The short answerAcross 108 posts on one B2B blog, ranked lists were 18.5% of the output and drew 53.1% of Google impressions, 66.7% of clicks and 72.6% of AI citations. Like for like, among posts aimed at a tracked AI question, 27 of 28 pages that rank or compare named options were cited, against 3 of 7 guides and explainers.
The 2026 numbers at a glance
| Finding | Number | Source, sample and window |
|---|---|---|
| Ranked "best X" lists: share of posts against share of results | 18.5% of posts; 53.1% of Google impressions (2.87x), 66.7% of clicks (3.6x), 72.6% of AI citations (3.92x) | LoudFace blog, 108 posts; Search Console 25 September to 22 October 2026; Peec AI 10 to 23 October 2026 |
| Citation rate, like for like (only posts aimed at a tracked AI question) | 27 of 28 ranked or compared posts cited (96.4%), against 3 of 7 guides and explainers (42.9%) | LoudFace, 48 posts, Peec AI, 10 to 23 October 2026 |
| Explainers, the largest format | 32.4% of posts; 0.5% of citations; clicks per impression 0.59x the site average | LoudFace, 108 posts, same windows |
| Guides: clicks per impression | 1.55x the site average, the highest of any format; 1.5% of citations | LoudFace, 108 posts, same windows |
| Alternatives pages | 7 of 7 cited in 14 days, the highest median citations per post of any format; 6.5% of posts, 15.1% of citations (2.34x) | LoudFace, Peec AI, 10 to 23 October 2026 |
| How concentrated results are | On each measure, the top 11 of 108 posts took 65.5% of impressions, 67.9% of clicks and 79.6% of AI citations (mostly different posts on each) | LoudFace, same windows |
| Posts aimed at a keyword with recorded search volume | Median impressions 13.5x the other posts (5x outside list-shaped formats) | LoudFace, 21 of 108 posts, Search Console 28 days |
| Content marketers whose blog delivers strong results | 13.9%, an all-time low (the figure sat between 20% and 30% for 12 years) | Orbit Media, 1,042 content marketers, published 8 September 2026 |
How we gathered the data
What we measured. Every post on the LoudFace blog at loudface.co/blog, a B2B agency blog written for B2B SaaS buyers. On 24 October 2026 we read each live post's page and its Google and AI results. 147 posts were live. We kept the 108 published at least 28 days earlier (September 2024 to 25 September 2026, with 88 of the 108 from 2026), so every post has a full Search Console window.
Formats. Each post sits in one format, assigned automatically from its title, slug and page structure: list (a ranked "best X" list), alternatives, comparison ("X vs Y"), study (original data), decision (pricing, hiring, red flags), guide (how-to, playbook) or explainer (everything else). The label is set by fixed rules with no human judgment, and a few posts sit on a border. The mix: explainer 35 posts, list 20, guide 17, study 10, comparison 10, decision 9, alternatives 7.
Google. Search Console impressions and clicks for each post over the 28 days to 22 October 2026 (25 September to 22 October).
AI answers. Peec AI asks the 150 questions LoudFace tracks once a day on ChatGPT, Perplexity and Google AI Overviews. A citation is a post's URL cited in one of those answers, counted over the 14 days to 23 October 2026 (10 to 23 October). Peec only sees answers to tracked questions, so a post written for a tracked question is partly favoured by construction. We show the comparison like for like where it matters.
Links. Ahrefs referring domains, read for 50 of the 108 posts (recent posts and studies).
What we publish. Shares, multipliers and counts of posts. No raw totals.
Public research. We read the Content Marketing Institute, Orbit Media, Ahrefs and Semrush studies on their own pages and give each figure with its sample and date. Budgets and buyer behaviour sit in our earlier report, The State of B2B SaaS Marketing and AI Search, and we do not repeat them here.
How to cite this report: LoudFace, "The State of B2B SaaS Content Marketing (2026 Report)", loudface.co/blog/state-of-b2b-saas-content-marketing, data windows 25 September to 23 October 2026. You may quote any figure and reuse the tables with that attribution and a link.
Which content formats earn Google impressions and clicks?
Ranked lists do. They are 18.5% of posts and drew 53.1% of Google impressions, 2.87x their share of posts, and 66.7% of clicks, 3.6x their share. None of that depends on which AI questions we track, so it is the cleanest number in the report.
Every page built around named options (lists, alternatives and comparisons, 34.3% of posts) drew 66.2% of impressions and 71.4% of clicks together, most of it from lists (53.1% of impressions and 66.7% of clicks). Guides and explainers together are 48.1% of posts and drew 28.2% of impressions and 26.2% of clicks.
Explainers are the weak spot. They are the largest format at 32.4% of posts, and they took 18.2% of impressions (0.56x) and 10.7% of clicks (0.33x). When an explainer does appear, few people click: clicks per impression run at 0.59x the site average.
Andy Crestodina, writing up Orbit Media's 2026 survey, describes the same thing from the marketer's side: "search traffic to articles that rank for information-intent phrases has been completely torched." Most of our explainers answer information-intent queries, and informational queries are where AI Overviews started (more on that below).
Guides behave differently, and that matters for your plan. They are 15.7% of posts with 10.0% of impressions (0.63x), yet they took 15.5% of clicks, almost exactly their fair share (0.98x). Their clicks per impression run at 1.55x the site average, the highest of any format.
| Format | Clicks per impression against the site average |
|---|---|
| Guide | 1.55x |
| List | 1.26x |
| Explainer | 0.59x |
| Alternatives | 0.50x |
| Decision | 0.45x |
| Comparison | 0.20x |
A how-to that shows up for the right query still earns the visit. The base is small, so treat the guide result as a signal to test on your own site.
Which content formats do ChatGPT, Perplexity and Google AI Overviews cite?
Pages that rank or compare named options. Lists, alternatives and comparisons drew 92.6% of AI citations across the whole cohort, and guides and explainers drew 2.0%.
That gap is partly built in, and you should know why before you quote it. Peec AI counts citations only in answers to the questions we track, and those questions are about choosing agencies. 28 of the 37 ranked or compared posts are aimed at one of those questions, against 7 of the 52 guides and explainers. So the fair test is the like-for-like one.
Like for like. Take only the 48 posts aimed at a tracked question, so every post is equally visible to Peec. 27 of 28 ranked or compared posts were cited (96.4%), against 3 of 7 guides and explainers (42.9%). Decision pages sat between, with 5 of 7 cited. Studies split, with 3 of 6 cited. Ranked and compared pages were 58.3% of these posts and drew 92.7% of their citations. One confound remains: the questions differ by format, since list posts answer "best X" questions.
Alternatives pages are the most reliable unit we publish. All 7 were cited in the 14 days, and their median citations per post were the highest of any format, more than double the list median. They are 6.5% of posts and drew 15.1% of citations, 2.34x their share. They also turn the most retrievals into citations: when an engine reads an alternatives page, it cites it about once per read (1.03), against 0.75 for lists, 0.53 for comparisons, 0.51 for guides and 0.28 for explainers. 6 of the 7 are aimed at a tracked question, so the same caveat applies.
Independent research points the same way. Ahrefs analysed ChatGPT responses across 750 top-of-the-funnel prompts in software, products and agency recommendations (December 2025) and found that "best X" blog lists were 43.8% of all page types among the sources, the largest single type. Ahrefs also found ChatGPT leans on fresh pages: of 1,100 blog lists used as sources, 79.1% had been updated in 2025 and 26% in the previous two months. 35% of the cited lists sat on low-authority domains, so authority is not the gate many teams assume.
A second Ahrefs study of 17 million citations (July 2025) found AI assistants cite content that is 25.7% fresher than what ranks in organic results. Google AI Overviews was the exception, citing content 16 days older on average than the organic results.
For the full picture of which sources AI engines pull from in this category, see What AI Actually Cites: The B2B SaaS AI-Citation Benchmark.
How concentrated are content results?
Very. On each measure, the top 10% of posts (11 of 108 posts) took 65.5% of impressions, 67.9% of clicks and 79.6% of AI citations. They are mostly different posts on each measure, but every measure is highly concentrated, and the posts at the top share a shape.
Of the 11 most-cited posts, 7 are lists, 3 are alternatives pages and 1 is a comparison. Of the 11 posts with the most impressions, 7 are lists.
What this means for planning: a content program is a portfolio with a few big winners. You cannot predict every winner, but you can stack the odds by format, and you should expect most of the return to come from a small group of pages you then keep fresh.
Does aiming at real demand matter more than format?
For Google, yes. Posts aimed at a keyword with recorded search volume (21 of 108) had median impressions 13.5x those of the other posts. That is partly a list effect, so we checked outside the list-shaped formats: the gap holds at 5x across the 8 posts there. Median age is about the same in both groups (142 against 140 days), so the older-posts explanation does not fit.
The volume-aimed posts were also cited more often: 14 of 21 (66.7%) against 27 of 87 (31.0%). Outside the list formats, 4 of 8 were cited against 9 of 63. The same Peec construction caveat applies here, since a post aimed at real demand is more likely to match a tracked question.
Aiming a post at a tracked AI question is a different lever. Those posts were cited 79.2% of the time against 5.0% for the rest, but that is mostly construction, because Peec only counts tracked answers. On Google impressions, the aimed posts led overall and trailed outside the list formats, so the impressions gap is a list effect, not a question effect.
The working rule: pick the demand first, then pick the format that wins for that demand.
Does original research still pay off?
The surveys and our data disagree, and both deserve a fair hearing.
The survey side. Orbit Media's 2026 survey of 1,042 content marketers found that "fewer marketers now publish new research studies, despite the fact that it improves the likelihood of performance by 50%." Orbit also names "Influencer collaboration" as "the single best predictor of success in the dataset." In CMI's 2026 research, only 15% of B2B marketers name research and insights among their top three areas for increased investment.
Our side. Studies are 9.3% of our posts (10 of 108) and drew 0.3% of impressions and 1.1% of citations in the windows above. 3 of the 10 reached a modest bar for links or AI citations. Only 2 of the 10 have any referring domain, against 11 of the 50 posts Ahrefs read. Among posts under 120 days old, the 17 carrying three or more first-party data signals did no better: lower median impressions, and 7 of 17 cited against 14 of 30.
Why both can be true. The two sides measure different things. Orbit asks marketers whether their program delivers strong results, a self-reported judgment over a year. We count 28 days of clicks and 14 days of citations. A study earns its keep through links and mentions, and those arrive slowly. Our dataset does not record whether each study was pitched to a publication, and most of them have earned no referring domain so far.
So the lesson from our data is narrower than "research does not work." Publishing a study is half the job. It earns links only when it reaches the people who write about the topic. A study needs the same effort after publishing as before it. If you want to see what that looks like, our list of original research and data study agencies for B2B SaaS and our study of digital PR for AI citations cover the distribution side.
Is AI making B2B content better or just faster?
Mostly faster. The Content Marketing Institute's B2B Content and Marketing Trends: Insights for 2026 surveyed 1,015 B2B marketers between June 24 and August 14, 2025. Among those using AI to create content, "87% say productivity has improved" and 58% say content quality has improved, but only 39% say content performance has improved. 12% say content quality decreased.
Orbit Media found the same split in 2026: 92.4% of content marketers now use AI for blogging, and "Using AI does not improve content performance." The average post is 1,312 words and takes 3 hours, 20 minutes to write.
Meanwhile, results are slipping. 13.9% of Orbit's respondents say their blog delivers strong results, the lowest in the survey's history, after 12 years between 20% and 30%. 18% say they do not know whether the blog delivers results at all.
CMI's numbers are kinder but not by much. Of marketers with a content strategy, 13% say it significantly improved and 48% somewhat improved. Their top three content challenges are creating content that prompts a desired action such as conversion (40%), resource constraints (39%) and measuring content effectiveness (33%).
When CMI asked effective teams what moved the needle, 65% named content relevance and quality and 53% team skills and capabilities. Budget allocation came in at 20%. AI lowers the cost of a post. It does nothing for a post written in the wrong format for the wrong demand, and in our format data, that is where our own blog's results were won and lost.
Are AI Overviews taking the clicks?
The evidence conflicts, and the method decides the answer.
The Semrush AI Overviews Study (10 million+ keywords, refreshed December 15, 2025) tracked how often AI Overviews appear: 6.49% of keywords in January 2025, nearly 25% in July, then 15.69% in November 2025. The mix shifted too. Informational queries were 91.3% of AI Overview triggers in January and 57.1% by October, as commercial and transactional ones grew.
On clicks, Semrush compared the same keywords before and after they showed an AI Overview and found the zero-click rate fell slightly, from 33.75% to 31.53%. Semrush notes that keywords with AI Overviews have higher zero-click rates on average, which may reflect their informational intent rather than the Overview itself.
Other studies, built on user panels and on the click rate of the top result, found fewer clicks when an AI summary appears. We covered those in our earlier report. The difference is method: a same-keyword zero-click rate against a click-through rate on one result or a panel of users.
For a content plan, the safe read is this. AI Overviews are moving from informational queries toward commercial ones, which is where B2B SaaS content earns its money. Write pages that win the citation inside the Overview, not only the blue link below it.
Do comparison tables, short answers and length change the result?
Comparison tables. Every list-shaped post we publish (37 of 37) carries a comparison table. Of the 23 posts with no table, none was cited. But 21 of those 23 are not aimed at a tracked question, and their median age is 224 days, so we cannot separate the table from the rest. Keep the habit. The cause is unproven.
A short answer at the top. 87 of 108 posts carry one. Posts with it were cited 39.1% of the time against 33.3% without, and 19.6% against 13.3% outside the list formats. Among posts with one, cited posts' answer boxes hold a median of 3 facts against 1. That is weak and directional. It still costs nothing to write, and our guide to structuring content for AI extraction shows how.
Length. Our median words per post by format: list 3,738, alternatives 3,035, decision 2,965, comparison 2,532, guide 2,495, study 2,035, explainer 1,984. Orbit's average post is 1,312 words. Length follows format here. The formats that win run long because they compare many named options in detail, so do not pad a post to hit a number.
Where should a 2027 content budget go?
Content keeps its place in the budget. In CMI's 2026 research, owned media is one of the top three areas for increased investment: 45% on AI-powered marketing tools, 33% on events and experiential marketing, and 32% on owned media (content assets, website, blog, email). The question for 2027 is order: what you ship first.
Our data suggests this order of moves:
- Pages that rank or compare named options, aimed at real buyer questions. Lists carry AI citations and Google clicks. Alternatives pages carry citations well above their share, and all 7 were cited. Alternatives and comparisons both draw fewer clicks than their share.
- Demand-led guides. Aim each one at a keyword people search. Guides convert impressions to clicks better than any other format we publish.
- Original research, with a distribution budget attached. Plan the pitching before you publish. A study that reaches publications can become the page others link to and engines quote.
- Explainers last, and only where a real query needs one. In our data they were 32.4% of posts and drew 10.7% of Google clicks, a third of their share.
Nothing on this list is banned. A blog made only of lists would chase the numbers in this report and lose the reader's trust in the brand. Lists earn the attention. The rest of the mix earns the trust and the links. Set the mix from your own measured results each quarter and move it as the results move.
Orbit's data backs breadth: marketers using six or more of the eight strategies it lists are nearly three times as likely to report strong results, and none of the respondents do all eight.
If you want your AI visibility measured per engine on your own site, that is how we run SEO and AEO programs for B2B SaaS. To see where your brand stands today, run a free AI visibility audit: it checks where your brand stands across ChatGPT, Claude, Gemini and Perplexity versus competitors.
What are this report's limits?
- One site, one niche, one window. These are 108 posts on one B2B agency blog, measured over 28 days on Google and 14 days in AI answers.
- The tracked questions are about choosing agencies. They reward pages that name and compare options. A post written for a tracked question is partly favoured by construction, which is why the like-for-like comparison is the one to quote for citations.
- Formats are automatic labels. They come from title, slug and page structure, and a few posts sit on a border.
- These are correlations from one blog. A format's result also reflects the keywords and questions we chose for it.
- Links are partial. Ahrefs referring domains were read for 50 of the 108 posts.
- A citation is not a mention. We count a post's URL cited in an answer. That is not the same as the brand being named in the answer text.
Sources
- Content Marketing Institute, B2B Content and Marketing Trends: Insights for 2026 (with MarketingProfs, sponsored by Storyblok; 1,015 B2B marketers, fielded June 24 to August 14, 2025; published October 8, 2025)
- Orbit Media, Blogging Statistics 2026 (Andy Crestodina; 1,042 content marketers, 24 questions; published September 8, 2026)
- Ahrefs, Do Self-Promotional 'Best' Lists Boost ChatGPT Visibility? (Glen Allsopp, December 4, 2025)
- Ahrefs, AI Assistants Prefer to Cite 'Fresher' Content (17 Million Citations Analyzed), July 28, 2025
- Semrush, AI Overviews Study (refreshed December 15, 2025)
- LoudFace first-party data: Search Console, 25 September to 22 October 2026; Peec AI, 10 to 23 October 2026; Ahrefs referring domains, read 24 October 2026
