that sessions needed to be positive for inventors – this included clarity about how LinkedIn was going to use its patent portfolio. After one session, a participant commented: “We didn’t know how the meeting was going to go. We thought we were going to get beat down by lawyers. Actually, the session was really positive. The patents are going to allow us to continue to operate the open source project.”

LinkedIn also needed to address a culture in which some inventors viewed patents – and software patents in particular – negatively. Within the company, we evangelised about defensive use of patents and the way that our founders had used them to help the broader community. For example, our founders were concerned that a foundational social media patent would fall into the hands of an NPE or someone that might harm emerging social media companies, so they bought it. We also educated inventors on how to apply for patents, what the benefits are and who to contact. We built an internal web page integrating inventor education and inventor recognition, and identified R&D team members who supported the goals, harnessing their advocacy and enthusiasm. We presented to the company in an all-hands meeting, where we gave out lots of t-shirts. We were complimentary about an engineering culture which believes in innovation and craftsmanship and wants to market its engineering brand by supporting strategic open source projects, as well as on the professional accomplishment of being recognised for innovation by adding a patent to your profile.

A few simple ideas worked well to keep the patent programme in the minds of inventors, such as distributing small branded giveaways across the entire company, focusing on items that inventors would keep on their desks.

Although cash awards for our inventors are an integral part of our incentive programme (ours are competitive), we believe that the combination of the white-glove service model and public recognition has had a higher overall impact on our success.

Looking to the future, our organic filing rate is likely to slow. However, we have been able to catch up on some of our lower filing rates and supplement our portfolio through strategic purchases. In June 2016 Microsoft announced the purchase of LinkedIn, so our plans will adapt to the new overall strategy after the purchase closes.

Assertion risk mitigation opportunity – patent acquisition

While the increase in LinkedIn’s filings helped to grow the total patent portfolio, challenges remain. First, while organic filings tend to focus on LinkedIn’s core technology and therefore help a great deal with counter-assertion against potential competitors, they are less helpful when it comes to large corporate asserters further outside LinkedIn’s core technology area. Second, the priority dates on all the new filings are recent (after 2011). Earlier priority dates (old inventions) help the most in counter-assertion, but LinkedIn would have had to file for those patents in the 2000s. Fortunately, the market for buying and selling patents is robust and allows companies to fill in where they have weakness in their portfolios. Focused patent buying allowed us to build a counter-assertion portfolio to help bolster any negotiations.

In our experience, the assets most typically used in assertion and counter-assertion negotiations range in age from eight to 16 years from the patent’s priority date. This is also supported by research (eg, Brian Love, “An Empirical Study of Patent Litigation Timing”, University of Pennsylvania Law Review, Vol 161, p 1309 (2013); Mark Lemley, John Allison and David Schwartz, “Understanding the Realities of Modern Patent Litigation”, 92 Texas Law Review 1769 (2014)). Older patents have a chance to be fundamental and pioneering to the industry, as well as to be used by many companies, including operating companies which assert patents. This is not to say that all older patents are useful for counter-assertions, as most cover technologies which were never widely adopted or which are unimportant for industry. Patents that are between eight and 16 years old are old enough to be adopted by corporate patent asserters, yet still have enough remaining life to be useful for a few years.

The importance of older patents is the reason that LinkedIn cannot mitigate today’s patent assertion risk with organic patent filings alone. Due to the eight-year lag from priority date, LinkedIn’s 2012 increases in filings will likely start to make a significant impact in counter-assertion negotiations in 2020.

LinkedIn’s phenomenal growth created another challenge: even if it had filed at a rate of 0.25 patents per $1 million spent on R&D, it would not have had sufficient patents to counter an assertion. Referring to Figure 6, if LinkedIn had filed at the target rate through 2008, the likely number of useful assets in its portfolio would still be vastly outmatched by those of an established corporate asserter. In an assertion, the asserter would likely try to map patents from 2004 to 2008 to LinkedIn’s 2016 revenue, but its organic target filing rate would not yet have created the necessary pool of older patents for potential mapping to the asserter. Additionally, as illustrated in Figure 7, generally only between 1% and 5% of patents in a portfolio are high value in the context of a patent assertion negotiation (see Suzanne Harrison et al, Edison in the Boardroom Revisited, 2011). With so few filings, the chances that we would find sufficient gems in those early filings were very low.

Figure 6. Organic portfolio growth by priority year

Figure 7. Value in context

Source: Adapted from Edison in the Boardroom by Suzanne Harrison

Unlike organic patent portfolio development, where you do not know which patents will be important, you can benefit from hindsight when creating a targeted patent buying programme. During a patent assertion negotiation, the important factor to consider is the value of each patent portfolio in context. Any patent in an asserting company’s portfolio which is outside LinkedIn’s technology area or which does not read on its products and revenue has a negligible value in the negotiation. For companies, this value distribution across their patent portfolios in any given negotiating context is log-normally distributed (see Figure 7).

Figure 8. Hindsight in targeted buying

Unlike the asserting company, LinkedIn knows what technologies were adopted by likely asserting companies and from which technologies those companies derive the most revenue. By using this knowledge to focus the buying parameters, LinkedIn can achieve the same value in a counter-assertion as the asserting company, while requiring far fewer assets. The blue areas in Figure 8 are where the majority of value is perceived and exchanged during these patent licence negotiations – LinkedIn’s patent buying allowed us to equalise these areas.

Patent acquisition process

LinkedIn’s targeted buying was focused on backfilling assets with early priority dates for counter-assertion against known corporate patent asserters. Additionally, targeted buying was and still is useful to develop areas in which LinkedIn is not developing organic technology and patents.

The first step in targeted buying is to identify the companies of concern. In the case of LinkedIn, we analysed the ecosystem to determine companies which might be of concern. Then we reviewed each of these on the basis of past assertion actions, financial strength and technological overlap with LinkedIn – as well as its current and expected business relationship with LinkedIn. In many cases, good business relationships mitigate much of the patent risk and minimise the level of concern. Then we reviewed the potential applicability of any current organic assets – although in 2012, this was an extremely short list. From there, an ROI analysis was carried out to determine whether the expected value of the risk reduction was worth the cost of patent acquisitions. Sometimes it makes business sense to keep a low level of risk rather than spend an exorbitant amount to remove it. Finally, we completed a revenue analysis for each company to determine the products and technology areas of greatest value to it. Once these technology areas were determined, we created buying criteria (see the first column of Table 1 for example companies and their projected risk). Richardson Oliver Law Group then monitored the brokered patent market, as well as privately sourced packages fitting LinkedIn’s buying criteria (eg, technologies, pricing, priority dates, existence of claim charts (evidence of a company, product or industry’s use of the patented technology)) (for further detail on the ROI analysis for buying, as well as methods for reducing the total spend, see “The strategic counter-assertion model for patent portfolio RoI”, IAM issue 72).

Figure 9. Targeted buying process flow

Patents are often sold by brokers and owners in small packages. The seller supplies a narrative around the value proposition of the package, asset lists, applicable market information, adoption by the market and pricing information (see “The brokered patent market in 2015 – driving off a cliff or just a detour?”, IAM issue 75). We found interesting patent packages on the brokered market, privately by directly approaching other companies and through our corporate development department.

As part of the buying programme, from 2012 to 2016 LinkedIn reviewed over 800 patent packages containing over 25,000 patent assets. Ultimately, 13 packages were purchased (equivalent to around 2%), which appears to be a typical rate for corporate purchasers.

If a package was found to meet LinkedIn’s buying criteria, it was reviewed in greater detail to determine its value in the context of a counter-assertion with a particular company of concern or in the context of general backfilling within LinkedIn’s core technology areas.

Table 1. Prepared counter-assertion strategy

Patent acquisition results

LinkedIn’s patent buying programme has substantially improved our readiness against 11 companies of concern (Table 1). LinkedIn now has a playbook response, discussed further below, for each of the 10 identified companies, including:

• creating evidence of use (EOU) materials – a full response to an assertion has been generated;
• identifying key patents to use in a counter-assertion – patents which will drive value in counter-assertion negotiations have been singled out and EOU materials created where necessary; and
• developing business relationships – the planned response is to leverage existing business relationships and the organic portfolio to mitigate patent risk. Additionally, existing acquisitions may be helpful, but no additional patent acquisitions are required.

Now that counter-assertion positions have been established, LinkedIn’s buying will likely slow, although targeted buying will remain an extremely useful tool for mitigating risk from new companies of concern outside LinkedIn’s core technology area.

Crucially, there are more patent risks than are presented here or directly addressed by the patent acquisition programme, which focuses on reducing the primary sources of risk. This is not to say that the purchased assets are not valuable against other sources. Because we focused on fundamentally enabling technologies, we can mitigate patent risk from other companies and we did so.

To date, we have purchased more than 900 patent assets across 13 deals (see Table 2). Most of the purchases are in LinkedIn’s core business area of social networking, although there is considerable diversity within those acquisitions (eg, technologies covered include video sharing, new contact identification and messaging). Other acquisitions were outside LinkedIn’s core (eg, telecommunications). Generally, we ensured that our acquisitions favoured smaller deals where the value driver patents were clearly identified. We bought from both the brokered market and directly from the owners. Finally, we have had some of our acquisitions reported in the news and recommend that you consider how you might respond when deals are reported.

We participated in a private study comparing over 100 patent acquisitions across multiple companies and also evaluated other purchase data. We found that we paid market rates (slightly below the average) for our acquisitions. Having access to a source of patent pricing data helps immensely when trying to price acquisitions.

The actual acquisition process was relatively straightforward. Deals usually closed within a few weeks of making the decision to acquire. We rarely found ourselves in pricing wars – if we thought that a package would end up having its price inflated, we would look elsewhere. Given that so many assets are available in the market, there is always another deal available tomorrow.

Prepared counter-assertion strategy

To mitigate the identified risks, we needed to know how well we were doing in executing this strategy and be prepared to use our patents for defence. If the new assets are not organised and are not easily identified when needed, the portfolio’s value is greatly diminished. We created a tracking system for LinkedIn’s patent assets. The assets are reviewed manually for applicability to companies of concern and re-evaluated when a new company is identified. Because we know how the patents apply to specific companies and technologies, we also understand how our portfolio might apply to other potential patent asserters.

We created playbooks for each company of concern. Each contains the list of patents applicable to that company and specifies which claimed inventions are used in which product. Additional information – such as business relationships which can be leveraged in a negotiation – can also be tracked. If a company of concern approaches LinkedIn, our legal team is ready to respond with vetted patents for counter-assertion.

The required number of patents in a given playbook varies based on both the size of the company of concern and our exposure to it. The general goal of a playbook is to shift the licensing amount purportedly owed by LinkedIn by $20 million to $200 million in our favour. In order to achieve this, we have found that a good playbook should contain between three and 10 patent families, with EOU for key patents. The goal of each playbook is to show infringement by the asserter’s products and services exceeding $1 billion revenue. We set specific goals for each one and tested its contents against them.

To help ensure a robust defensive portfolio, LinkedIn included diverse technologies across the acquisitions. This allows some flexibility in case a new company begins asserting patents or an asserter stops producing an infringing product.

Concrete, successful results

The results of LinkedIn’s patent strategy programme are clear. The company now owns a patent portfolio which allows it to mitigate risk from operating company patent assertions. The portfolio covers a range of technologies and has a range of priority years. As a result, LinkedIn has been able to use the purchased patents effectively to address inbound licensing assertions. Both the large and smaller purchases have proven valuable in discussions with corporate asserters. Although we cannot share specifics, we are confident that these purchases have had a positive ROI.

Figure 10 shows how the acquired portfolio supplements the organic portfolio to backfill for earlier priority years. LinkedIn now has many assets with a priority date well before the increase in organic filings and even includes some inventions from before the company was founded. This broad spread of technology and priority dates would be impossible without a patent strategy which integrates organic patent filings and patent acquisitions.

Figure 10. Total portfolio size by priority year