- 01Seed niche creators instead of buying ads
- 02Publish a real technical opinion, not a launch post
- 03Build one landing page per buyer segment, not a blog
- 04Run paid ads as a first-person message, not brand copy
01 Snapshot
The one-glance read on who they are and how they grow. Each point is verifiable from the receipts above.
An AI research assistant that automates literature reviews and evidence synthesis for scientists, spun out of nonprofit Ought and now pushing into pharma and life-sciences enterprise buyers, running on a $9M seed round raised from Fifty Years in 2023.
Organic-led, powered by founder technical content and global creator UGC, with a discrete enterprise-paid overlay for life sciences.
~1.73M monthly visits as of 2026-07-12; the founder's X account has 3,289 followers and 1,001 posts; LinkedIn's ad library shows 19 total ads with 4 active as of 2026-07-12.
As of 2026-07-12, the active LinkedIn ads center on conference and webinar moments (three are distinguishable in the library: a JPM26 networking mixer, a webinar recap, and a Bio-IT World causal-modeling talk) rather than evergreen product demand-gen creative.
Elicit runs two nearly disconnected motions on one product: unpaid, multilingual student-creator UGC feeds free-tier top-of-funnel volume, while a separate founder-and-cofounder-voiced paid/outbound track sells the same tool into pharma medical-affairs teams.
02 How it unfolded
The order the channels came online. Sequence is strategy: what they did first, and what they layered on once demand existed.
03 Traffic & channel mix
Estimated demand, the channel split behind it, and the keywords and referrers doing the work. Directional modeling, not audited analytics.
Traffic sources
Top search keywords
| Keyword | Volume | Weight | CPC |
|---|---|---|---|
| elicit | 201k | $0.79 | |
| elicit ai | 51k | $0.80 | |
| ellicit | 7.0k | $1.29 | |
| elict | 1.7k | $1.99 | |
| ai for research | 8.6k | $1.50 |
Top referring sites
Closest competitors by audience
Elicit pulls an estimated ~1.73M monthly visits, down 23.4% over the last 3 months. Direct traffic dominates at 54.1%, consistent with a habitual, bookmarked/logged-in user base rather than discovery-driven visits; Search Organic follows at 30%, and a notable 5.5% comes from Gen AI referral sources, meaning AI answer engines are themselves sending Elicit traffic, a self-reinforcing signal for a company whose product is AI-powered search. The remainder, Referrals, Social Organic, Email, Search Paid, Display Ads, and Social Paid, totals under 12% combined and confirms paid acquisition is not doing meaningful work here. The clearest ranking terms are the brand terms themselves: "elicit" pulls 200,620 monthly searches and "elicit ai" pulls 51,370, while the category term "ai for research" (8,590/mo) shows Elicit is also chasing generic-intent demand. Two of the three named referrers, uvigo.es (Universidad de Vigo) and pknstan.ac.id (an Indonesian academic institution), are university domains, pointing to organic embedding in institutional library or course-resource pages rather than a formal partnership; digitalocean.com's presence looks incidental, likely a third-party tutorial or blog post linking out. Among the named competitors, consensus.app is the closest head-to-head rival (another AI answer engine for research questions), while semanticscholar.org is a free, institutionally-backed academic search engine that competes for the same discovery moment without the AI-synthesis layer.
04 Organic content
The specific pages earning their organic search traffic, and the pattern behind why they rank. Adapt the format, not the topic.
Elicit's organic footprint is concentrated almost entirely on its own product pages rather than a content-marketing library: the homepage carries the large majority of organic visits, followed by a handful of narrow solution and vertical pages (an education-industry page, a systematic-review solution page, a search solution page, and the sign-in page). All five of its top-ranking keywords, including "ai research assistant" and "ai for research," sit at position #1, which reads as strong brand-term lockout rather than broad non-branded discovery. The pattern is a product-led SEO structure, a small set of vertical/use-case landing pages absorbing category-adjacent search intent, not a blog, comparison-page, or programmatic content engine.
Top pages by organic traffic
05 Channel map
Not traffic share. How much weight the growth system actually puts on each channel, with a one-line read on the role it plays.
What drove growth
06 The motions doing the work
For founder-led SaaS the breakdown shifts from ads to traction: where the first users came from, how the founder grows it in the open, and the compounding organic surface.
Elicit's traction was never a single launch moment, it was a company that changed shape twice while a steady drip of founder-authored technical essays kept resurfacing it to a developer audience on Hacker News.
The tool first surfaced on Hacker News in November 2021 under the research nonprofit Ought (109 points), then Ought formally announced Elicit as its own product in September 2023, shortly before the company split off as an independent entity.
The founder's own account puts the seed round from Fifty Years at "summer 2023," landing in the same window as the spinout, funding the transition from a research lab to a commercial product company.
Dig deeper+1▾
Rather than one launch, the founder periodically resubmits engineering deep-dives that break through, 241 points for a search-architecture essay in December 2023 and 111 points for a product-focused post in May 2024, each acting as a mini relaunch to a technical audience rather than a single acquisition event.
Elicit's build-in-public motion runs through one person, CEO and founder Andreas Stuhlmüller, posting from his personal account rather than a separately active brand handle.
All of the company's tracked X activity is tagged to Stuhlmüller's own account, while the brand handle linked from the homepage has no measured activity of its own, meaning the "voice" of Elicit online is literally the founder's.
His posting cadence is infrequent but substantive, feature announcements, a funding/mission reflection on the two-year seed anniversary, and technical benchmark threads, rather than high-frequency engagement bait.
Dig deeper+2▾
His highest-performing post by far, a head-to-head benchmark of Opus 4.5, Sonnet 4.5, and Gemini 3 Pro on Elicit's own research-QA accuracy metrics, drew 345 likes and 40,237 views, well above his funding or product-update posts.
Unlike the HN essays covered above, there's no evidence of the founder posting his own threads to Reddit, the Reddit items on record are all third-party mentions, not owned build-in-public activity.
Search Organic accounts for 30% of Elicit's total traffic mix, but the underlying content system is thin, whether by deliberate positioning or resource constraint: there's no visible blog, comparison ("alternative to X"), or template library, just the homepage and a handful of vertical solution pages absorbing category and brand search.
The organic engine is a small set of use-case landing pages (education, systematic review, search) rather than a scaled content operation, so the lesson here is narrow and defensible positioning per segment, not volume.
Nothing in the site evidence points to a partner or affiliate program, so distribution rests entirely on product-led SEO plus the creator and founder channels covered above.
Dig deeper+1▾
LinkedIn is the one confirmed paid channel; its strongest recurring creative is copy written as a first-person InMail from cofounder Jungwon Byun ("Hi %FIRSTNAME%, I'm Jungwon, one of the cofounders of Elicit...") targeting medical-affairs and pharma titles directly, an approach expanded as Play 4 below. Google's ad-library entries, by contrast, carry unrelated third-party advertiser names, so they don't read as a genuine paid-search motion.
07 Paid ads
This founder-led SaaS also runs paid acquisition. Here are the live ads doing the work, each with the X-ray and a play you can adapt.
- Running
- 376 days (longevity = it's converting)
- Hook
- Sponsored Elicit www.elicit.com/ Elicit: AI for scientific research Search for research papers, summarize papers, extract data, and synthesize your findings Use...
- CTA / offer
- Sign Up For Free
- Audience
- Researchers in academia and industry, indicated by the text 'Used by over 2 million researchers in academia and industry' and the product's focus on 'scientific research'.
Why it works. Follows the branded headline with a plain list of the actual actions researchers take: search, summarize, extract, synthesize, proving it maps to their workflow.
Headline directly addresses the core problem and solution: 'Elicit: AI for scientific research'. List the exact verbs your buyer's job already involves right under your headline, instead of abstract benefit language.
- Running
- 296 days (longevity = it's converting)
- Hook
- Elicit Official Website - Your Solution Starts Here
- CTA / offer
- Choose Smart Choose Brand
- Audience
- Individuals seeking solutions or services, indicated by the phrase "Your Solution Starts Here" and the call to "Explore more now with confidence."
Why it works. Runs on branded search terms so the ad appears exactly when someone is already typing the company name into Google.
Headline stating a direct benefit: "Elicit Official Website - Your Solution Starts Here" Bid on your own brand name in search so switching-intent visitors land on your page instead of a competitor's comparison post.
08 The playbook
The channels are not separate. They are one system where each stage feeds the next. Here is the read, then the plays to run tomorrow.
The inferred playbook
Elicit's loop starts with two nearly separate top-of-funnel engines feeding one product, then narrows sharply once a lead becomes an enterprise prospect.
Global student creators discover Elicit organically and drive free-tier signups, while HN essays and the founder's X account earn a separate, higher-intent technical audience, and the two never really overlap.
Once a prospect looks like pharma or life sciences, the motion shifts entirely to paid LinkedIn outreach voiced by the cofounder, a deliberate move to a higher-touch, higher-ACV sales conversation.
Dig deeper+3▾
Nothing in the organic engine (a handful of solution pages, no blog or comparison content) compounds the way the HN essays or creator UGC do, so the SEO side of the business is static while the community side keeps growing.
A free Basic tier, an unusually cheap ~$11/month Plus tier (billed at $132/year), and a $39-49/month Pro tier point to a self-serve, prosumer-priced core product, with two higher tiers (a team plan and a custom-priced enterprise plan) not disclosed in detail. Given the seed-only funding stage, this traffic level, and no public MRR figures, a reasonable order-of-magnitude estimate is low-to-mid six-figure ARR from the self-serve tiers alone, this is an estimate, not a reported figure.
free-to-paid conversion rate, churn, CAC, and current MRR.
Seed niche creators instead of buying ads
The proofElicit never paid for its multilingual creator wave (Indonesian, Brazilian Portuguese, Spanish, Thai, French, Vietnamese TikTok/Instagram posts), and the biggest single post pulled 152,892 engagements with zero ad spend behind it.
The adaptationPick one underserved language or niche in your category, find 5-10 creators there already talking about adjacent problems, and send each of them free access plus a specific workflow script to demo, not a sponsorship pitch. Track which language or niche produces outsized engagement and put your next round of outreach there instead of spreading evenly.
Cost: $0 · Time to signal: weeks · Works pre-PMF: yes
Publish a real technical opinion, not a launch post
The proofThe founder's 241-point Hacker News essay on search architecture, covered above, outperformed every product-announcement post the company ever submitted there.
The adaptationWrite one specific, defensible technical argument about a tradeoff in your own product (a data-model choice, an architecture decision you'd defend against pushback), post it under your own name on your blog, and submit it yourself to the one or two technical communities where your buyer actually reads. A generic "we launched" post will not get the same reception; the argument has to be real enough to survive disagreement in the comments.
Cost: $0 · Time to signal: days · Works pre-PMF: yes
Build one landing page per buyer segment, not a blog
The proofElicit's strongest organic pages beyond its homepage are narrow, segment-specific pages (an education page, a systematic-review page, covered above), each ranking #1 for its exact category phrase rather than competing on broad content volume.
The adaptationName your top two customer segments, and for each, build one landing page using that segment's own vocabulary and proof points instead of a general blog post. Target the literal phrase that segment searches, and let that page, not a content calendar, do the SEO work.
Cost: under $500 · Time to signal: weeks · Works pre-PMF: conditional, needs at least one segment specific enough to write for
Run paid ads as a first-person message, not brand copy
The proofElicit's longest-running LinkedIn creative, referenced above, reads as a personal InMail from a named cofounder to a specific buyer title, not brand-voice ad copy.
The adaptationFor a narrow enterprise or professional audience, write your next paid social ad as if you personally messaged your ideal buyer, naming yourself and their specific role pain, and test it on a small budget against a tightly defined audience. The mechanism: a first-person, specific message reads as outreach instead of an ad, which earns a response rate brand copy doesn't.
Cost: under $500 · Time to signal: weeks · Works pre-PMF: conditional, needs a specific enough buyer persona to name credibly, generic language collapses the illusion
Publish your own benchmark, including where you lose
The proofThe founder's model-benchmark thread, covered above, drew 40,237 views, far above any of his other posts, by publishing real comparative numbers rather than a hype claim.
The adaptationPick one measurable claim central to your product (accuracy, speed, cost) and publish the actual comparative numbers on your own founder account, including the cases where a competitor or alternative approach wins. Credibility comes from showing the loss column, not hiding it.
Cost: $0 · Time to signal: days · Works pre-PMF: yes Not transferable at an earlier stage: the enterprise LinkedIn motion assumes named buyer titles and life-sciences-specific pain (medical affairs, KOL prep) that a pre-PMF founder in a different category won't yet have language for, and the scale of the creator wave reflects years of accumulated free-tier usage among students worldwide, not a tactic that reproduces in a single push.
You just read how Elicit does it. Tomorrow it is someone else.
Every morning we take one company that is actually growing and break down where its customers come from: the ads still running after a year, the channel doing the real work, and the play you can run this week.
Systemaic · directional intelligence. Traffic, spend, and reach figures are SimilarWeb-style estimates and qualitative reads of public data, not audited numbers. Built on real public receipts.
// teardowns with a similar GTM motion
sources & receipts (8)
Live ad libraries: linkedin ad library · google ad library
Launch archives: Hacker News: Build a search engine, not a vector DB · Hacker News: SARS-CoV-2 infects human adipose tissue and elicits an infla · Hacker News: Unsupervised Elicitation of Language Models · Hacker News: Elicit – AI Research Assistant · Hacker News: Elicit · Hacker News: Scaling your impact: A polynomial model of career growth
Traffic, spend, and revenue figures are estimates as noted in the report; the links above are the primary public artifacts.