HARTING Technologiegruppe — Patent Portfolio Analysis 2014–2024

Executive Summary

Between 2014 and 2024 the HARTING Technologiegruppe first filed 704 DOCDB patent families, pursued through 2,334 applications at 16 patent offices — an average of 3.32 applications per family. The portfolio is unusually concentrated: 535 of the 704 families (76.0%) carry a code in the single IPC subclass H01R, electrically conductive connections and coupling devices. Within H01R, 476 families sit in H01R 13/00 alone, the details of coupling devices — that is 67.6% of the entire portfolio in one main group. This is the patent footprint of a company that does one thing.

The route to protection is international rather than national. 432 families (61.4%) have at least one WO member, so the PCT is the default rather than the exception, and Germany, the PCT, the EPO, China and the United States each receive filings from between 43.8% and 76.4% of all families. The pattern is a German priority filing followed by a PCT application and a broad national phase.

Comparing two complete four-year cohorts — families first filed 2014–2017 (252) against 2018–2021 (223) — shows a portfolio narrowing onto its core and shifting east and west at the same time. The H01R share rises from 69.8% to 84.3%, while the retail and vending cluster that HARTING Systems used to feed (A47F shop fittings, G07F coin-freed apparatus) falls from 15 and 14 families to 4 each. Geographically, Germany's share of applications falls from 23.9% to 20.4% while China rises from 14.7% to 18.7%, the EPO from 14.7% to 18.2% and the United States from 13.6% to 16.5%. Russia, Canada and Japan effectively disappear from the filing programme after 2017.

Almost all of this is done alone. Only 15 of 704 families (2.1%) carry more than one applicant, and exactly one of those has an external partner — the Massachusetts Institute of Technology, on two shared applications. The other 14 are co-filings between HARTING's own legal entities. The inventor base behind the portfolio comprises 525 distinct Latin-script inventors, and every one of the fifteen most frequently named is recorded in Germany.

The single most consequential decision in this report is not analytical but definitional. Building the scope on person_name rather than on PATSTAT's harmonised han_name matters here because the group fragments across 40 separate han_name groups in this window; and searching only Latin script would have missed the Chinese and Japanese renderings under which HARTING files at CNIPA and the JPO. Adding the per-variant family counts naively gives 1,422; the deduplicated answer is 704. The gap of 718 is what consolidation is for.

Scope & Applicant Harmonisation

How the 704 families were identified — and what a shortcut would have cost.

PATSTAT performs no entity resolution. A company exists in it only as a set of spellings, one person_id per rendering, per country, per typo. For HARTING a substring search on person_name across a wide discovery window returned 182 candidate name rows, of which 65 survive review inside the report window as genuine group entities. Everything below documents which were kept, which were dropped, and on what evidence.

What was kept, what was dropped

Group	Decision	Evidence
Latin corporate variants (HARTING Electric, Electronics, Systems, Automotive, Customised Solutions, International Innovation, IT Software, Applied Technologies, Deutschland, Zhuhai, AG & Co. KG, KGaA, Stiftung, Corporation)	Kept	Name begins with HARTING and denotes a legal entity
Russian transliterations (`KHARTING ELEKTRIK GMBKH UND KO. KG`, `KHARTING EHLEKTRIK…`, `KHARTING ELEKTRONIKS GMBKH`, `KHARTING APPLAJD TEKNOLODZHIZ GMBKH`)	Kept	Cyrillic-to-Latin re-transliteration of the same entities; all filed at RU
Chinese renderings (浩亭…, 哈廷…) and Japanese renderings (ハルティング…)	Kept	Confirmed by DOCDB family overlap with the Latin variants, then verified independently
Machine transliterations (`HARYNTE ELECTRONIC LIMITED COMPANY AND TWO-IN-ONE COMPANY`, `HARBIN ELECTRONIC FOUNDATION TWO-IN-ONE COMPANY`)	Kept	Word-by-word renderings of the Chinese names; same families
`HARTIN Electric Stiftung & Co. KG`	Kept	Typo variant found only via family overlap — a `%HARTING%` search cannot reach it
`HARTING CORPORATION` (no country code)	Kept	4 of its 5 families are shared with HARTING International Innovation and the holding
Chilean individuals (Harting Glade, Harting Eckman, Harting S.A.) and their Chinese renderings 托马斯·弗朗西斯·哈廷格莱德 / 史蒂文·李·哈廷埃克曼	Excluded	Natural persons and an unrelated Chilean company; no family overlap with the group
HARTINGER, HARTINGH, HARTINGTON, HARTINGS, STANDHARTINGER, VAN HARTINGSVELDT, Karl Hartinger Kranbetrieb, VOLUME CHARTING METHOD COMPANY	Excluded	Different names that merely contain the search string

Why `han_name` is a cross-check and not the scope

PATSTAT's harmonised applicant name was checked in both directions, because it fails in two different ways and each check catches only one of them.

Contamination — none found. Every person_name sharing a han_name group with the HARTING scope is itself a HARTING entity. No foreign organisation has been merged in.
Fragmentation — severe. The group scatters across 40 distinct han_name groups. Picking any single one, or even the largest three, would have produced a portfolio a fraction of the real size. A group that passes the contamination check can still be the wrong unit of analysis.

Summed naively, the 40 han_name groups account for 1,239 family entries for applications filed 2014–2024. Summed naively over the 65 person_name variants inside the report window, the figure is 1,422. Deduplicated at family level it is 704. The difference is not error — it is the same invention appearing under several spellings, which is precisely what a family-level COUNT(DISTINCT) is there to collapse.

The renderings a Latin search cannot reach

HARTING files at CNIPA and the JPO under names written in the local script. No search on Latin characters returns them, and the failure is silent — the query succeeds and simply returns less. They were found by looking at who else appears on the other members of HARTING's own families, then verified by searching each harvested string directly, without any family restriction. That second pass is what measures the gap rather than merely confirming it: up to seven families are reachable only through these renderings — six under 哈廷电子有限公司及两合公司 and one under its machine transliteration, which may or may not be the same family, so six is the floor and seven the ceiling.

Data table: han_name fragmentation and office-side renderings

han_name group	Families (applications filed 2014–2024)
HARTING ELECT GMBH & CO KG	267
HARTING ELECT STIFTUNG & CO KG	218
HARTING (ZHUHAI) MANUFACTURING CO., LTD.	153
HARTING ELECTS GMBH	111
HARTING SYSTEMS GMBH	63
HARTING ELECTRIC FOUNDATION GMBH & CO. KG	63
HARTING INT INNOVATION AG	59
HARTING ELECTRONICS GMBH & CO. KG	36
HARTING AUTOMOTIVE GMBH	35
HARTING AG&CO KG	28
HARTING KGAA	28
HARTING CUSTOMISED SOLUTIONS GMBH & CO KG	22
28 further groups (combined)	156
Total across 40 groups (naive sum)	1,239

Office-side rendering	Office	Families	Of those, not reachable by any Latin spelling
哈廷电子有限公司及两合公司	CN	123	6
浩亭电子有限公司	CN	39	0
哈廷电子基金会两合公司	CN	21	0
ハルティングエレクトリック …	JP	19	0
ハルティングエレクトロニクス …	JP	13	0
浩亭汽车部件有限公司	CN	5	0
HARYNTE ELECTRONIC LIMITED COMPANY AND TWO-IN-ONE COMPANY	CN	4	1
浩亭股份有限公司	CN	3	0
ハルティングオートモーティヴ …	JP	3	0
浩亭（珠海）制造有限公司	CN	2	0
浩亭信息技术软件开发有限及两合公司	CN	2	0
HARBIN ELECTRONIC FOUNDATION TWO-IN-ONE COMPANY	CN	2	0
ハルティングアクチエンゲゼルシャフト …	JP	1	0
浩亭国际创新股份公司	CN	1	0

* The han_name table counts families with an application filed 2014–2024; the portfolio total of 704 counts families whose earliest filing falls in that window. The two definitions answer different questions and are not interchangeable.

Filing Trend 2014–2024

Patent families by earliest filing year, with the applications those families generated.

The two series move in opposite directions after 2021, and the reason is mechanical rather than strategic. Family counts reach their highest values of the window in 2023 (76) and 2024 (91), while applications per family collapse from a 2014–2021 range of 3.4 to 4.7 down to 2.94 in 2022, 2.04 in 2023 and 1.66 in 2024. A family is counted as soon as its first member is published; its PCT and national-phase members appear one to three years later. On a Spring 2026 edition those later members are simply not in the data yet. The family figures for 2023 and 2024 are therefore lower bounds, and the application figures for those years are substantially incomplete.

Do not read the last three years as a trend. Every year from 2022 onward is still filling up, and the effect is largest exactly where the numbers look most interesting. The PCT rate is a good illustration: it reads 37.4% for 2024 against a 2014–2021 range of 54.5% to 78.0%, not because HARTING abandoned the PCT route but because a PCT application filed on a 2024 priority publishes too late for this edition. For the same reason no growth rate in this report is anchored on the final year.

Across the complete part of the window the portfolio is remarkably steady rather than growing: 252 families first filed in 2014–2017 against 223 in 2018–2021, an output oscillating in a band of roughly 50 to 77 families a year with no sustained direction. Whether the higher 2023 and 2024 counts survive the remaining publication lag cannot be decided from this edition.

Data table: Filing Trend 2014–2024

Year	Patent families	Applications	Applications per family	Families with a WO member	PCT rate
2014	49	214	4.37	33	67.3%
2015	74	271	3.66	55	74.3%
2016	52	197	3.79	33	63.5%
2017	77	265	3.44	42	54.5%
2018	54	201	3.72	30	55.6%
2019	50	195	3.90	33	66.0%
2020	50	237	4.74	39	78.0%
2021	69	266	3.86	52	75.4%
2022*	62	182	2.94	40	64.5%
2023*	76	155	2.04	41	53.9%
2024*	91	151	1.66	34	37.4%
Total	704	2,334	3.32	432	61.4%

* 2022–2024: incomplete data due to the publication delay of roughly 18 months plus national-phase lag (PATSTAT Global, Spring 2026 / patstat_2026a). Families are assigned to the earliest filing year of any patent-of-invention member of the family, computed without a year restriction and only then filtered to 2014–2024.

Geographic Strategy

Where HARTING files — and how the balance moved between two complete filing cohorts.

Germany is the entry point for almost the whole portfolio — 538 of 704 families (76.4%) have a DE filing — and the PCT is the second step for 431 of them. The interesting movement is in the third step, the national phase. Between the 2014–2017 and 2018–2021 cohorts, China's share of applications rises from 14.7% to 18.7%, the EPO's from 14.7% to 18.2% and the United States' from 13.6% to 16.5%, while Germany's own share falls from 23.9% to 20.4%. In the same comparison Russia goes from 34 applications to 3, and Canada and Japan from 18 each to none. The absolute German figure falls too (226 to 183), so this is a portfolio being pushed further out from the same home base rather than one being relocated.

Both cohorts were chosen to end before the incomplete years. A boundary drawn at 2023 would have put missing national-phase entries inside the later period and rendered the same data as a broad geographic retreat — a lag artefact, not a decision by the company.

Data table: Filing offices 2014–2024

Office	Families	Share of 704 families	Applications
DE — Germany (DPMA)	538	76.4%	538
WO — PCT (WIPO)	431	61.2%	433
EP — European Patent Office	374	53.1%	389
CN — China (CNIPA)	365	51.8%	376
US — United States (USPTO)	308	43.8%	317
KR — Korea (KIPO)	135	19.2%	136
RU — Russia	37	5.3%	37
GB — United Kingdom	35	5.0%	35
PL — Poland	23	3.3%	23
CA — Canada	19	2.7%	19
JP — Japan (JPO)	18	2.6%	18
CH — Switzerland	6	0.9%	6
DK — Denmark	2	0.3%	2
ES — Spain	2	0.3%	2
MX — Mexico	2	0.3%	2
BR — Brazil	1	0.1%	1
Total (applications)	—	—	2,334

Family counts per office sum to more than 704 because one family is normally filed at several offices. The application column sums exactly to the portfolio total.

Data table: Geographic shift between cohorts

Office	Applications, cohort 2014–2017	Share	Applications, cohort 2018–2021	Share
DE	226	23.9%	183	20.4%
WO	164	17.3%	154	17.1%
CN	139	14.7%	168	18.7%
EP	139	14.7%	164	18.2%
US	129	13.6%	148	16.5%
KR	62	6.5%	64	7.1%
RU	34	3.6%	3	0.3%
CA	18	1.9%	0	0.0%
JP	18	1.9%	0	0.0%
PL	14	1.5%	9	1.0%
Other (DK, ES, CH, BR)	4	0.4%	6	0.7%
Total	947	100.0%	899	100.0%

Cohorts are defined by the family's earliest filing year: 252 families in 2014–2017 and 223 in 2018–2021. Both end before the years affected by publication lag.

Technology Profile

IPC subclasses and WIPO technology fields, aggregated over every member of each family.

535 of 704 families (76.0%) carry a code in H01R, and the next subclass, H05K, reaches 51. That is a ratio of more than ten to one between first and second place. The WIPO view says the same thing at a coarser grain: 562 families (79.8%) fall in the single field "Electrical machinery, apparatus, energy". What sits outside the core is a thin but readable set of neighbourhoods — cable installation (H02G), electric vehicle supply (B60L), printed circuits (H05K), shop fittings and coin-freed apparatus (A47F, G07F), and electrolytic coating (C25D). For a connector manufacturer these read as adjacent rather than as diversification, though the classification alone cannot establish that.

IPC or CPC — measured rather than assumed

Running the same 704 families through CPC instead of IPC gives 526 families in H01R against 535 under IPC. Both schemes reach 692 of the 704 families overall, so neither is systematically blind here, and for this applicant-scoped portfolio IPC has marginally the better reach in the core subclass. CPC does add something IPC cannot: the Y tagging codes, which have no IPC equivalent.

CPC tagging code	Meaning	Families
`Y02T`	Climate change mitigation technologies related to transportation	24
`Y02E`	Reduction of greenhouse gas emissions related to energy generation, transmission or distribution	8
`Y10T`	Technical subjects covered by former US classification	4
`Y02P`	Climate change mitigation technologies in the production or processing of goods	3

Two counting caveats. First, a family normally carries several classification codes, so the columns above sum to more than 704 — the fifteen IPC subclasses shown account for 855 family-subclass pairs across at most 704 distinct families. Percentages are always of 704, never of that sum, and no pie or doughnut chart is used for this data. Second, the Y codes are tags applied on top of a family's ordinary technical subclass, so a family counted under Y02T is also counted under H01R or B60L; and because Y codes are CPC-only, they under-represent offices that do not assign CPC — which for this portfolio means the Chinese and Japanese filings.

Data table: Top 15 IPC subclasses

IPC subclass	Title	Families	Share of 704
H01R	Electrically-conductive connections; structural associations of mutually-insulated electrical connecting elements; coupling devices; current collectors	535	76.0%
H05K	Printed circuits; casings or constructional details of electric apparatus	51	7.2%
H02G	Installation of electric cables or lines, or of combined optical and electric cables or lines	35	5.0%
B60L	Propulsion of electrically-propelled vehicles; supplying electric power for auxiliary equipment	33	4.7%
G06F	Electric digital data processing	28	4.0%
G01R	Measuring electric variables; measuring magnetic variables	23	3.3%
A47F	Special furniture, fittings or accessories for shops, storehouses, bars, restaurants; paying counters	22	3.1%
G07F	Coin-freed or like apparatus	20	2.8%
H01B	Cables; conductors; insulators; selection of materials for conductive, insulating or dielectric properties	18	2.6%
H04L	Transmission of digital information	17	2.4%
G02B	Optical elements, systems or apparatus	16	2.3%
F16B	Devices for fastening or securing constructional elements or machine parts together	15	2.1%
C25D	Processes for the electrolytic or electrophoretic production of coatings; electroforming	15	2.1%
G06Q	ICT specially adapted for administrative, commercial, financial or managerial purposes	15	2.1%
B61G	Couplings specially adapted for railway vehicles; draught or buffing appliances	12	1.7%

692 of the 704 families carry at least one IPC code. Titles from patstat-mtc.classification.dpt_ipc, IPC edition 20260101.

Data table: WIPO technology fields

WIPO sector	Technology field	Families	Share of 704
Electrical engineering	Electrical machinery, apparatus, energy	562	79.8%
Electrical engineering	Audio-visual technology	61	8.7%
Mechanical engineering	Transport	56	8.0%
Instruments	Measurement	44	6.3%
Electrical engineering	Computer technology	38	5.4%
Instruments	Control	37	5.3%
Mechanical engineering	Mechanical elements	24	3.4%
Other fields	Furniture, games	22	3.1%
Electrical engineering	Telecommunications	20	2.8%
Chemistry	Surface technology, coating	18	2.6%
Instruments	Optics	17	2.4%
Electrical engineering	Digital communication	17	2.4%
Electrical engineering	IT methods for management	15	2.1%
Mechanical engineering	Handling	12	1.7%
Mechanical engineering	Other special machines	12	1.7%
Mechanical engineering	Machine tools	10	1.4%

WIPO 35-field mapping via tls230_appln_techn_field. Fields below 10 families are omitted from the table. A family can belong to several fields, so the column sums above 704.

H01R Detail & Technology Shift

Inside the core subclass, and how the balance between core and periphery moved.

Within H01R the portfolio is concentrated again, one level down. 476 of the 535 H01R families (89.0%) carry H01R 13/00 — the details of coupling devices: contacts, housings, seals, locking, shielding. Three further main groups form the rest of the working set: direct conductive connections (H01R 4/00, 131 families), the machinery for making and assembling connectors (H01R 43/00, 102) and two-part coupling devices (H01R 24/00, 89). The presence of H01R 43/00 at that level is worth noting on its own: roughly a fifth of the H01R portfolio protects how the connectors are manufactured, not only what they are.

Between the two complete cohorts the portfolio moves inward. H01R's share rises from 69.8% to 84.3% of families, and the adjacent electrical subclasses rise with it — cable installation H02G from 3.2% to 8.5%, printed circuits H05K from 6.7% to 9.0%, electric-vehicle supply B60L from 2.4% to 4.9%. The movement in the other direction is just as clear and concerns a different business: shop fittings (A47F) fall from 15 families to 4, coin-freed apparatus (G07F) from 14 to 4, data processing (G06F) from 12 to 3. That cluster is consistent with HARTING Systems, the group's retail-systems entity, which accounts for 45 families across the window — but this report does not cross-tabulate applicant entity against classification, so the attribution is an inference from the subject matter, not a measurement.

Data table: IPC main groups within H01R

Main group	Title	Families	Share of 535 H01R families
H01R 13/00	Details of coupling devices of the kinds covered by groups H01R 12/70 or H01R 24/00–H01R 33/00	476	89.0%
H01R 4/00	Electrically-conductive connections between two or more conductive members in direct contact	131	24.5%
H01R 43/00	Apparatus or processes specially adapted for manufacturing, assembling, maintaining or repairing line connectors	102	19.1%
H01R 24/00	Two-part coupling devices, or either of their cooperating parts, characterised by their overall structure	89	16.6%
H01R 12/00	Structural associations of mutually-insulated connecting elements specially adapted for printed circuits	63	11.8%
H01R 9/00	Structural associations of mutually-insulated connecting elements, e.g. terminal strips or blocks	49	9.2%
H01R 11/00	Individual connecting elements providing two or more spaced connecting locations	35	6.5%
H01R 31/00	Coupling parts supported only by co-operation with counterpart	33	6.2%
H01R 25/00	Coupling parts adapted for simultaneous co-operation with two or more identical counterparts	16	3.0%
H01R 101/00	One pole (indexing code)	14	2.6%
H01R 27/00	Coupling parts adapted for co-operation with two or more dissimilar counterparts	12	2.2%
H01R 35/00	Flexible or turnable line connectors	12	2.2%

Non-exclusive: a family is normally classified in several main groups, so the column sums to 1,032 across at most 535 distinct families.

Data table: Technology shift between cohorts

IPC subclass	Families, cohort 2014–2017	Share of 252	Families, cohort 2018–2021	Share of 223
H01R	176	69.8%	188	84.3%
H05K	17	6.7%	20	9.0%
H02G	8	3.2%	19	8.5%
B60L	6	2.4%	11	4.9%
A47F	15	6.0%	4	1.8%
G07F	14	5.6%	4	1.8%
G06F	12	4.8%	3	1.3%
G01R	11	4.4%	4	1.8%

Both cohorts end before the years affected by publication lag. Shares are of all families in the cohort and do not sum to 100% — a family can carry several subclasses, and the eight subclasses shown are not the whole portfolio.

Corporate Entities

Which legal entity of the group appears as applicant, and how often.

HARTING Electric — under both its GmbH & Co. KG and its Stiftung & Co. KG form — is the filing entity for 365 families and 1,415 applications, more than half the portfolio and 60.6% of all applications. It is also where the international strategy is concentrated: 3.88 applications per family against 3.32 for the portfolio as a whole. HARTING Electronics follows at 102 families, then the holding entities at 69 and HARTING International Innovation AG at 67. The last of these is the newest name in the filing programme: every one of its spellings carries a first filing year of 2023.

Entity group	Type	Families	Applications	Applications per family
HARTING Electric / Electric Stiftung & Co. KG	Company	365	1,415	3.88
Office-side renderings (CN / JP)	Rendering	187	205	1.10
HARTING Electronics GmbH (& Co. KG)	Company	102	324	3.18
Holding — HARTING AG & Co. KG / KGaA / Stiftung & Co. KG	Company	69	109	1.58
HARTING International Innovation AG	Company	67	87	1.30
HARTING Systems GmbH (& Co. KG)	Company	45	59	1.31
HARTING Automotive GmbH	Company	31	62	2.00
HARTING (Zhuhai) Manufacturing Co., Ltd.	Company	22	28	1.27
HARTING Customised Solutions GmbH & Co. KG	Company	16	27	1.69
HARTING IT Software Development / Applied Technologies	Company	7	13	1.86
HARTING Corporation	Company	5	5	1.00
HARTING Deutschland GmbH & Co. KG	Company	1	1	1.00

Two things this table is not. First, the rows are not disjoint: a family can name more than one entity, so the family column sums to 917 across 704 distinct families. Second, "Office-side renderings" is not a legal entity but a bucket. It holds the Chinese and Japanese spellings under which CNIPA and JPO record HARTING's filings; those records do not identify which German entity is behind each one, so attributing them would be invention rather than measurement. They are counted in the totals and shown separately here.

Two entity questions were resolved rather than assumed. HARTING CORPORATION, which carries no country code in PATSTAT, was accepted into the scope because four of its five families are shared with HARTING International Innovation and the holding — a name alone would not have been enough. And the split between "GmbH & Co. KG" and "Stiftung & Co. KG" is a change of legal form, not two companies: the GmbH & Co. KG spellings carry filings from 2008 to 2023 and the Stiftung & Co. KG spellings from 2013 to 2025, overlapping rather than succeeding one another in the data. Both are counted as one entity here, which is why that row spans 365 families across both forms.

Data table: Applicant entity groups

Entity group	Families	Applications	Share of 2,334 applications
HARTING Electric / Electric Stiftung & Co. KG	365	1,415	60.6%
Office-side renderings (CN / JP)	187	205	8.8%
HARTING Electronics	102	324	13.9%
Holding (AG & Co. KG / KGaA / Stiftung & Co. KG)	69	109	4.7%
HARTING International Innovation AG	67	87	3.7%
HARTING Systems	45	59	2.5%
HARTING Automotive	31	62	2.7%
HARTING (Zhuhai) Manufacturing	22	28	1.2%
HARTING Customised Solutions	16	27	1.2%
HARTING IT Software / Applied Technologies	7	13	0.6%
HARTING Corporation	5	5	0.2%
HARTING Deutschland	1	1	0.0%
Sum of rows (non-exclusive)	917	2,335	—
Distinct total	704	2,334	100%

Grant Status

How much of the portfolio has been granted, by office and by filing cohort.

Across the whole window 325 of 704 families (46.2%) and 962 of 2,334 applications (41.2%) are recorded as granted. Read only the cohorts old enough to have finished examination: for families first filed 2014–2020 the granted share of applications sits in a narrow band of 50.0% to 60.5%, and it is that band, not the portfolio average, that describes HARTING's conversion rate. Among the larger offices the spread is wide — 78.7% at KIPO and 66.2% at the USPTO against 47.3% at the EPO and 24.7% at the DPMA. The low German figure is consistent with the DPMA acting as the priority-establishing first filing here — 538 of 704 families file at the DPMA, more than at any other office — though this report does not trace individual applications through withdrawal, continuation or grant.

Grant rates for recent cohorts are artificially low and are not a quality signal. Examination takes three to five years, so the drop from 53.6% (2020 cohort) to 17.7% (2021), 3.8% (2022), 2.6% (2023) and 4.6% (2024) measures pendency, not refusal. Two further reading notes: WO applications are excluded from the office table because the PCT phase produces no grant, and the 35 GB applications show a granted count of zero, which for a national office of that size is more likely a gap in PATSTAT's grant flag than a real outcome — it should not be read as ten years of unsuccessful UK filings.

Data table: Grant status by office

Office	Applications	Granted	Granted share
RU	37	37	100.0%
PL	23	23	100.0%
CH	6	5	83.3%
KR	136	107	78.7%
US	317	210	66.2%
CN	376	199	52.9%
EP	389	184	47.3%
JP	18	5	27.8%
DE	538	133	24.7%
CA	19	1	5.3%
GB	35	0	0.0%
DK	2	2	100.0%
ES	2	2	100.0%
MX	2	0	0.0%
BR	1	0	0.0%
WO (no grant in the PCT phase)	433	—	—

Data table: Grant status by filing cohort

Cohort (earliest filing year)	Applications	Granted	Granted share
2014	214	107	50.0%
2015	271	164	60.5%
2016	197	119	60.4%
2017	265	160	60.4%
2018	201	115	57.2%
2019	195	105	53.8%
2020	237	127	53.6%
2021*	266	47	17.7%
2022*	182	7	3.8%
2023*	155	4	2.6%
2024*	151	7	4.6%
Total	2,334	962	41.2%

* 2021–2024: still in examination. These rows measure pendency, not outcome.

Inventors

The named inventor base behind the portfolio, and how broadly it is spread.

525 distinct Latin-script inventors are named across 686 of the 704 families, so the portfolio rests on a broad base rather than a handful of people: the most frequently named inventor appears on 44 families, 6.3% of the total, and no one reaches 50. Reading the recent column alongside the total shows two different profiles inside that group. Daniel Wesseler's 20 families are all from 2020–2024, and Marc Genau's 26 of 28 and Sebastian Griepenstroh's 26 of 35 are as well, while Lutz Tröger (1 of 27), Heiko Herbrechtsmeier (1 of 19) and Bernard Schlegel (1 of 16) contributed almost entirely in the first half of the window.

The number of distinct inventors named per cohort moves from a 2014–2020 range of 64 to 90 up to 127 in 2021, 106 in 2023 and 125 in 2024, on family counts that did not rise proportionally. Per family with at least one named inventor that is roughly 1.3 inventors in 2014–2018 against 1.6 to 1.8 from 2019 onward — larger inventor teams per invention. Because inventor names on later family members are still being published, the figures for 2023 and 2024 are lower bounds like every other count in this window.

How inventor names were consolidated, and what is excluded. The same person appears in PATSTAT as MEIER, HEIKO and MEIER HEIKO; both are merged onto a punctuation-stripped key. Chinese and Japanese renderings of inventor names are deliberately not merged and are excluded from this ranking: they carry only a single Latin initial, so the same key would collapse every CN inventor sharing that initial into one fictitious person. This exclusion costs no families where a Latin spelling also exists — 686 of 704 families are covered — but it means the counts here are of Latin-script renderings. Different people with identical names cannot be separated by this method either.

Data table: Most frequently named inventors

Inventor	Country	Families 2014–2024	of which first filed 2020–2024
Meier, Heiko	DE	44	9
Griepenstroh, Sebastian	DE	35	26
Friesen, Markus	DE	33	17
Loske, Felix	DE	31	14
Kropiewnicki, Norbert	DE	30	17
Genau, Marc	DE	28	26
Tröger, Lutz	DE	27	1
Beimdieck, Carsten	DE	26	18
Riechmann, Till	DE	22	13
Kliever, Wilhelm	DE	21	9
Ferderer, Albert	DE	21	11
Wesseler, Daniel	DE	20	20
Herbrechtsmeier, Heiko	DE	19	1
Schlegel, Bernard	DE	16	1
Hellige, Denny	DE	16	11

Data table: Inventors per filing cohort

Cohort	Distinct inventors	Families with a named inventor	Inventors per such family
2014	64	49	1.31
2015	90	72	1.25
2016	73	52	1.40
2017	88	77	1.14
2018	72	54	1.33
2019	80	50	1.60
2020	79	46	1.72
2021	127	69	1.84
2022*	96	62	1.55
2023*	106	76	1.39
2024*	125	79	1.58
All cohorts (distinct)	525	686	—

* 2022–2024: incomplete, as elsewhere in this report. The "all cohorts" row counts each inventor once even if they appear in several cohorts, which is why it is far below the column sum.

Most Cited Families

Incoming citations to any publication of a HARTING family.

354 of 704 families (50.3%) have been cited at least once, drawing 2,755 citations in total. The distribution is steep: the single most cited family — Holding frame for plug connectors, first filed in 2014 — accounts for 139 citations on its own, more than twice the second place. Four of the top ten are holding frames and three more are modular connectors or connector modules — together the mechanism that lets one industrial connector carry interchangeable inserts, and the part of HARTING's portfolio that other applicants' documents cite most. Two further entries — Liquid-cooled contact element (2016, 57) and Terminal unit for an actively cooled cable (2016, 40) — sit in cooled high-current contacts, and together with Car charging plug-in connector (2017, 41) they point at fast charging.

Citation counts favour older families and are not a quality score. A family can only be cited after it publishes, and it keeps accumulating citations for years, so the top of this list is necessarily drawn from the early part of the window — nine of the ten entries were first filed in 2014–2017. A 2022 family with the same intrinsic significance would not yet appear. Citations here are incoming references to any publication of the family from any citing document, without distinguishing examiner from applicant citations.

Data table: Ten most cited families

DOCDB family	Title	Earliest filing year	Incoming citations
`49475936`	Holding frame for plug connectors	2014	139
`56925929`	Holding frame for plug connector modules that is fixible in a housing	2015	67
`52396325`	Holding frame for a plug-type connector	2014	57
`59501125`	Liquid-cooled contact element	2016	57
`53054818`	Modular plug connector	2015	52
`63372503`	Car charging plug-in connector	2017	41
`59631518`	Terminal unit for an actively cooled cable	2016	40
`56741022`	Holding frame for holding plug connector modules	2015	40
`63407004`	Modular connector for circuit boards	2017	38
`67184754`	Plug connector module for an industrial plug connector	2018	38
All families cited at least once			354 families / 2,755 citations

Titles are taken from the EP, WO, US or GB member of each family.

Collaborations

Who else appears as applicant on a HARTING application.

15 of 704 families (2.1%) name more than one applicant. Exactly one of them has an applicant that is not a HARTING group entity. On the remaining 14 families every co-applicant is another HARTING entity — most often HARTING Electric appearing alongside itself under two spellings of its legal form, and in a few cases HARTING Electric together with HARTING IT Software Development or the holding.

Co-applicant, exactly as recorded	Type	Relationship	Families	Applications
The Massachusetts Institute of Technology	University	External partner	1	2
HARTING Electric Stiftung & Co. KG	Company	Group-internal	9	9
Harting Electric GmbH & Co. KG	Company	Group-internal	8	9
HARTING Electric Stiftung &Co. KG	Company	Group-internal	2	2
HARTING ELECTRIC GMBH & CO., KG	Company	Group-internal	2	2
HARTING Electric & Stiftung & Co. KG	Company	Group-internal	1	1
HARTING IT SOFTWARE DEVELOPMENT GMBH & CO. KG	Company	Group-internal	1	1
HARTING IT SOFTWARE DEVELOPMENT GMBH & CO,. KG	Company	Group-internal	1	1
HARTING Electric Stiftung & Co., KG	Company	Group-internal	1	1
Harting Electric GmbH & Co., KG	Company	Group-internal	1	1
HARTING AG & CO. KG	Company	Group-internal	1	1
HARTIN Electric Stiftung & Co. KG	Company	Group-internal (typo variant)	1	1
HARTING ELECTRIC GMBH &CO. KG	Company	Group-internal	1	1
HARTING Electric Stiftung GmbH & Co. KG	Company	Group-internal	1	1

Rows are individual PATSTAT name records, not consolidated entities, and they are not disjoint — one family can carry two spellings of the same company, which is exactly why the rows below MIT add up to more than the 14 group-internal families.

This is, on the evidence, a company that patents alone. A co-application rate of 2.1%, with exactly one external partner across eleven years, is a defining structural feature of the portfolio and stands in contrast to the breadth of the filing programme — HARTING reaches 16 offices but shares almost none of its inventions with anyone outside the group. The one exception, with the Massachusetts Institute of Technology, appears on two applications of a single family. What this measurement cannot see is collaboration that never surfaces as a co-applicant record: joint development where only one party files, or work assigned before filing, leaves no trace in tls207_pers_appln.

Why co-applicants are joined on the application, not on the family. Sharing a DOCDB family is not co-filing. Two names surfaced on other members of HARTING families without ever appearing on the same application — Molex LLC on one Chinese member, and DIAC Holdings LLC on the EP, US and WO members of one family. Counted at family level they would read as partners; counted at application level, which is what co-filing means, they are not, and they are excluded here. The same rule removed the Chinese and Japanese renderings of HARTING's own name from the partner list, where they would otherwise have ranked as the company's largest collaborators by a wide margin.

Methodology

Data basis, search strategy, counting rules and known limitations.

Data basis note — 15 August 2026

This report was computed on EPO PATSTAT Global, Spring 2026 (patstat-mtc.patstat_2026a), with classification titles and hierarchy taken from IPC edition 20260101 (in force 1 January 2026) and CPC edition 202608 (in force 1 August 2026), read from patstat-mtc.classification.dpt_edition rather than hard-coded. Every figure in this report comes from a query executed against that edition on that date; none is carried over from any other source. Figures computed on a different PATSTAT or classification edition are not comparable with these — the difference between editions looks exactly like real change.

Search strategy, counting rules and data source — the full method

Search strategy — an applicant scope, not a classification scope

The patent set is defined by who filed, not by classification codes. The scope is built on person_name in tls206_person, joined through tls207_pers_appln with applt_seq_nr > 0 (applicants only).
Discovery: a substring search UPPER(person_name) LIKE '%HARTING%' over filing years 2008–2026 returned 182 candidate name rows. A prefix search would have missed ZHUHAI HARTING LTD.; the substring search also returns unrelated names, which were pruned by hand.
Office-side renderings: applicant names on the other members of HARTING's own DOCDB families were listed, which surfaced eight Chinese renderings (浩亭…, 哈廷…), four Japanese ones (ハルティング…) and two machine transliterations. Each was then searched directly, without any family restriction, to measure what it adds rather than merely confirm it: six to seven families that no Latin spelling reaches (six under one Chinese rendering and one under its machine transliteration, which may be the same family). Two hits were rejected as the Chinese renderings of Chilean individuals.
Kept: 65 person_name variants inside the window, including the Russian re-transliterations (KHARTING ELEKTRIK GMBKH UND KO. KG and siblings) and the typo variant HARTIN Electric Stiftung & Co. KG, which only family overlap could find. Excluded: the Chilean individuals Harting Glade and Harting Eckman and the Chilean Harting S.A.; and every HARTINGER, HARTINGH, HARTINGTON, HARTINGS, STANDHARTINGER, VAN HARTINGSVELDT and similar name that merely contains the search string.
han_name was used only as a cross-check, in both directions. Contamination: none — every person_name sharing a han_name group with the scope is itself a HARTING entity. Fragmentation: severe — the group scatters across 40 distinct han_name groups, so no single group could have served as the scope.
Classification code-set completion does not apply to this report. Subtree expansion, predecessor unions from the reclassification tables, and the retention of deprecated symbols are the correctness rules for a scope derived from a code list. This scope comes from applicant harmonisation; no code list defines the family set, so there is nothing to expand or union, and there is no code-set sample to draw. Classification is used here only to describe the set that names already selected. In that descriptive use, no is_active filter was applied and title lookups against dpt_ipc/dpt_cpc were written as LEFT JOINs, because retired symbols carry NULL titles and an inner join would drop them as silently as an is_active filter; no untitled symbol appeared in the reported profile.

Family counting and the definition of a year

The counting unit is the DOCDB patent family (docdb_family_id), always via COUNT(DISTINCT …). Pseudo-families (docdb_family_id = 0) are excluded.
ipr_type = 'PI' restricts the set to patents of invention. Utility models and designs are excluded — see Scope Limitations for what that removes.
A family's year is the earliest filing year of any patent-of-invention member of that family, computed with no year restriction and only then filtered to 2014–2024. Taking the minimum over the filings inside the window instead would pin families first filed before 2014 to the first year of the window and inflate it. The sum-equals-total check passes identically under both readings and cannot distinguish them, which is why the definition is stated rather than assumed.
Under the alternative reading — minimum over every member regardless of ipr_type — the window would hold 703 families instead of 704, and 3 families in the reported set have a non-PI member filed before 2014.
PCT rate is families with at least one WO member divided by total families (432 / 704 = 61.4%), never the WO application count. Grant rate is granted applications over applications, per office or per cohort, never a mix of family and application level. Co-application rate is co-filed families over total families, and co-applicants are joined on appln_id, never on docdb_family_id.
Applications counted for a family include every application in the scope belonging to that family, including any filed just outside the calendar window, since the family as a whole belongs to its earliest-filing cohort.

Multi-assignment note

A patent family can carry several IPC and CPC subclasses, belong to several WIPO technology fields, name several applicant entities and be filed at several offices. Every such table in this report therefore sums to more than 704: the fifteen IPC subclasses shown sum to 855, the twelve H01R main groups to 1,032, the entity table to 917, and office family counts to 2,296. This is correct behaviour, not double counting, and it is why no pie or doughnut chart appears anywhere in this report — those forms imply a partition, and this data is not partitioned. All percentages are of 704 families or of 2,334 applications, never of a column sum.
The CPC Y codes (Y02T, Y02E, Y02P, Y10T) are tags layered on top of a family's ordinary technical subclass, so a family counted under Y02T is also counted under H01R or B60L. Because Y codes exist only in CPC, they under-represent offices that do not assign CPC.

Data source and stack

EPO PATSTAT Global, Spring 2026 edition, dataset patstat-mtc.patstat_2026a, on Google BigQuery.
Classification views patstat-mtc.classification.dpt_ipc, dpt_cpc and dpt_edition, joined to PATSTAT on symbol_patstat.
Core tables: tls201_appln, tls202_appln_title, tls206_person, tls207_pers_appln, tls209_appln_ipc, tls211_pat_publn, tls212_citation, tls224_appln_cpc, tls230_appln_techn_field, tls901_techn_field_ipc.
Access via the patstat-mcp server; analysis and visualisation with Claude. All ten queries are reproducible and shipped with the report in queries.sql.

Scope Limitations

Utility models and designs are excluded, and here that is a material exclusion. Within the same applicant scope and filing years 2014–2024, PATSTAT holds 244 utility-model families (259 applications: 160 at CN, 90 at DE, 1 at JP) and 9 design families (14 applications, all US). For a connector manufacturer with Chinese manufacturing this is a real part of the protection activity, not a rounding error. It is outside this report by definition, not because it is unimportant.
2022–2024 are incomplete because of the roughly 18-month publication delay plus national-phase lag. Applications per family fall from a 2014–2021 range of 3.4–4.7 to 2.94, 2.04 and 1.66 in those three years. All family counts for them are lower bounds and no period comparison in this report ends inside them.
12 of 704 families carry neither an IPC nor a CPC code, so every classification percentage rests on 692 classified families while being expressed against 704.
Inventor counts are of Latin-script name renderings only. Chinese and Japanese inventor renderings are excluded from the ranking because their consolidation key would collapse to a single Latin initial and merge unrelated people. 686 of 704 families are covered. Distinct people sharing a name cannot be separated.
The Chinese and Japanese office-side renderings cannot be attributed to a specific German legal entity from the PATSTAT record, so in the entity breakdown they form their own bucket (187 families) rather than being distributed across the operating companies.
Undiscovered spelling errors remain possible. The typo variant HARTIN Electric Stiftung & Co. KG was found only because it shares a family with correctly spelled filings. A family filed exclusively under a misspelling, with no correctly-spelled family member, would not be found by any method used here.
No independent external benchmark was obtained (QA Check 9). A web search for third-party portfolio figures returned only a prior mtc.berlin publication, which is circular and was therefore rejected rather than used. The scope is instead evidenced internally: 182 candidates reviewed, 65 variants kept, both han_name directions checked, and the office-side renderings verified independently of family overlap.
Two grant-flag caveats. Grant rates for the 2021–2024 cohorts are artificially low because examination takes three to five years. Separately, the 35 GB applications carry a granted count of zero, which is more plausibly a gap in PATSTAT's grant flag for that office than a real outcome.
Applications filed before 2014, and families whose earliest patent-of-invention filing falls outside 2014–2024, are not part of this report.

Glossary — patent terms explained

DOCDB patent family: A group of patent applications protecting the same invention in different countries. Counted once, so that a broadly filed invention does not outweigh a narrowly filed one.
person_name vs. han_name: person_name is the applicant name exactly as the office recorded it — one record per spelling. han_name is PATSTAT's attempt to harmonise those spellings into one name per organisation. It is used here as a check, never as the scope.
ipr_type: The kind of industrial property right: PI patent of invention, UM utility model, DP design patent. This report uses PI only.
IPC / CPC: International and Cooperative Patent Classification — hierarchical schemes for classifying inventions by technology. CPC is the finer of the two and carries the Y tagging codes; IPC is assigned by more offices.
H01R: The IPC subclass "Electrically-conductive connections; structural associations of a plurality of mutually-insulated electrical connecting elements; coupling devices; current collectors" — HARTING's core technology area.
Y codes (Y02, Y04, Y10): CPC-only tagging schemes applied across the whole classification, chiefly for climate-change mitigation technologies. They sit on top of a family's ordinary subclass rather than replacing it.
WIPO technology fields: A mapping of IPC codes onto 35 technology fields grouped into five sectors, used for a high-level view that is comparable across industries.
PCT / WO: The Patent Cooperation Treaty. A single international application (office code WO) preserves the right to enter national or regional examination in the member states for up to 30 months.
Filing office codes: DE Germany (DPMA), EP European Patent Office, US United States, CN China (CNIPA), KR Korea (KIPO), JP Japan (JPO), RU Russia, GB United Kingdom, PL Poland, CA Canada, CH Switzerland, WO PCT.
Grant rate: The share of applications recorded as granted. Rates for recent filing years appear low because examination takes years, not because the applications were refused.
Incoming citation: A reference from a later patent document to a publication of this family. Older families accumulate more, so citation counts favour the early part of any window.

Report Files

This report ships as a set of files: meta.json (metadata), cover.html and kpi-strip.html, thirteen section files under sections/, and queries.sql — the complete, commented, runnable SQL behind every number shown, including the exact scope predicate, the edition header, and the reasoning for each counting decision. Every insight box in this report carries an HTML comment naming the query it comes from.

Like what you see?

This report was built with a fully reproducible pipeline: EPO PATSTAT Global on BigQuery, a custom MCP server, and Claude for analysis and visualisation. Ten SQL queries, every figure traceable to one of them, and the applicant scope documented decision by decision.

Custom Patent Report Need a similar analysis for your technology area, company, or research question? We build tailored patent intelligence reports — including the unglamorous part, getting the applicant scope right.

The Toolchain Want to run your own AI-powered patent analyses? We help you set up PATSTAT on BigQuery with MCP server integration.

Arne Krueger — LinkedIn Book a Call depa.tech mtc.berlin