The one-glance read on who they are and how they grow. Each point is verifiable from the receipts above.
An AI voice-tutor app for spoken-language fluency, valued at $1B in its December 2024 Series C, with 10M+ learners across 20+ countries.
Organic-anchored in Korea, paid-led in newer markets, running on top of a launch-led origin story that started in South Korea in 2019.
~623,693 monthly visits (+15.4% over the last 3 months), 104 active Google ads as of 2026-07-04 out of a roughly 2,000-ad historical library, and a 2.4M-follower owned Instagram account.
Co-founder and CEO Connor Zwick: "Fun fact: @speak has 65% brand recognition in South Korea and close to 10% of the population has used us to improve their English fluency."
As of 2026-07-04, Speak is running a fresh wave of Google video, image, and text ads launched June 23, alongside Meta creative testing "no sabo" (slang for US-raised heritage Latinos who do not speak fluent Spanish) and "gringo"-targeted Spanish-fluency hooks.
Speak runs two separately-cast ad funnels under one product, English fluency for Asia and Europe versus Spanish fluency for US heritage speakers, a bilingual split most single-language apps never build.
The order the channels came online. Sequence is strategy: what they did first, and what they layered on once demand existed.
Estimated demand, the channel split behind it, and the keywords and referrers doing the work. Directional modeling, not audited analytics.
| Keyword | Volume | Weight | CPC |
|---|---|---|---|
| speak | 50k | $1.10 | |
| 스픽 | 4.6k | $2.31 | |
| speak ai | 8.8k | $1.17 | |
| speak アプリ | 1.4k | $2.29 | |
| speak app | 5.6k | $1.98 |
Speak pulls an estimated 623,693 monthly visits, up 15.4% over the last 3 months. Search organic leads the mix at 36%, nearly matched by search paid at 26.6%, with direct visits close behind at 13.2%, a pattern that reads as a brand with a real organic base that still buys a large share of its own search real estate. Social paid (6.3%), affiliate (5.5%), and referrals (5%) each add a modest secondary slice, and the affiliate share likely reflects the coupon-link posts seen on Reddit and in Korean discount-focused search ads rather than a formal partner program. The rest, Gen-AI referrals, display, organic social, and email, together total under 8%.
The keyword "speak" pulls 49,790 monthly searches globally and "speak ai" pulls 8,770, showing the brand competes inside a broader AI-tutor category as much as on its own name, while the Korean brand term "스픽" carries 4,600 searches, a sign the name itself is a search term in its proven market. Among referrers, hq-hq.co.jp and lifehacker.jp are Japanese media and blog sites, suggesting press pickup feeds a slice of Japan-market discovery, while collabfund.com, a venture-focused blog, more likely reflects funding-news pickup than customer acquisition. Of the listed competitors, rarejob.com stands out as a human live-tutor conversation service dominant in Japan, meaning Speak's real competitive set includes tutoring incumbents, not just AI-only rivals like speakpal.ai.
The specific pages earning their organic search traffic, and the pattern behind why they rank. Adapt the format, not the topic.
The organic surface is anchored almost entirely in Korean-language pages: the speak.com/ko homepage alone draws about 6,686 organic visits a month, 18.3% of tracked organic traffic, followed by the Korean app-login and content-hub pages. The blog runs short "how to say X in English" expression posts, and a Korean help-center refund article ranks #1 for its own support query, "스픽 환불," meaning branded and account-support terms carry as much organic weight as top-of-funnel content. The pattern points to an SEO surface built to serve and retain an existing Korean install base more than to attract new markets through content.
Not traffic share. How much weight the growth system actually puts on each channel, with a one-line read on the role it plays.
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.
Speak's traction story is a single-market proof play: it launched in South Korea in 2019, became the country's top-ranked education brand, and used that concentrated win to fund and justify global expansion, not a Product Hunt or Hacker News-style launch spike (no evidence of either surfaced for this company).
by early 2025 that home-market concentration, close to 10% of the population having used Speak to improve their English, was a scale rarely seen for a single market.
the founder stated learner numbers more than doubled year over year for five straight years since the 2019 Korea launch, describing accelerating rather than one-time growth.
the Series B extension in mid-2024 ($20M) and the Series C in December 2024 ($78M, led by the venture firm Accel, valuing the company at $1B) each landed alongside founder X threads recapping the year, including a Google Play Best App of 2024 win in South Korea, Hong Kong, and Taiwan and a Speak for Business milestone of hundreds of enterprise customers in Asia.
no Show HN or Product Hunt activity for the company appears in the data, so unlike founder-led SaaS that break out via a single launch-day post, Speak's early growth ran through country-by-country expansion and paid acquisition instead.
Speak's build-in-public engine is small in raw following but consistently timed to real news, run by CEO Connor Zwick with an assist from CTO Andrew Hsu, not a high-frequency personal-brand cadence.
Zwick's X account (3,479 followers, 235 posts) favors recap threads over daily updates: a 2023 raise-and-US-launch announcement, a mid-2024 growth thread on five years of compounding, and an October 2024 post announcing Speak as a launch partner for OpenAI's Realtime API, a speech-to-speech AI tool, which drew his highest engagement in the sample at 204 likes and 60,722 views.
Andrew Hsu (3,664 followers) used the same Series C week to post a hiring call for engineering roles, pairing the funding headline directly with recruiting.
the only Reddit post found is a referral-discount promo from a founder-tagged account, so growth-in-public here runs mostly through timed X threads rather than sustained subreddit engagement or AMAs.
Organic search is Speak's single largest channel at 36% of all traffic, but the pages earning it skew Korean-market and support-driven rather than a broad content-marketing footprint.
the Korean homepage alone pulls roughly 6,686 organic visits a month, with Korean app-login and content-hub pages next, meaning the SEO surface is built around the one market that already proved the product.
the blog runs short Korean expression posts alongside branded and support pages, including a refund help-center article ranking #1 for its own query, a small top-of-funnel layer riding on top of account-support search demand.
no partner or affiliate page surfaced on the owned site, so the 5.5% affiliate-channel traffic share more likely reflects third-party coupon and discount-link posts than a run program.
the 104 active Google ads noted above run alongside dozens of localized Meta variants; the Korean text ads bidding on "스픽" and discount-link terms function as defensive brand-term capture, while the English, French, and German UGC and challenge-recruitment creative function as broad demand generation.
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.
Why it works. Borrows credibility from a 'no sabo turned Spanish teacher' persona to deliver a contrarian, mistakes-first framing instead of a straightforward feature pitch.
Persona-based authority (0:00) "from a no sabo turned spanish teacher". Find a creator in your niche who used to struggle and now teaches it, and have them list the top mistakes to avoid before mentioning your product.
Why it works. Structures the UGC testimonial around someone else's visible reaction, letting the payoff, what was said and the compliment that followed, unfold as a story rather than a claim.
Real-world 'hidden camera' validation (0:04-0:32) Open your testimonial with another person's reaction to your product's result, then reveal what triggered that reaction later in the video.
Why it works. Uses a low-stakes travel win, being understood at a hotel desk, as proof of real-world fluency, appealing to learners who doubt app practice transfers outside it.
POV Text Overlay (0:00): "POV: the hotel receptionist actually understood everything I said 😎" Show your product working in one small, low-stakes real-world moment instead of a dramatic success story.
Why it works. A French coach claims one 'first rule' cuts through years of grammar-textbook overwhelm, appealing to frustrated adult English learners in France.
Contrarian positioning (0:07-0:10: 'Ça ne sert à rien la grammaire...') Open with a single counterintuitive 'rule' claim in your niche, then spend the rest of the ad proving it with one concrete example.
Why it works. Uses a bearded Dominican-Spanish learner's own words to show a hard regional dialect is learnable, building trust with dialect-anxious learners.
Direct Competitor Anchoring (0:24-0:35) Film a customer mid-sentence talking about a niche subgroup of your market instead of scripting a clean opening line.
Why it works. Borrows a familiar short-form video format and a naming-withheld product mention to blend into organic feed content instead of announcing itself as an ad.
POV text overlay "pov: you become bilingual with THIS app" at 0:00. Open with 'pov:' plus your core outcome claim, and refer to your product as 'this app' instead of by name to delay the reveal and hold attention.
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 loop starts with a paid or press-driven first touch, a localized ad or a funding headline, into a free trial, compounds hardest where a market has already tipped like Korea, and still runs with no visible affiliate or community layer to carry growth between paid pushes.
the same ad budget runs an English-fluency funnel for Asia and Europe (Korean, French, German creative) and a separate Spanish-fluency funnel for US heritage speakers (the no-sabo and Gringo-themed creative), each with its own hooks and creators.
Korea's brand recognition and organic dominance show where the loop compounds on its own; other markets have no comparable organic base yet, so expansion there leans more heavily on paid spend.
recruitment ads for 14, 21, and 28-day immersion "challenges" (plus a 100-day Korean challenge) double as both signups and a testimonial pipeline for the next round of ads, expanded as Play 2 below.
Premium runs about $14 to $20 a month or roughly $99 a year behind a 7-day free trial (a paid model, not freemium, since there's no permanent free tier), Premium Plus adds unlimited personalized lessons at an undisclosed higher price, and Speak for Business sells the same engine to employers, with hundreds of business customers in Asia per the founder.
free-to-paid conversion, churn, CAC, blended ARPU, and a current revenue figure (no revenue was disclosed at the December 2024 Series C; the $100M+ ARR figure was announced separately in November 2025).
The proofSpeak's entire growth curve traces back to concentrating on South Korea alone until it hit 65% brand recognition before expanding elsewhere.
The adaptationInstead of testing five markets or segments at once, pick the single niche where your product's pain is most acute and build your first landing page, content, and outreach entirely around it. Go deep enough to become the obvious answer in that one pocket before spending anything on a second. Start by identifying the segment where your existing users show the strongest word-of-mouth and write your next month of content only for them.
Cost: $0 · Time to signal: weeks · Works pre-PMF: yes
The proofSpeak runs 14, 21, 28, and 100-day "immersion challenge" recruitment ads targeting specific groups like nurses and heritage-Spanish speakers, generating both signups and testimonial footage.
The adaptationPost a free "7-day [outcome] challenge" in a relevant community or DM a small group of prospects directly, asking them to commit publicly to a short structured trial of your product. The commitment device drives usage, and their before/after results become your next batch of proof-of-concept content at zero production cost. First step: recruit 5-10 people from your existing waitlist or a niche subreddit and set a fixed start date.
Cost: $0 · Time to signal: days to weeks · Works pre-PMF: yes
The proofConnor Zwick's highest-engagement X post wasn't a routine update, it was tied to Speak becoming an OpenAI Realtime API launch partner, an actual external event.
The adaptationSave your public build-in-public posts for moments with real external validation, an integration going live, a partnership, a press mention, rather than posting on a fixed schedule regardless of news. Draft the post the moment the real event happens and lead with the concrete detail, not a generic milestone recap. The mechanism: audiences engage with news, not with process updates dressed up as news.
Cost: $0 · Time to signal: days · Works pre-PMF: yes
The proofSpeak's Korean refund help-center article ranks #1 for its own support query, turning a routine account page into a real organic-traffic asset.
The adaptationPublish plain, findable help-center articles for your product's actual transactional queries, cancellation, refunds, pricing, so they rank for the searches your own users (and prospects comparing options) are already typing. First step: pull your five most-asked support questions from tickets or chat logs and write one focused page for each.
Cost: $0 · Time to signal: weeks to months · Works pre-PMF: conditional, it needs enough existing users or search volume around your brand name for these queries to have traffic behind them.
The proofSpeak runs an entirely separate Spanish-fluency-for-heritage-speakers creative track alongside its core English-learning campaigns, aimed at a US audience its main product story doesn't address.
The adaptationLook for a second, underused audience already adjacent to your product, one with a different pain than your primary buyer, and build a distinct content or creative track speaking directly to them rather than folding them into your main message. Start with organic content for that audience before committing ad spend, then layer paid once you see engagement.
Cost: dedicated budget or hire once paid, $0 for the organic-content first pass · Time to signal: weeks · Works pre-PMF: conditional, it needs a genuinely distinct secondary audience to already exist around your product. Not transferable at an earlier stage: the scale of Speak's paid infrastructure, a roughly 2,000-ad Google library and dozens of professionally localized Meta variants across four languages, requires funding most readers won't have yet. Adapt the concentration and challenge-cohort mechanics behind it, not the ad budget itself.
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.
Live ad libraries: meta ad library · google ad library
Launch archives: Hacker News: Tim Cook Speaks Up · Hacker News: Bose has released API docs and opened the API for its EoL So · Hacker News: Kagi Translate now supports LinkedIn Speak as an output lang · Hacker News: “Let her speak please” · Hacker News: It's time for us in the tech world to speak out about crypto · Hacker News: Google chief: I'd disclose smart speakers before guests ente
Traffic, spend, and revenue figures are estimates as noted in the report; the links above are the primary public artifacts.