Benchmark – Product Mapping

Maersk logo
Maersk logo

×

Stilta logo
Stilta logo

September 2026

September 2026

From patent to portfolio analysis: a benchmark on product mapping

Stilta mastered the analysis of a single patent: litigation-grade and evidence-backed. That same standard is now scaled to portfolio-level. Together with Maersk’s IP team, Stilta benchmarked its new capabilities against a subset of Maersk’s well documented patent portfolio, linked to their products

Stilta mastered the analysis of a single patent: litigation-grade and evidence-backed. That same standard is now scaled to portfolio-level. Together with Maersk’s IP team, Stilta benchmarked its new capabilities against a subset of Maersk’s well documented patent portfolio, linked to their products

Stilta mastered the analysis of a single patent: litigation-grade and evidence-backed. That same standard is now scaled to portfolio-level. Together with Maersk’s IP team, Stilta benchmarked its new capabilities against a subset of Maersk’s well documented patent portfolio, linked to their products

0%
correct against Maersk’s own product labels
0+
limitation-evidence pairs analyzed
Q3 26
portfolio analysis launches in beta

Background

Great products are built in steps. Over the past year Stilta has focused on doing one thing exceptionally well: analysing a single patent. Every claim is broken into its individual limitations, and each limitation is matched to real, cited, public evidence.

Next, Stilta’s natural next step is to scale this to the portfolio: enriching and mining across a whole portfolio to infer what matters, and how it connects. Starting with infringement (patent-to-product analysis). To ensure quality, this is done with a staged approach. First, evaluate the capabilities on a company’s patents against its own products. From there, naturally extending it to include competitor products. This benchmark covers that first step

Great products are built in steps. Over the past year Stilta has focused on doing one thing exceptionally well: analysing a single patent. Every claim is broken into its individual limitations, and each limitation is matched to real, cited, public evidence.

Next, Stilta’s natural next step is to scale this to the portfolio: enriching and mining across a whole portfolio to infer what matters, and how it connects. Starting with infringement (patent-to-product analysis). To ensure quality, this is done with a staged approach. First, evaluate the capabilities on a company’s patents against its own products. From there, naturally extending it to include competitor products. This benchmark covers that first step

Great products are built in steps. Over the past year Stilta has focused on doing one thing exceptionally well: analysing a single patent. Every claim is broken into its individual limitations, and each limitation is matched to real, cited, public evidence.

Next, Stilta’s natural next step is to scale this to the portfolio: enriching and mining across a whole portfolio to infer what matters, and how it connects. Starting with infringement (patent-to-product analysis). To ensure quality, this is done with a staged approach. First, evaluate the capabilities on a company’s patents against its own products. From there, naturally extending it to include competitor products. This benchmark covers that first step

The benchmark

Stilta’s new portfolio capabilities were tested together with Maersk, a team of leading patent practitioners with a well-understood, business-linked patent portfolio that had already been mapped to products by the practitioners. With only patent IDs and identified public evidence, Stilta operated at the patent limitation level, surfacing mappings to Maersk’s own products and services based on how completely the found evidence addressed each claim. Results were compared against Maersk’s own mappings.

Benchmark results
88%143 of 163 patents
of Stilta’s mappings, derived from public evidence, were correct against Maersk’s own product labels
97%30 of 31
where every limitation was taught by the public evidence found
matched Maersk’s own product label
  • Scope: 500+ patents across 150+ invention families, with 2,000+ claim limitations

  • Product match: Stilta found a limitation-based product match for 163 patents upon analyzing 10,000+ limitation-evidence pairs

  • Overall agreement: 88% of those mappings were correct against Maersk’s own product labels (143 of 163 patents)1

  • Complete evidence mappings: 31 patents had a full, public-evidence-based limitation mapping (i.e., all patent limitations were taught by the found evidence); in that subsample, 97% mapped correctly to the products

  • Scope: 500+ patents across 150+ invention families, with 2,000+ claim limitations

  • Product match: Stilta found a limitation-based product match for 163 patents upon analyzing 10,000+ limitation-evidence pairs

  • Overall agreement: 88% of those mappings were correct against Maersk’s own product labels (143 of 163 patents)1

  • Complete evidence mappings: 31 patents had a full, public-evidence-based limitation mapping (i.e., all patent limitations were taught by the found evidence); in that subsample, 97% mapped correctly to the products

  • Scope: 500+ patents across 150+ invention families, with 2,000+ claim limitations

  • Product match: Stilta found a limitation-based product match for 163 patents upon analyzing 10,000+ limitation-evidence pairs

  • Overall agreement: 88% of those mappings were correct against Maersk’s own product labels (143 of 163 patents)1

  • Complete evidence mappings: 31 patents had a full, public-evidence-based limitation mapping (i.e., all patent limitations were taught by the found evidence); in that subsample, 97% mapped correctly to the products

1 The comparable set is smaller than the full 500+ patents scoped as it excludes products where Stilta could not find public evidence to map against.

1 The comparable set is smaller than the full 500+ patents scoped as it excludes products where Stilta could not find public evidence to map against.

1 The comparable set is smaller than the full 500+ patents scoped as it excludes products where Stilta could not find public evidence to map against.

What’s next

Portfolio analysis is the next capability on Stilta’s platform. We’re starting with portfolio-wide infringement (and product-patent work), running on the same evidence-first engine as the single-patent work. Following this success on an internal-product benchmark, the capability will launch in beta for users during Q3. Subsequently, a pilot targeting external products (infringement) kicked off in August, again with an enterprise holding a leading portfolio.

Portfolio analysis is the next capability on Stilta’s platform. We’re starting with portfolio-wide infringement (and product-patent work), running on the same evidence-first engine as the single-patent work. Following this success on an internal-product benchmark, the capability will launch in beta for users during Q3. Subsequently, a pilot targeting external products (infringement) kicked off in August, again with an enterprise holding a leading portfolio.

Portfolio analysis is the next capability on Stilta’s platform. We’re starting with portfolio-wide infringement (and product-patent work), running on the same evidence-first engine as the single-patent work. Following this success on an internal-product benchmark, the capability will launch in beta for users during Q3. Subsequently, a pilot targeting external products (infringement) kicked off in August, again with an enterprise holding a leading portfolio.

Get started with Stilta

Evidence-first patent analysis that builds the strongest case. Book a demo to see it on your own work.

Book a Demo

Input. Only patent IDs from the portfolio were provided as input. For evaluation, Maersk’s internal data on product-patent mappings were used.

Evidence. Product evidence was drawn entirely from public sources gathered via web search. No internal or confidential material was used except for the patent-product mappings used for evaluation.

Mapping. Each patent was decomposed into its claim limitations. Candidate products were identified from public data and scored on how completely they cover the claim limitations, and the strongest was retained. For some patents the mapping was full, with every limitation covered by a documented read; for most, only a share of limitations carried a public-evidence citation.

Comparison. Stilta's inferred product was compared against Maersk's own product label, together with the patent team.

Join the waiting list

Join the waiting list

Get portfolio analysis updates and early-access news in your inbox.