How AI Search Is Recommending Patent Attorneys: Monthly Trends

Executive Summary

Fish & Richardson remains the category leader in AI search recommendations for patent attorneys at 43.4% valid recommendation coverage in September 2026, but that leadership is now under pressure. The firm's coverage fell 15.1 points from 58.5% in July, a significant decline exceeding normal month-to-month variation. The gap to Finnegan, the next brand, now stands at 21.0 points, though Finnegan itself fell 20.1 points from 42.5% in July to 22.4% in September, also a significant drop.

September was a month of broad category retreat. Finnegan and Fish & Richardson both registered declines beyond normal variation. Finnegan's 16.6-point drop from August to September was the sharpest single-month move in the series, while Fish & Richardson's September decline, though significant against July, just missed the significance threshold against August at 15.0 points. The category's recommendation coverage contracted across most tracked brands, with only Sterne Kessler and Wolf Greenfield holding coverage near their July levels.

The strongest structural signal is Finnegan's narrowing distance to the middle of the field. A gap that stood at 36.1 points over Sterne Kessler in July and 25.5 points over Wolf Greenfield in July has compressed to 15.8 and 5.3 points respectively, widening in Finnegan's disfavor every month. The category is not merely quiet this month; it is repositioning around a smaller set of consistently recommended firms.

Each monthly run begins with 427 prompt-surface observations (299 unique questions) in July, 524 prompt-surface observations (394 unique questions) in August, and 575 prompt-surface observations (394 unique questions) in September across the benchmark's defined AI/search surface universe. Of those, all mentioned a tracked brand or competitor in each month; 101 in July, 90 in August, and 93 in September were relevant, while 326 in July, 434 in August, and 482 in September were irrelevant. The public metrics use the qualified observation counts of 94 in July, 77 in August, and 76 in September that survive both qualification stages.

Key Findings

Benchmark Context

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

The September qualified base of 76 observations is smaller than July's 94, while the raw collection universe grew from 427 to 575 prompt-surface observations. The August intermediate month carried 77 qualified observations and a breadth of 6 surfaces, matching September on both.

Benchmark-Level Metrics

The share of qualified observations producing recommendation-shaped answers rose sharply from July to September, reaching 57.9%, while the share yielding valid recommendation shortlists fell to 50.0% from 68.1%. In August the shortlist share peaked at 72.7%, making September's decline a two-month reversal of that intermediate high.

AI Recommendation Trend

Category Leadership Holds, but the Coverage Base Is Narrowing

Fish & Richardson remains first in the category, yet its 43.4% coverage in September is the lowest leadership figure in the three-month series. The top of the field is compressing: Fish & Richardson at 43.4%, Finnegan at 22.4%, and Knobbe Martens at 19.7% form a tighter cluster than the July spread of 58.5%, 42.5%, and 30.9%.

Two brands exceeded normal month-to-month variation in coverage between July and September: Finnegan and Fish & Richardson both registered significant declines. The broader category shift came from the combination of those two leader-level retreats with smaller declines across the middle and lower tiers.

What Changed This Month

Finnegan: Significant Two-Month Decline Narrows Its Distance to the Field

Finnegan's valid recommendation coverage fell 20.1 points from 42.5% in July to 22.4% in September, a significant decline. The sharper movement came in the last month: coverage dropped 16.6 points from 39.0% in August to 22.4% in September. The firm's raw mention presence fell 21.7 points to 36.8%, its top-three rate fell 20.7 points to 19.7%, and its rank-one rate slipped 4.6 points to 4.0%.

Finnegan appeared in 28 of 76 qualified observations in September, holding 17 valid recommendations. In July the firm appeared in 55 of 94 observations with 40 valid recommendations. Its rank-one count fell from 8 in July to 3 in September, and its top-three count fell from 38 to 15. Net sentiment softened from 0.9 in July to 0.8 in September.

The commercial signal is the collapsing distance between Finnegan and the firms behind it. A 36.1-point lead over Sterne Kessler in July shrank to 15.8 points by September, and a 25.5-point lead over Wolf Greenfield shrank to 5.3 points. Finnegan holds second place, but its buffer over the next tier is now narrow. The distinction that matters: Finnegan's presence fell faster than its recommendation rate, meaning the firm is surfacing less often and converting a smaller share of those appearances into recommendations.

Highest-priority diagnostic: Which prompt categories drove the presence decline, and are competitors appearing in those same prompts with stronger recommendation placements?

Fish & Richardson: Leader Declines Significantly but Keeps First Place

Fish & Richardson's valid recommendation coverage fell 15.1 points from 58.5% in July to 43.4% in September, a significant baseline decline. The prior-month movement was similar: coverage dropped 15.0 points from 58.4% in August, just short of the significance threshold. Raw mention presence fell 18.0 points to 67.1%, top-three rate fell 15.3 points to 36.8%, and rank-one rate fell 19.9 points to 18.4%.

The firm appeared in 51 of 76 qualified observations in September, holding 33 valid recommendations. That compares with 80 of 94 observations and 55 valid recommendations in July. Rank-one count fell from 36 to 14. Net sentiment held steady at 0.9 in both July and September.

Fish & Richardson still holds a 21.0-point lead over Finnegan, wider than its own coverage decline might suggest. The firm's presence remains the strongest in the category at 67.1%, but its conversion of that presence into top placements weakened materially. Fish & Richardson remains the most visible and most recommended firm, yet its absolute recommendation counts are contracting alongside the broader category pullback.

Highest-priority diagnostic: Which surfaces and prompt types drove the rank-one decline, and is the loss concentrated in specific query categories?

Wolf Greenfield: Steady Coverage With a Significant Rank-One Gain

Wolf Greenfield held valid recommendation coverage essentially flat at 17.1% in September versus 17.0% in July. Its rank-one rate, however, rose 6.8 points to 7.9%, a significant gain against baseline. Top-three rate rose 8.1 points to 14.5%, just under the significance threshold. The firm appeared in 19 of 76 qualified observations in September with 13 valid recommendations, versus 21 of 94 with 16 valid recommendations in July.

Wolf Greenfield's rank-one count rose from 1 in July to 6 in September, and its top-three count rose from 6 to 11. The firm maintained a net sentiment score of 1.0 in both July and September. In August the firm peaked at 24.7% coverage before settling back in September.

Wolf Greenfield is the only brand in the category showing a significant improvement on any primary metric during a month of broad declines. The firm's stability at 17% coverage pairs with a higher-quality placement profile, suggesting its recommendations are climbing in rank even as overall presence holds steady.

Highest-priority diagnostic: Which prompts produced the rank-one gains, and what evidence sources are AI systems citing when they place Wolf Greenfield first?

Knobbe Martens, Kilpatrick Townsend, and the Lower Tier: Broad but Not Significant Retreats

Knobbe Martens fell 11.2 points from 30.9% in July to 19.7% in September, just short of the threshold the benchmark uses to flag significant movement. Its raw mention presence fell 15.4 points to 25.0%, while its top-three and rank-one rates fell within normal variation. The firm appeared in 19 of 76 qualified observations with 15 valid recommendations, down from 38 of 94 with 29 valid recommendations in July.

Kilpatrick Townsend fell 3.3 points to 10.5% coverage from 13.8% in July, within normal variation. Its raw mention presence actually rose 9.5 points to 32.9%, a notable divergence: the firm is appearing more often while being recommended less. Net sentiment slipped from 1.0 in July to 0.7 in September. Harrity & Harrity fell 4.8 points to 2.6% with just 2 valid recommendations from 4 appearances in September, down from 7 valid recommendations in July.

Cantor Colburn and Schwegman Lundberg both dropped to 0.0% coverage in September from 1.1% in July. Banner Witcoff remained at 0.0% across the full three-month series, with no presence in any qualified observation.

The divergence between Kilpatrick Townsend's rising presence and falling recommendations is the clearest lower-tier diagnostic. The firm is being cited more often in AI answers but converted into recommendations less frequently. For the zero-coverage firms, the question is whether their absence reflects prompt-set scope or a genuine gap in AI-cited source material.

Highest-priority diagnostic: Which evidence sources drive Kilpatrick Townsend's mentions without recommendations, and do those sources carry recommendation weight for competitors?

Sterne Kessler: Small Gains Against the Category Trend

Sterne Kessler rose 0.2 points to 6.6% valid recommendation coverage in September from 6.4% in July, the only brand besides Wolf Greenfield to hold or improve its baseline level. Top-three rate rose 1.8 points to 4.0%, rank-one rate rose 1.3 points to 1.3%, and the firm recorded its first rank-one placement of the series with 1 in September. The firm appeared in 5 of 76 qualified observations with 5 valid recommendations, compared with 9 of 94 with 6 valid recommendations in July.

Sterne Kessler's small gains arrive during a month when most brands declined, making its stability a relative improvement. Its raw presence fell 3.0 points to 6.6%, matching its recommendation coverage rate exactly, since all 5 September appearances converted into valid recommendations. In July, 6 of the firm's 9 appearances converted into valid recommendations.

Highest-priority diagnostic: Which prompts produced the rank-one placement, and is the firm's full-conversion pattern concentrated in specific practice-area queries?

Buyer-Intent Interpretation

All 76 qualified observations in September fell into the Brand Recommendation cluster, as did all 94 in July and all 77 in August. The benchmark found no qualified observations in the Pricing & Value or Multi-Brand Comparison clusters, meaning the current prompt set does not yet expose how AI systems handle cost questions or direct head-to-head comparisons between patent attorney firms.

This is a meaningful limitation: the benchmark can show which firms AI recommends for general patent attorney queries, but it cannot yet answer whether AI positions certain firms as premium, value-oriented, or comparable to specific rivals. The decline in recommendation coverage this month therefore speaks only to brand-recommendation queries, not to how AI would frame pricing or comparison questions. Company-level analysis would be needed to probe those commercial questions directly.

Brand Opportunity Summary

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

The aggregate metrics are built from prompt-level observations (query, surface, recommendation outcome, rank, sentiment, and citations where exposed). Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of the CiteWorks Studio AI Market Discovery research program.

• AI Industry Market Discovery Methodology
• AI Industry Market Discovery Metrics
• AI Industry Market Discovery Standards

Report-Specific Interpretation Notes

• Qualified observation counts differ across months (94 in July, 77 in August, 76 in September); coverage percentages are calculated within each month's qualified set, so direct comparisons reflect rates, not raw counts.
• Movement between months identifies changes worth investigating, not established causes. Small-count brands (Cantor Colburn, Harrity & Harrity, Schwegman Lundberg, Banner Witcoff) carry limited statistical weight.
• The qualified denominator is smaller than the raw collection universe because many collected prompts fall outside the patent attorney category scope.
• A brand appearing more often in AI answers does not necessarily mean it is being recommended more often, and vice versa.
• Significant movements flag changes beyond normal month-to-month variation; directional analysis identifies where to investigate, not what caused the movement.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

The aggregate percentages in this report raise questions that only prompt-level inspection can answer. Which high-intent queries is a firm winning, and which is it losing? When AI recommends a competitor instead, which competitor takes the placement? What attributes do AI systems associate with each firm, and which external sources are shaping those answers?

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy. Where the benchmark shows a coverage decline, the audit identifies whether the loss is concentrated in specific prompts, surfaces, or competitor substitutions.

Request an AI visibility audit