How Liberty Mutual scores now
Twelve signals across the three pillars, this month. Each bar shows how the brand compares with the category leader, which sits at 100. The dark bar marks a signal where this brand leads the set.
At a glance
The strategic read on the signals above — where Liberty Mutual sits, what is holding it back, and where the openings are. Who else is moving in the set is broken out under the trajectory chart on the next tab.
5th of 8 on 62, four points clear of Nationwide and 16 behind State Farm. Chosen is its best relative pillar (4th, carried by a 4.3 Trustpilot score no rival gets near); Understood is the weakest (7th). The AI gap is sharpest in ChatGPT: 3.8% share of voice against 18.6% in Google AI Overviews.
AI share of voice 13.2%, 6th of 8, against Progressive's 56.7%; 3.8% in ChatGPT.
AI sentiment net +9.1, the lowest in the set (State Farm +37.4). Models list Liberty and quote its price; they rarely praise it.
Thinnest press coverage: six live press-set sources, and MarketWatch, Bankrate and Policygenius lost this year.
Brand demand of 360K a month, a fifth of Progressive's 2.0M.
Get named in the ranking tables AI reads. NerdWallet's and U.S. News's best-car-insurance pages are cited in 107 and 99 Liberty answers; a stronger place in them lifts every model.
Give the models a reason to choose Liberty. They repeat the $950 bundling saving and the high price; carry the 4.3 Trustpilot rating and accident forgiveness into the sources they cite.
Build on the search lead. Liberty is the only brand growing top-3 rankings (+16%) and leads on category capture (71%); its teens, cheap-car and same-day pages are already cited.
Organic Discovery Score over time
Overall score by month across the set. Where is the field tightening or splitting apart?
Who is moving, and why
The specific brands driving the trajectory chart above — what changed, and the piece of content, product event or market shift behind it. Write one card per brand that materially moved (rising / sliding / outlier / holding), with the actual metric change and the named driver.
The only brand in the set adding top-3 rankings over the year (rivals -1% to -39%), and its top-10 footprint held best (-4% against -12% to -41%). Organic traffic still fell 26% with the category, so the gain is in position, not yet in visits.
The AI leader (56.7% share of voice, 59.2% in ChatGPT) is the biggest organic faller in the set: non-brand traffic halved (3.7M to 1.5M) as its quote and comparison terms moved under AI Overviews.
Still #1 on five of eight head terms and the largest non-brand traffic in the set (3.4M), but brand demand fell 18% and its AI sentiment is the second weakest (net +15.6).
Leads on link base (49.6K referring domains, DR 90), brand traffic and AI sentiment. Its traffic fell 24%, in line with the category, and 77% of it is brand traffic.
Held its rankings while larger rivals lost them, and carries 40% AI share of voice with the second-best sentiment. The smallest traffic fall of the big four.
Four points behind Liberty Mutual. The only brand growing brand demand, with eleven live press-set sources including Bloomberg (188 links) and WSJ.
Signal by signal, against the leader
Twelve signals side by side. Now is where we sit against the top rival this month. 13m indexed move is how our position against the set has shifted over the year — positive when we gain relative ground, even if raw values slide.
Why raw and indexed can disagree. Every signal is indexed to the leader each month (leader = 100). If rivals decline harder than we do, our indexed position rises even when the raw number falls. The pillar scores are indexed; the raw column shows the absolute change over the same window.
Non-brand keyword coverage
Ranking positions across a curated non-brand keyword set, grouped by buyer intent. Which brand owns which corner of the category.
Signals moving the wrong way
The tracked signals whose raw values have slid over the thirteen months. Even where the indexed position holds against the field, a falling raw number still needs fixing.
Category context
SERP pattern · AIOs · Trustpilot floor · non-scored competitors
The structural shape of US car, home and renters insurance — who owns the SERP, where AI Overviews show up, which brand names inflate raw AI SoV, and the aggregators that sit outside the scored set but capture most of the demand.
How AI sees us
How AI sees Liberty Mutual
Two different questions the models answer: how often a brand is named, and how clearly and warmly it is described. The first is visibility, the second is interpretation. Both are measured across six AI surfaces (ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Copilot) on brand-free insurance prompts.
AI Share of Voice, by model
How the models actually answer
Four representative responses from four models, all to prompts that name no brand. Rival names in charcoal, Liberty Mutual in teal.
What we say vs what others say
Liberty Mutual's positioning from its own site, side by side with how the AI models and its reviewers describe the brand. The diff below shows which claims land on both sides, which are only in the brand's voice, and which external framings the brand isn't addressing.
Where AI's picture comes from
The sources the models cite when they discuss each brand. This is where earned, community and creator presence shows up — and where leaning on your own site becomes a fragility, because the models have less independent material to trust.
Editorial ranking tables (NerdWallet, U.S. News, MoneyGeek) and comparison marketplaces (Insurify, The Zebra, LendingTree) write roughly 70–80% of what AI reads about every brand here. Owned pages are a small slice everywhere, but Liberty Mutual's is the largest in the set and its site is the single most-cited domain in its own answers. Shares are of the top 60 cited domains per brand; "Other" is rival carriers, regulators, banks and retailers.
Who the press recommends
Live dofollow coverage from twenty titles that shape US insurance buying, ten mainstream nationals and ten insurance and personal-finance specialists, weighted by each source's authority. When the press and the ranking tables keep naming a brand, buyers are being pointed towards it.
Press endorsement
Where the audience actually is
Cited Reddit threads · AI-cited YouTube creators · prompt themes
The Reddit threads, YouTube channels and prompt themes AI cites when it discusses car, home and renters insurance.
Cited Reddit posts & threads
YouTube creators cited in AI answers
Prompt themes & what they ask
Where Liberty Mutual is losing ground
The specific constraints behind the score, zoomed in — a gap to a named rival, or a signal moving the wrong way. This is the shortlist a Discovery Roadmap turns into work.
The gap
What to fix
The weakest signals, turned into things to work on — the starting point for a roadmap.
How the score is built
How the Organic score is put together — the twelve signals, the weights, the adjustments and floors, and what is tracked month to month.
The twelve signals
Each signal is read from the Ahrefs web app (US) or Trustpilot, indexed against the leader in the set each month, then rolled into its pillar. Pillars combine into the headline score by the weights below. "Tracked" signals have monthly history; "point-in-time" ones are held at their latest reading.
| Pillar | Signal | What it reads | Basis |
|---|
Reading it: level, direction, consistency
Three properties, not one
Level is how good it is — read against 100, not only against rivals; leading a pillar the whole category is bad at signals an opening, not a solved problem. Direction is which way it is heading — treat the score as the lagging indicator and the trend as the leading one. Consistency is how steadily it holds — measured as variation around the brand's own 13-month average; presence built and withdrawn has to be re-earned, and machine surfaces reward sustained signal over bursts.
Relative, and it moves under you
Each signal scores value ÷ category-best × 100 each month. Relative is the important word: if rivals improve and you stand still, your score falls even though nothing about you changed. The latest partial month is held out so the trend reads clean.
Weighting, the brand/non-brand split, and the caveats
Weights
Understood-led: Found 30 · Understood 45 · Chosen 25. Within Found, AI Share of Voice and non-brand traffic carry 35% each and link authority 30%; within Understood, AI clarity 25%, anchor relevance 24%, topic footprint 20%, category capture 19% and AI sentiment 12%; within Chosen, brand demand 30%, top-3 rankings 25%, press endorsement 23% and Trustpilot 22%. Multiply the two to get a signal’s share of the headline score — AI clarity is the heaviest single input at 11.3%, AI sentiment the lightest at 5.4%. Switch to "Even" in the control bar to test how the pillar weighting changes the picture.
Twelve signals, not fourteen
Two pairs measured one thing twice. Domain Rating is computed from the referring-domain graph, and the two indices typically correlate above r=0.8, so they are averaged into a single link authority signal. Brand traffic is roughly brand demand multiplied by ranking for your own name, so brand demand absorbs it, 65/35. Nothing is lost; the model stops paying twice for the same fact, and brands that run on brand-ness alone no longer get credited for it in two pillars. AI sentiment carries 12% of Understood rather than 25% for a different reason: it is an analyst read of sampled responses rather than a measured quantity, and weight should track how reliably a signal is measured. Re-weighting correlated signals moves less than it looks like it should — topic footprint and top-3 rankings routinely correlate above r=0.95 — so expect most positions to hold and read any rank change against the size of the gap it crossed.
Brand vs non-brand traffic
Found counts these separately: being found for your own name is not the same as being found for the category. The split is measured from each brand's non-branded organic keywords across the window. Category capture in Understood reads the same non-brand share as a percentage — absolute reach in Found, mix in Understood.
The AI layer
AI Share of Voice, sentiment and representation clarity come from Ahrefs Brand Radar (Explore) across six prompt surfaces — ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity and Copilot (Grok is paused and the Google search-query views are excluded). Scope: US, every brand matched as a company entity (its name plus "<brand> insurance"), and every prompt that names any brand in the set excluded. Sentiment is Ahrefs' positive/negative classification of the brand's mentions, shown as 50 + 1.5 × net; clarity is how consistently share of voice holds across the six surfaces.
Held at their latest reading
- The AI signals, anchor relevance and Trustpilot are shown at their most recent reading rather than as a monthly trend, so they set the level but do not drive month-to-month movement.
- The long-tail keyword count is unreliable in the most recent months, so topic footprint tracks the stable top-10 count rather than the total.
This run: sources, route and what changed
Route
Every search, link, SERP and demand signal was read from the Ahrefs web app in Chrome on 21 October 2026, and the AI layer from Ahrefs Brand Radar in the same session: 0 Ahrefs API units, one data provider throughout. History runs September 2025 to September 2026; October 2026 is a partial month and is held out.
Signals and sources
- Traffic, referring domains, DR, top-3 and top-10 counts: Site Explorer, subdomains, US, monthly, "best links: no spam".
- Brand vs non-brand split: Ahrefs' monthly branded-traffic series. Ahrefs files Progressive's and Farmers' own brand under "other brands", so for those two the own-brand series is branded + other-branded; checked against a brand-term keyword pull (top 50 rows per variant).
- Brand demand: Keywords Explorer volume history for each brand's main navigational term ("liberty mutual", "state farm", "allstate", "progressive", "geico", "nationwide", "farmers insurance", "travelers insurance"); the bare "farmers" and "travelers" are mostly non-brand.
- Anchor relevance: share of the top 50 anchors by referring domains that name the brand. State Farm's top 50 is sitewide boilerplate from a partner network, so it is unscored and takes the set median.
- Press endorsement: live dofollow links from 20 publications (Forbes, WSJ, NYT, CNBC, CNN, USA Today, Business Insider, Washington Post, Reuters, Bloomberg; NerdWallet, Bankrate, MarketWatch, Consumer Reports, ValuePenguin, MoneyGeek, Investopedia, U.S. News, The Zebra, Policygenius). Bankrate dropped its links to seven of eight brands between 23 and 26 August 2026, and MarketWatch to six of eight in April 2026; that pattern reads as link re-routing on those sites rather than lost coverage.
- Non-brand keyword coverage and SERP landscape: 40 category terms and 8 head terms, each checked on the latest Ahrefs SERP (top 20, 12 to 21 October 2026).
- Trustpilot: ratings and review counts as read on 20 October 2026.
- Audience: SparkToro report 2df6d08c3040 (US adults who research and buy car, home and renters insurance), prompt themes.
Read with care
- Liberty Mutual's referring domains step up from 13.5K to 16.1K in May 2026 with no matching change at any rival. The scored figure is Ahrefs' "no spam" count, but the jump looks more like injected links than earned coverage; the 41% rise is not read as a PR result.
- Perplexity carried a data-provider outage notice in Brand Radar from 19 October 2026.
- The prompt excerpts come from Eric Hoover's 20 October Brand Radar sample (5,000 Liberty responses); figures in this report come from the 21 October re-read.
Against the earlier version of this report
The first version mixed DataForSEO (referring domains, domain rank, anchors, press, SERP, demand) with Ahrefs (traffic, AI). This version reads everything from Ahrefs, uses the v16 template's twelve signals, scores AI on prompt surfaces only, and matches every brand as a company entity. The result moves Liberty Mutual's AI share of voice from 16.2% to 13.2% and its press finding from "no mainstream links" to "thinnest coverage, with live Forbes Advisor, NYT, Business Insider and Bloomberg links".
Adjustments, floors and category scoping
AI SoV common-word adjustment
Four names in this set are common words (Progressive, Nationwide, Farmers, Travelers). Rather than text matching, every brand is matched as a company entity, which removes the everyday uses at source: "all entities containing Progressive" returns 226K responses (it catches progressive lenses), the Progressive company entity 67K; "Farmers" returns 187K, the Farmers company entity 14K. No ratio adjustment is applied, so no brand carries an "adj" flag.
| Brand | Text / all-entity match | Company entity (scored) | Why |
|---|---|---|---|
| Progressive | 226K responses | 67K responses | Common word (lenses, politics) |
| Farmers | 187K responses | 14K responses | Common word (farmers markets) |
| Nationwide | 102K responses | 49K responses | Common word; UK building society |
Trustpilot volume floor
A rating is only scored when a brand carries at least 500 lifetime reviews. Below that, the rating is carried in the report but does not drive the Chosen pillar — small samples are too noisy for indexing.
Category scoping is mandatory
No tracked Brand Radar prompt set exists for this competitor set. Three saved Explore views exist for Liberty Mutual (General non-brand, Home, Car/Auto), but they cover six brands and use text matching, so they were not used. This run uses the brand-excluded wide scope with no topic filter, so the score reflects total AI presence on brand-free prompts.