The ETSI declaration database contains over 555,000 declared patents spanning more than 100,000 patent families. In response to court rulings prohibiting enforcement of late-declared patents, many ETSI-declaring companies are adapting their strategies and patents are now often self-declared at the earliest possible stage, frequently including unpublished provisional or priority numbers. 

The self-declaration process introduces significant variability when it comes to what is disclosed – some companies only provide provisional numbers, while others declare application and grant numbers as well as all jurisdictional counterparts. This lack of consistency results in an ambiguous ETSI database, which poses challenges for analysis and interpretation. So-called ‘top-down’ analysis to identify a 4G or 5G patent owner’s share can become inherently biased when based on unprocessed data as it can lead to underrepresenting the portfolio size of companies that share less information than others. 

Over the past decade, patent data providers have developed sophisticated rules for matching, cleaning and normalising patent numbers, enabling them to identify and link many declared patents. However, the complexity of this process has deterred some data providers from fully committing to producing accurate data. As a result, several 5G patent reports have presented conflicting and often confusing findings, creating a distorted view of the true landscape.

For many 5G patent experts, the presence of multiple reports on 5G patent ownership with conflicting rankings is a cause for concern, particularly when reports consider publicly available ETSI-declared patents where all the data originates from the same public source. Both SEP owners and standards implementers – as well as IP lawyers and economists – are asking: 

• Which 4G or 5G patent report can be trusted? 
• Which one serves as the benchmark for the highest data quality? 
• How can we reconcile the differences between these studies? 

While it is not transparent how certain reports conclude their patent shares, public rankings may have an impact on SEP licensing. More cases are incorporating the top-down approach for FRAND determinations, in which a patent owner’s portfolio size is compared to the total patent stack to calculate a 4G or 5G share. Additionally, when assessing the comparability of licence agreements, courts are increasingly relying on declaration data to determine whether an agreement is truly comparable. Reliable 4G and 5G patent data is playing an ever-growing role in SEP licensing and litigation, so declaration data integrity is becoming critical.

Matching various patent number formats

Companies submit patent numbers in various formats and types, leading to inconsistencies in matching ETSI-declared patent families. Figure 1 highlights the number of non-matched ETSI-declared patent families (INPADOC) by declaring company when relying solely on unprocessed patent number declarations. 

Figure 1. The number of non-matched ETSI-declared patent families (INPADOC) based on unprocessed patent number declarations (as of 31 October 2024)

Figure 1 underscores how differences in declaration practices can significantly affect matching performance. Companies like Xiaomi, Lenovo and Oppo provide consistent declarations that disclose all published patent numbers. Notably, companies with a high proportion of provisional or priority numbers in their declarations (eg, Ericsson, Qualcomm, Nokia, Huawei and Intel) show particularly high numbers of non-matched ETSI-declared patent families, when based on unprocessed patent number declarations 

Failing to account for all patent number formats risks underrepresenting the size of certain companies’ patent portfolios. Without a comprehensive matching process, entire patent families may be overlooked, which highlights the importance of rigorous data validation to ensure accurate and equitable 4G and 5G patent analysis.

Identifying missing patent family counterparts

Another major challenge is that the ETSI database does not comprehensively identify all patent family counterparts for declared patents. One key reason is that ETSI identifies patent family counterparts only at the time of declaration. If a patent is declared before its family counterparts are published, the database will not be updated later to include those additional counterparts when they become available in other jurisdictions.

While some companies proactively re-declare family counterparts as they are published, others only declare the initial patent application. This practice leaves many global patent family counterparts unaccounted for in the ETSI database.

Figure 2 highlights this issue, showing that 9,244 granted US patent counterparts are missing from the ETSI database despite the corresponding basis patent being declared. Likewise, a significant number of granted counterparts for Chinese, European, Japanese and Korean patents are also absent. The family expansion considers the ETSI IP rights policy, which stipulates that when a basis patent is declared, its counterparts are also subject to the FRAND commitment. Here, counterparts are determined based on ETSI’s more restrictive simple patent family definition compared to INPADOC.

Relying solely on ETSI-declared patents introduces a bias in jurisdiction-specific patent analysis as it skews the data based on declaration behaviour. This leads to the omission of a substantial number of granted patents, undermining the accuracy and completeness of such analyses. To ensure balanced and reliable rankings, it is essential to account for all family counterparts – regardless of their inclusion in the public ETSI database – so long as one family member is declared.

Identifying patent ownership and corporate tree information

Often, ownership of valuable patents (eg, cellular SEPs) changes as SEPs are sold or acquired If patent ownership data is outdated and such changes are not rigorously tracked in global patent offices, patent owner rankings can become highly inaccurate and misleading.

Figure 3 illustrates the difference between rankings based on cleaned, normalised and up-to-date ultimate owner data – identified through corporate trees – versus those relying on raw ETSI-declared company names. For example, it shows that a significant portion of the Intel portfolio acquired by Apple still appears as being owned by Intel. Companies like Huawei, LG or Samsung, which have sold parts of their portfolios, show inflated patent counts. Conversely, patents may be omitted if owner names are not normalised or subsidiaries are not recognised, as seen with Nokia, Interdigital, CICT, Qualcomm or Lenovo.

Figure 3. The difference in patent counts when comparing cleaned, normalized up-to date corporate tree identified ultimate owner data to raw ETSI declaring company counts (as of 31 October 2024).



In the context of these challenges, LexisNexis Intellectual Property Solutions launched the Cellular Verified declaration data-cleaning initiative with the aim of making more accurate data accessible to IP professionals and allowing for more reliable SEP analytics. This project involved collaboration with more than 30 ETSI-declaring companies, which compared their internal declaration records with the information available on ETSI and the LexisNexis IPlytics database. Given the sheer volume of declared patents, achieving a perfect match rate is nearly impossible without external validation – so feedback from these companies was incorporated into the process. Their feedback was instrumental in refining critical processes, such as improving patent number matching rules, identifying missing family counterparts, removing incorrectly declared patents, verifying global ownership and ensuring accurate classification of patents into specific standard generations (eg, 3G, 4G or 5G).

Figure 4 illustrates the progress of matching and cleaning efforts, showing how the average matching rate improved from t1 (initial matching efforts) at just above 80% to t7 (final application of matching processes) at about 95% to Cellular Verified at 99.9%, when the data was checked and validated by the ETSI-declaring companies. 

Figure 4: The matching performance process after each cleaning step from t1-t7 and after the completion of the Cellular Verified data-cleaning initiative

Mastering SEP licensing challenges with high-quality data 

These databases provide critical insights into 4G and 5G patent ownership, serving as a foundational resource for stakeholders involved in patent licensing negotiations, patent commercialisation and pool formation. To navigate these processes effectively, access to comprehensive data on declared patents, pooled assets and standards contributions is essential. SEP portfolios are inherently dynamic, evolving alongside advancements in patent activity and progression of standards. Understanding and calculating market share for a specific standard (eg, 4G or 5G) requires a detailed analysis based on several key factors. 

4G and 5G patent stack calculation

• The denominator represents the total number of declared, active and granted 5G patents relevant to the standard in question, while the numerator corresponds to the patent owner’s specific 5G declared patent portfolio. Dividing the numerator by the denominator yields the 5G patent holder’s share of the total 5G patent stack, providing a benchmark for comparison within the industry. 

5G patent data validation

Rigorous data cleaning ensures the commitment to only consider cleaned and harmonised patent data to ensure unbiased ranks, despite huge differences in declaration behaviours of different companies. 

Thorough data enhancement ensures the commitment to track worldwide legal status data (eg, active, expired, lapsed, revoked or abandoned) and patent ownership information, as well as corporate tree hierarchies to reflect M&A. 

Data validation ensures the commitment to work directly with more than 30 ETSI-declaring companies to validate that public ETSI patent declaration records match each company’s internal declaration records with 100% accuracy. 

By leveraging these data points, stakeholders can assess the value of patent portfolios available for licensing and prepare for FRAND negotiations. This approach also enables benchmarking SEP licensing offers against comparable agreements.