LinkedIn's 60% Non-Brand Traffic Claim: What It Said, Changed and Left Out

LinkedIn's 60% non-brand traffic figure is usually retold as proof that LinkedIn lost most of its B2B search traffic. The first statement of it, on 28 January 2026, was narrower: "across the industry", in "a subset of B2B topics". By August the same figure described "our own marketing sites". No version carries a baseline, a topic list or a date range.

The short answer: LinkedIn published a ceiling, not a measurement. It says non-brand traffic fell by up to 60%, first across a subset of B2B topics industry-wide and later on its own sites. It has not published a baseline, topic list, date range or any quantified result from its response.

The 60% claim, traced across 3 LinkedIn posts

LinkedIn has stated the figure three times on its own marketing blog. Each time a different author attached it to a different population. The number stayed still while the claim around it moved, and that movement is the first thing a careful reader should notice.

28 January: the industry, and a subset of topics

The first version sits in a post by Inna Meklin, with Cassie Dell credited as co-author. Under a section on why the old playbook fell short, it reads: "Across the industry, non-brand, awareness-driven traffic declined by up to 60% across a subset of B2B topics". The next line says "Click-through rates softened while rankings stayed stable".

Neither line has a footnote or a link. The surrounding text does frame the piece as LinkedIn's own experience, and it dates the impact to "early 2025". The sentence itself still names the industry. "Our traffic" appears nowhere in the post.

10 February and 11 August: LinkedIn's own sites

Thirteen days later, Brooke Weller linked back to that post. Weller wrote: "At LinkedIn, we've seen the impact firsthand, with B2B non-brand keyword traffic dropping up to 60%", after AI Overviews launched. "Awareness-driven" had become "keyword". The subset had gone.

On 11 August, Davang Shah, a LinkedIn marketing VP, went furthest: "our own marketing sites have seen non-branded search traffic decline by up to 60%". I treat that as the most authoritative statement available, because it is the most specific about whose traffic this is. It is also the furthest from the only version that stated a limit.

Trade coverage flattened it further. PPC Land's February headline said LinkedIn was abandoning traditional SEO after a "60% traffic loss". LinkedIn's own January words were that its "established playbook wasn't broken but it needed updating". Watch for two unrelated 60% figures in the same material as well. LinkedIn cites a study that about 60% of US and EU searches end without a click. It also says roughly 60% of LinkedIn content citations go to articles.

LinkedIn's disclosure ledger: 7 statements, 0 fully specified

LinkedIn presents this series as evidence, and its January post closes on the promise of data-driven results. So I tested it the way this site tests any vendor claim. I took each statement about traffic or test performance and asked whether it carries a baseline, a named sample and a time window.

Zan Digital reading of the three posts, retrieved 8 November 2026. A statement passes only with all three of baseline, named sample and window. The guide PDF and the gated August guide were not assessed, as explained in the limitations section.

The count is 0 of 7. "Triple-digit" is the most revealing entry, because it reads like a number and behaves like a range. It covers everything from 100% to 999%, a tenfold spread, on a starting base LinkedIn does not give.

LinkedIn's citation rank depends on whose study you open

LinkedIn's standing in AI answers gets the same loose handling. The January post calls LinkedIn the "#2 most-cited source". A Semrush study of 325,000 prompts from January and February 2026 did rank it second, appearing in 11% of AI responses on average. Semrush says it collaborated with LinkedIn on the work.

Meltwater's May 2026 study of 9.5 million citations, described as "our new research with LinkedIn", also ranks it second, with 0.53% of citations against YouTube's 1.52%. Those two figures use different units, so 11% and 0.53% cannot be compared. The August post then calls LinkedIn the "#1 most-cited domain for professional queries", citing Profound. The two second-place rankings both came from studies run with LinkedIn, and a reader should weigh that.

What LinkedIn says it changed after the traffic decline

LinkedIn's account of its own response is the most useful part of this series. It is operationally specific even where the outcomes are not. The table separates what the company says it did from the evidence it published that the action worked.

The measurement rebuild is the most concrete change

Three of the eight actions are measurement plumbing, and they are the ones you could reproduce next month. A separate channel for LLM traffic costs almost nothing. Reading CMS logs for bot behaviour needs a developer for a day. LinkedIn also says plainly what the rebuild has not solved: "We simply couldn't quantify how visibility within LLM responses impacts the bottom line". That sentence is the most credible one in the series, because it costs the author something to write it.

Owned content is the one tactic given a direction

LinkedIn ranks one lever above the others: "owned content delivering the fastest and most scalable gains so far". That is notable coming from a company that sells distribution on its own platform. It would be more useful with a number attached, and it is still a directional finding rather than a test result. The practical reading is that the content you already control is a sensible first place to work.

What LinkedIn recommends, as distinct from what it reported

A case study and a playbook are different documents, and this series moves between them without marking the join. If a tactic appears as a recommendation with no reported effect behind it, treat it as informed opinion.

The owned-content guide is advice with numbers in it

The February announcement promises guidance on writing, technical signals, structure and media. It calls these "current recommendations" that LinkedIn expects to change. PPC Land's summary lists opening blocks of 30 to 80 words, sentences under 20 words, Grade 9 to 11 readability and meta descriptions of 140 to 160 characters. Those figures look like test outputs. In the material I could verify, they are presented as guidance.

Schema is the case to examine. The February post says to "implement schema markup (such as FAQPage) to improve AI discoverability". PPC Land quotes the guide as calling that markup "currently the highest-impact markup for increasing LLM discoverability in our internal testing". No lift figure accompanies it there. The independent tests of schema and AI citation point the other way, so I would hold this claim to a higher standard before funding it.

The recommendation with a commercial interest

The August post tells you to publish long-form articles on LinkedIn, post 2 to 3 times a week and turn employees into regular posters. Its closing advice is one in-depth article and two supporting posts on the platform. The accompanying guide sits under LinkedIn's advertising resources path. The body of the post never states that the publisher profits when B2B marketers move attention and budget to LinkedIn.

The advice may still be good. LinkedIn's advertising business depends on B2B budgets, though. A seller recommending its own channel is a Tier 3 source for that recommendation under the standard this site applies. In this one case, I would want third-party evidence before moving spend.

Whether LLM traffic can replace lost non-brand organic traffic

The January post sets falling non-brand traffic beside triple-digit LLM growth, which invites you to net one against the other. LinkedIn has not published the LLM share of its traffic, so the sum cannot be done from its numbers. It can be bounded. Assume 100 non-brand visits on affected topics before the decline, take the 60-visit loss at LinkedIn's ceiling, and test a range of starting LLM volumes.

Zan Digital model, not LinkedIn data. Share replaced = starting visits x growth / 60. Multiple = (60 + starting visits) / starting visits. The 60-visit loss is LinkedIn's stated maximum, so a typical topic would need less.

The 1.4 row comes from the only comparable published denominator I could verify. An analysis of 94 ecommerce stores covering January to December 2025 found 135,000 ChatGPT sessions against 9.46 million organic sessions, or 1.4%. It is ecommerce and covers ChatGPT alone, so it calibrates the range without describing LinkedIn.

The usual reply is that LLM visitors convert better, and LinkedIn does say it can track how they convert, without giving a rate. In the 94-store panel, ChatGPT sessions converted 31% better than non-branded organic. Apply that edge to the 1.4 row at +100% and the replaced share moves from 2.3% to about 3.1%. Our breakdown of AI referral conversion claims explains why that multiple is smaller than the one usually quoted. A better rate on a small base remains a small number.

Where this reading of LinkedIn's disclosure is weakest

Three objections deserve a straight answer, and the first could undo the headline finding.

January's wording may just have been loose

The likeliest explanation for the drift is not that LinkedIn changed its story. The January bullet sits in a passage about LinkedIn's own experience, followed by "The loss was undeniable". The authors may have had internal data in hand and written "across the industry" to soften it. If so, the later wording is a correction and the figure was always LinkedIn's. Nothing published can settle that, which is the reason to quote the wording and date instead of asserting one version.

AI Overviews may not be the only cause

LinkedIn attributes the decline to AI search, and independent data supports the mechanism. Ahrefs' February 2026 update, built on 300,000 informational keywords and aggregated desktop Search Console data, associated AI Overviews with about 58% lower position-one click-through rate. Our analysis of zero-click reporting covers the wider pattern. The same Ahrefs table also shows position-one CTR on keywords without AI Overviews falling from 0.076 in December 2023 to 0.039 in December 2025, about 49%. Ahrefs puts that down to other search page features. So part of any non-brand decline in this period may have nothing to do with AI answers. LinkedIn has not said how it isolated the effect.

A marketing blog owes you no methodology, and I did not read everything

LinkedIn is not a filer disclosing its marketing-site traffic, and no rule requires it to publish a denominator. Most companies publish nothing at all, so this series is more candid than the norm, and I give it credit for that. My ledger also covers three blog posts. I could not extract text from the owned-content PDF, I did not read the full gated August guide, and Search Engine Land's coverage returned an access error. A quantified result in either guide would change the count in the ledger, and I would update it.

What to copy from LinkedIn's B2B content strategy

Copy the measurement rebuild and be sceptical of the content checklist. The first has a logic you can check on your own data. The second is unquantified advice from a publisher with a stake in where you post.

Keep your reporting split by query type. LinkedIn's most useful observation is that clicks fell while rankings held. A blended organic dashboard would have shown a ranking-neutral decline with no explanation. Separating non-brand from brand, by topic cluster, makes the loss legible before anyone proposes a fix.

Name the gap in your own attribution as LinkedIn did. If you cannot connect citation visibility to pipeline, say so in the review and report it as a leading indicator. A citation share chart presented as a revenue proxy is the mistake this whole category is drifting toward. I would rather show a board an honest unknown than a precise proxy.

Above all, do not put "LinkedIn lost 60% of its traffic" in a deck. If you need the citation, quote the sentence, name the date and say "up to". The stronger version of this argument does not need the exaggerated one.

Frequently asked questions

Where to start this week

Open Search Console and export 24 months of non-brand queries, grouped into your five largest topic clusters. For each cluster, chart clicks, impressions and average position on one line. Any cluster where clicks fall and position holds is your own version of LinkedIn's finding, with the denominator LinkedIn did not publish.

Then add one number to your next report: LLM visits gained divided by non-brand visits lost, for the same clusters and months. That replacement ratio will settle the "AI traffic is making up for it" conversation faster than any citation chart, and it takes under an hour to build.

Related reading