A claim chart is a mapping: asserted claim on the left, accused instrumentality on the right. That much everyone reading this knows.
What is less often acknowledged is how the document gets made. In most engagements, a chart is scoped as a deliverable, with a fixed page count, a delivery date, a review pass and a PDF. It is treated as documentation of an infringement theory that already exists in someone’s head.
That is the wrong frame, and the gap it creates is where cases quietly weaken. A chart is an exhibit in waiting. It is the first draft of what an expert will defend under oath, what a Rule 12 motion will attack, what opposing counsel will mine for admissions, and what a licensee’s engineering team will read looking for the seam. Nobody on the other side will read it charitably. Almost everyone on your side does.
The difference between a chart that was produced and a chart that was built to be attacked shows up months later, and by then it is expensive to fix.
Key Takeaways
- Claim limitation mapping has to reach the implementation level. Marketing copy describes outcomes; claims recite mechanisms. A cell citing a product page instead of an API call or register field has not mapped the limitation.
- Every broad cell in an infringement chart is a free invalidity contention: what reads on the product, if wide enough, also reads on prior art.
- Evidence without reasoning is as weak as reasoning without evidence. The analyst’s connective work is the product.
- Means-plus-function terms require a two-step mapping: corresponding structure from the specification first, then equivalent structure in the accused product. Skipping step one is an unanswered question.
- The adversarial desk test, writing the best non-infringement argument against your own chart, is where quality shows up before litigation forces it.
Where Patent Claim Analysis Fails: Seven Failure Modes
Marketing Copy Standing in for Implementation
This is the single most common defect in claim limitation analysis. A claim requires determining a routing path based on a stored user profile; the chart cites a product page promising “personalized delivery.” Courts have seen enough of this to name it. Contentions built on generic marketing literature and screenshots have been rejected as insufficient, because advertising material does not give notice of a specific theory of why the product infringes. Marketing describes outcomes. Claims recite mechanisms.
Assertion Where Evidence Should Be
The cell reads “the accused system caches the query result,” and nothing follows. The mirror-image failure is just as bad: evidence dumped without connective tissue. In MasterObjects v. Amazon, the court found that the patentee opened by parroting the claim language and then supplied eighteen pages of document screenshots with no supporting commentary. Volume is not proof. The analyst’s reasoning, showing that this field, in this call, performs this claimed step, is the product.
Parroting the Claim Language
You can spot this in ten seconds: the right column uses the claim’s vocabulary instead of the product’s. A chart that maps “a data processing module” to “the accused data processing module” has done nothing. Contentions that merely repeat claim language fail the local patent rules precisely because they avoid committing to named technology in the accused system. Analysts sometimes do this deliberately to preserve flexibility. It reads as evasion, because it is.
Blurring Functional Limitations
A product that does something adjacent to the claimed function has not met it. If the claim requires a determination performed before transmission, by a specified entity, using a stated input, each of those is a separate proof obligation. Method claims add another: all steps must be attributable to a single actor absent direction or control (Akamai v. Limelight). Charts that blur a multi-party system into one undifferentiated “accused system” are hiding a divided-infringement problem, not solving it.
Means-Plus-Function Charted Without Structure
After Williamson v. Citrix, terms like module, unit, or engine readily invoke § 112(f). The statute limits such claims to the corresponding structure described in the specification and its equivalents, and inadequate disclosure renders the claim indefinite. That makes the charting obligation two-step: identify the corresponding structure or algorithm in the specification, then show the accused implementation has identical or equivalent structure. Skipping step one is an unanswered question that will be asked later. The PTAB has shown the same failure in mirror image: a petitioner that identified no corresponding structure in the specification could not identify anticipating structure in the prior art either.
No Traceability
A screenshot with no URL, no capture date and no archived copy. A citation to firmware with no build number. A specification page with no version. Two years later the page has changed, the app has shipped six releases, and no one can authenticate the exhibit. The patent evidence mapping may have been right. It is now unprovable.
The Mapping That Proves Too Much
This is the most under-appreciated failure in patent infringement mapping. A limitation stretched broadly enough to capture the accused product is, by construction, broad enough to read on prior art. The maxim from Peters v. Active Manufacturing is old and still load-bearing: that which would infringe, if later, anticipates, if earlier. Every over-broad cell in an infringement chart is a free invalidity contention drafted by the patent owner, and a competent defendant will hand it straight to their IPR counsel.
What Separates a Claim Chart That Holds: Patent Infringement Evidence That Works
It maps element by element in the accused system’s own nomenclature (API names, register fields, message types, configuration parameters) so a reader can verify the mapping rather than accept it. Where a term is likely to be disputed, the chart states the construction it assumes and charts to it, rather than leaving the assumption implicit and discoverable.
It respects an evidence hierarchy, shown in Figure 2. Source code and binaries sit at the top; then protocol captures, teardowns, bench measurement and standards cross-references; then developer documentation, API references and regulatory filings; then, last and weakest, marketing. Good practice pushes every element as far up that stack as the pre-suit posture allows, and says plainly which elements are currently supported only from below.
For many elements, public documentation simply stops before the claim does. On-device processing, cache behaviour, timing relationships, silicon-level structure and proprietary protocol internals are not documented anywhere a search will reach. This is where reverse engineering and product testing earn their cost: packet traces for network limitations, decompiled application analysis for on-device steps, instrumented bench testing for timing and threshold claims, and teardowns and die-level analysis for hardware structure. It converts an assertion into a result someone can reproduce. For the source code side of that work, see Patent Infringement Analysis Through Source Code Forensics.
It is traceable at the cell level: source, version, build, capture file, retrieval date, archived copy. An exhibit you cannot authenticate is an exhibit you do not have.
And it is drafted adversarially. The practical test is to write the best non-infringement argument against your own mapping for every element and see whether the cell still stands. A second analyst who did not build the chart should attempt to break each cell before it leaves the building. Figure 3 sets out that loop.
Claim Limitation Mapping: How Quality Converts Into Outcome
Pre-Suit: Rule 11 and the Sanctions Record
Rule 11 requires the comparison to happen before filing. In Judin v. United States, the Federal Circuit found a clear violation where the plaintiff and counsel observed the accused device only from a distance and never tried to obtain one for closer examination, and held that post-filing work could not cure it. Q-Pharma and View Engineering set the same floor: apply the claims to the accused device before suit. A chart with an implementation-level foundation is also your sanctions record.
Pleading
Charts attached to complaints now get read on the merits. Courts have dismissed complaints where allegations merely parroted claim language without showing how the accused products plausibly practice the claimed steps, even when packaged in a claim chart.
Licensing
A chart that cites a target’s own engineering documentation, its own field names and its own released build reaches the engineer the licensing team forwards it to. That engineer either recognises their system or does not, and the answer sets the negotiation. A chart built on datasheets invites a two-line rejection at no cost to the recipient.
Claim Construction and IPR Defence
Your mapping is a commitment. Defendants read infringement contentions as proposed constructions and test that breadth against prior art. Charts written with the § 102 and § 103 consequences in view keep each limitation only as wide as it needs to be.
Expert Defensibility Under Amended Rule 702
Check out our Claim Charts Services
This is where chart quality becomes visible to a judge. Amended Rule 702, effective December 2023, places the burden on the proponent to establish admissibility by a preponderance and requires that the opinion reflect a reliable application of the expert’s methods to the facts. The amendment is expected to drive more exclusion motions in patent cases. Courts have rejected expert analysis as conclusory even where it proceeded claim by claim, because the cited material did not address a feature the claims required. An expert who inherits an unsupported chart is being asked to certify someone else’s assumptions.
Five Things to Check on the Next Chart That Lands on Your Desk
- Read only the three hardest limitations. Is there cited evidence in those cells, or a restatement of the claim?
- Check the vocabulary. If the right column uses the claim’s words instead of the product’s, the mapping has not been made.
- Trace one citation to source. Can you reach that document, at that version, today?
- Find every functional or “module”-style term. Was corresponding structure identified, in the specification and in the product?
- Ask what else the broadest cell reads on. If it reads on prior art, you have written the defendant’s petition.
If a chart clears all five, it was built as evidence. If it does not, the case is carrying the difference.
How Lumenci Builds Claim Limitation Mapping for Patent Owners and Defendants
The seven failures above share a root cause: the chart was built to fill cells, not to survive someone trying to break them. Most of Lumenci’s claim chart work is for patent owners, where the chart has to hold. For defendants, the same standards show exactly where an opposing chart will give way.
- Every limitation is mapped to the accused product’s own nomenclature. API names, register fields, message types, configuration parameters. If the right column reads like the left column, the cell goes back.
- Where public documentation stops short of what a claim requires, our reverse engineering and product testing teams close the gap with packet traces, decompiled application analysis, instrumented bench testing, or die-level teardowns. An assertion backed by a reproducible test result is a different exhibit from one backed by a datasheet.
- A second analyst who did not build the chart runs the adversarial pass before it ships. Their job is to write the strongest non-infringement argument against each cell and see whether the mapping holds.
- Every cell carries a versioned, dated, archived citation. Two years from now, when the page has changed and the app has shipped six releases, the exhibit still authenticates.
- Scope is checked against prior art at the cell level. The Peters problem, where a broad mapping hands the defendant a free invalidity contention, is caught in review rather than in an IPR petition.
- For defendants, the seven failure modes become the review checklist for a plaintiff’s infringement contentions. We test every cell for marketing-only evidence, parroted claim language and missing § 112(f) structure, and flag mappings broad enough to read on prior art. Those cells go straight to prior art search for invalidity.
Whichever side we are on, our expert testimony practice treats the finished chart as the evidentiary base a testifying expert will defend line by line at deposition, under the amended Rule 702 standard.
Talk to Lumenci about a claim chart engagement.
Table of Contents
Frequently Asked Questions
What is patent claim analysis?
It is the process of reading a patent’s claims against an accused product, element by element, to determine whether the product practices the invention. The output is a claim chart that maps each claim limitation to specific evidence in the implementation.
What is claim limitation mapping?
It maps each element of a patent claim to a specific feature, function, or structure in the accused product. A strong mapping uses the product’s own nomenclature and cites versioned, traceable evidence rather than restating the claim language.
Why do claim charts fail in litigation?
The seven most common failures are: citing marketing instead of implementation, asserting without evidence, parroting claim language, blurring functional limitations, skipping means-plus-function structure, losing traceability, and mapping too broadly against prior art.
What is the evidence hierarchy for claim charts?
Source code and binaries are the strongest evidence. Protocol captures, teardowns, and bench testing follow. Developer documentation and API references sit below those. Marketing material is the weakest, and a chart supported only by marketing is vulnerable to challenge.
How does claim chart quality affect licensing?
A chart that cites the target’s own field names and released builds reaches their engineering team with something they can verify. A chart built on datasheets invites a quick rejection. Specificity sets the negotiation.
What changed with the 2023 Rule 702 amendment?
The amendment places the burden on the proponent to show admissibility by a preponderance. An expert opinion must reflect a reliable application of the expert’s methods to the facts, which raises the standard for claim charts that serve as the evidentiary base for testimony.
What is the Peters problem in claim charting?
The maxim from Peters v. Active Manufacturing holds that what would infringe if later, anticipates if earlier. A claim limitation mapped too broadly to capture the accused product may also read on prior art, giving the defendant a free invalidity contention.
