In reply to my @demotruk/lets-build-a-web-of-trust" target="_blank" rel="noopener noreferrer">last post, @stayoutoftherz asked where he can see the retention statistics of the different onboarding methods that we have on Hive.

Where can I see retention stats? Have they ever been published?

By checking how well the onboarded users do over a longer term. I am not sure exactly, but to give people 5 HBD or so for doing one post like in CheckInWithXYZ is apparently not the way to do it.

Fair enough to give incentives only to trusted people, so this trust thing has its merit.

@stayoutoftherz/re-demotruk-tdfas1" target="_blank" rel="noopener noreferrer">https://peakd.com/hive-107705/@stayoutoftherz/re-demotruk-tdfas1

I've written on this topic @demotruk/ocd-leading-the-way-in-user-retention" target="_blank" rel="noopener noreferrer">in the past, but my numbers are old, perhaps outdated. Since I have been using Claude Code, I know that it can do a better job of it than I can anyway, so, rather than relitigate it myself and similar to another of my recent posts, I'll let Claude Opus 4.6 dig into this topic...

The rest of this post is researched and presented by Claude, guided and directed by myself.

---

Methodology

This analysis uses HiveSQL to query the full on-chain history of both the Hive and legacy Steem blockchains. Every account creation on-chain records a creator field — the account that paid the fee or used a claimed account token — which lets us definitively attribute accounts to the service or project that created them. For retention, we use the last_root_post field (the last time an account published a top-level post) as the primary metric, with last_post (includes comments) and last_vote_time as secondary signals.

There are important caveats. The creator field tells us who created the account, but not always why. Some creator accounts (like ocdb) serve as infrastructure for multiple downstream projects. Users who arrive through community programs like OCD's onboarding or Curie's curation are tracked via beneficiary rewards and voting records, which introduces selection bias — we can only identify users who already posted. Where these biases affect interpretation, we note them.

All data was queried on April 13, 2026.

Part 1: A History of Onboarding on Steem and Hive

The story of onboarding on this blockchain is a story of evolving philosophy: from scarcity and gatekeeping, to abundance and open doors, and a growing realization that getting people in was never the hard part.

The Genesis: Mined Accounts (March 2016)

When Steem launched on March 24, 2016, there was no signup page. Account creation was fused with block production through Proof of Work mining. You ran the steemd daemon, configured your desired username as the miner name, and if you successfully mined a block, your account was born. 13,953 accounts were created this way before mining was removed in Hardfork 17 (January 2017).

Of those 13,953 mined accounts, only 387 ever published a post and just 10 are still posting in 2025. Most were name-squatters or bots — the infamous "rabbit" accounts that eventually hijacked the mining queue are a relic of this era.

The Steemit.com Faucet (2016–2020)

Steemit.com became the singular gateway to the blockchain when it launched in July 2016. The process was centralized: submit your email and phone number, wait for manual approval, and eventually receive an account with ~15 SP delegated to get you started. Wait times were notoriously variable — Steemit stated "up to 7 days," but during the late 2017 crypto boom, some users waited weeks.

The steem account created 1,261,927 accounts over its lifetime — more than any other single creator. This includes the massive 2018 peak when over 600,000 accounts were created in a single year.

Third-Party Alternatives Emerge (2017–2018)

The friction of Steemit's signup process created market demand for alternatives:

• AnonSteem (by @someguy123) offered anonymous account creation for Bitcoin or Litecoin payment, with no identity verification and a Tor hidden service. 8,833 accounts created.
• BlockTrades added account creation as a paid service (~$3-6 in crypto). Instant creation, no wait. 18,347 accounts created.
• SteemConnect (by @good-karma and the Busy.org team) provided OAuth2-based creation, later evolving into HiveSigner after the fork.
• SteemInvite (by @pharesim) introduced shareable invite links.
• Steem Ninja pioneered credit card purchases for ~$2 with a referral reward system.

The Protocol Evolves

Two hardforks fundamentally changed how account creation worked:

Hardfork 17 (late 2016) introduced account_create_with_delegation, allowing cheaper creation by delegating Steem Power instead of paying the full fee. This is how Steemit Inc funded mass free accounts. It was deprecated in HF20.

Hardfork 20 (September 2018) introduced Resource Credits and Account Creation Tokens — the system still in use today. Any account with sufficient Hive Power can burn RC to claim free account creation tokens, then use those tokens to create accounts at zero cost. This shifted account creation from a centralized bottleneck to a distributed capability. Today, roughly 2,100 tokens are claimed daily across the network, with over 3 million unused tokens in reserve.

The Hive Fork and the Cambrian Explosion (2020–Present)

When Hive forked from Steem in March 2020, all existing infrastructure carried over — and then diversified rapidly. Multiple frontends and dApps began running their own onboarding pipelines:

• HiveOnboard (by @roomservice) — A free account creation service funded by community HP delegations. SMS verification for anti-abuse. Integrated with PeakD and Hive Keychain. ~1,600 accounts created directly, but many more via partner services that used its flow.
• Ecency — Free email signup with 5 HP delegation to every new account. One of the top account creators by volume with **47,000 accounts** (including as esteemapp).
• InLeo / LeoFinance — Pioneered "Lite Accounts" where users sign up via Twitter or MetaMask and never see their Hive keys unless they choose to. The @leo.voter account became the #1 active account creator in 2025, creating over 29,000 accounts that year alone.
• 3Speak — Video platform with Web2-style email signup and custodial key management. 29,000+ accounts.
• Splinterlands — The play-to-earn card game was historically the single largest driver of Hive account creation. The @steemmonsters account created 713,000+ accounts, including 659,000 in 2021 alone. Initially every signup got a Hive account; later, players used proxy accounts until purchasing the Spellbook.
• DBuzz — Microblogging platform. 9,000+ accounts.
• BlockTrades — Continued as a paid service on Hive (~$3). Now also available through Hivedex.io as promoted by PeakD.
• Actifit — Fitness tracking dApp, from $2 per account.

The OCDB account deserves special mention: it created 181,000+ accounts but functions as infrastructure for the broader OCD ecosystem rather than a single-purpose onboarding service. Many accounts created through OCDB are service accounts, bots, or accounts created on behalf of downstream projects.

The timeline tells a dramatic story: Steemit.com's dominance through 2019, the Splinterlands supernova of 2021 (which created more accounts in three months than Steemit.com created in its entire history), and the current era of diversified, lower-volume creation from multiple competing services.

Community-Driven Onboarding and Retention Projects

Beyond the account creation services, a parallel ecosystem of community projects has always focused on the harder problem: not just getting people accounts, but keeping them.

Curie (founded August 2016) was one of the earliest — over 100 curators worldwide browsing the platform daily to discover and reward new authors. Strict guidelines ensured quality: authors had to be under reputation 62, posts needed to be between 150 minutes and 20 hours old with under $1 pending payout. Curie expanded to support 8+ languages and served as proof of concept for many new users that earning on the platform was genuinely possible.

The Minnow Support Project / PALnet (founded June 2017) took a different approach, building a Discord community where new users could request upvotes and find mentorship. At its peak, it had 3,400+ followers and 123,000 SP. Welcome Wagon and Newbieresteemday focused on hands-on human mentorship rather than automated support — choosing 3 newbies per week for intensive help, or resteemed small-account content to give newcomers visibility.

On Hive, these evolved into more structured programs:

OCD's Onboarding Program (led by @acidyo) generates unique, tracked invite links through HiveOnboard, with a 3% beneficiary cut to the onboarder, 1% to HiveOnboard, and 1% to OCD. Onboarders are vetted through Discord and can nominate their invitees' posts for curation by the @ocdb account. Over 27 compilation posts have been published showcasing newly onboarded users.

CheckInWithXYZ uses a "Proof of Person" model — a selfie with a brief introduction, approved by one of 150+ Trusted Onboarders. Rewards include 2 HBD and ~$1 in Bitcoin. A structured "Journey of a Hivean" task sequence guides users through their first experiences. The program has onboarded 837 tracked users since launching in mid-2025 (tracked via the @pioneersupporter follow list), with a notable 38-user batch from the Cripto Latin Fest in Medellín, Colombia.

Aliento (led by @eddiespino) focuses specifically on the Hispanic community, providing 15-100 HP delegations, Discord-based support in Spanish, and hands-on mentorship. They operate as a Hive witness and have created 3,671 accounts directly.

HiveBuzz (by @arcange) takes a gamification approach — badges, achievements, rankings, and the Hive Tour for new users. When users earn a badge, HiveBuzz drops a notification comment, creating a dopamine loop that encourages continued activity.

Part 2: The Data

With the history established, let's look at what actually works. We can measure retention in several ways, but the most meaningful question is: of the accounts created by each service, how many are still posting months or years later?

The Big Picture: Retention by Account Creator

For accounts created in 2023 (giving us a full year+ of retention data), here is how the major creators compare:

OCD stands alone. With 55.5% of their 627 accounts ever posting, 32.5% still posting after 3 months, and 16.3% after 12 months, OCD's vetted onboarding model produces the highest retention rates of any account creation service. The runner-up cluster — DBuzz, Ecency, 3Speak, Appreciator, OCDB — all land in the 7-10% range at 6 months. Leo.voter trails at 1.6%.

How Fast Do Users Drop Off?

The decay curves reveal the shape of retention loss over time:

Every creator shows the same basic pattern: a steep drop in the first 1-3 months, then a gradual flattening. The crucial difference is where each curve flattens. OCD's curve stabilizes around 16% — meaning roughly 1 in 6 of their onboarded users becomes a long-term contributor. Leo.voter's curve drops to 1% and flatlines — their Lite Account model brings users in easily but doesn't create lasting posting habits.

Community Support vs. Raw Account Creation

When we compare users who received community support (HiveOnboard's guided onboarding flow, or a Curie vote within their first 90 days) against users who simply received an account from a creation service, the gap is enormous:

Curie-voted new users retain at 47.5% after 12 months. HiveOnboard beneficiary users retain at 29.2%. Compare these to Ecency's 5.8% or Leo.voter's 0.9%.

An important caveat: these community cohorts have selection bias. The HiveOnboard beneficiary is only set on posts that were actually published, so this cohort already demonstrated intent to participate. Curie curators actively selected for quality content. These numbers don't tell us that community support caused better retention — it's more likely that community programs are better at identifying users who would have retained anyway, and then reinforcing that behavior.

That said, the reinforcement effect shouldn't be dismissed. A well-timed Curie vote or curation support early in a user's journey provides social proof that the platform works — that content gets seen, and effort gets rewarded.

The Payout-Retention Curve

This is the single most important chart in this analysis.

For the 2023 cohort across all creators, we divided users by what they earned on their very first post. The relationship is nearly linear: users who earned $5-$20 on their first post retain at 3x the rate of users who earned nothing. Users who earned $20+ retain at nearly 5x the rate of zero-payout users at the 12-month mark.

This finding is consistent with what I reported in @demotruk/the-single-most-impactful-thing-you-can-do-for-hive-upvote-a-newbie" target="_blank" rel="noopener noreferrer">my previous analysis. Early rewards are the strongest predictor of whether a new user will stick around.

What Creators Pay Their Newbies

This naturally raises the question: how much do different onboarding services' newbies actually earn?

OCD's average first-post payout of $44.44 dwarfs every other creator. Aliento comes second at $17.24. Everyone else is under $5. This largely explains OCD's superior retention — they don't just create accounts, they ensure those accounts' early content receives substantial curation.

Among retained users (those who stuck around 6+ months), the payout gap vs. churned users was most extreme for Leo.voter: retained users earned 7x what churned users earned on their first post. The message is clear — even within a Lite Account system, early rewards are what separate stayers from leavers.

The #introduceyourself Signal

We also checked whether using the traditional #introduceyourself tag on an early post correlates with retention:

Users who write an intro post retain at 3-6x the rate of those who don't, depending on the creator. The gap is most extreme for Aliento (30.4% vs. 5.5%) and Leo.voter (27.7% vs. 1.6%).

This likely reflects self-selection — users who take the time to introduce themselves are already more committed. But it also suggests that encouraging new users to write a proper introduction post (rather than diving straight into content or transactions) may serve as both a commitment device and a social on-ramp that connects them to the community.

The Steem Era: Long-Term Retention (8 Years of Data)

The 2017 Steem cohort gives us the longest possible retention window — 8+ years:

Among users who ever posted, Steemit.com retention drops to 16.2% after a year, then continues declining to 0.5% after 8 years. AnonSteem (paid, anonymous) and BlockTrades (paid, instant) both outperformed Steemit.com at every stage, confirming that users who pay for accounts — even small amounts — retain better due to self-selection.

After 8 years, the long-term curves converge. Time eventually catches everyone.

Boom vs. Post-Boom: 2021 vs. 2023

How did retention change between the Splinterlands-driven boom of 2021 and the quieter, more intentional period of 2023?

Every single creator improved their 6-month retention rate from 2021 to 2023, even as volumes dropped dramatically. OCD went from 5.0% to 23.3%. Ecency from 5.5% to 8.1%. This strongly suggests that boom-era signups were disproportionately low-quality — speculators and bots inflated the denominator. When the hype subsided, what remained was a higher proportion of genuine users.

Splinterlands: A Different Kind of User

Splinterlands created 703,000+ accounts on Hive post-fork, yet only 1.2% ever published a post. Does this mean Splinterlands onboarding failed? Not exactly — these accounts were created to play a card game, not to blog. When we look at voting activity (which includes game actions broadcast as custom_json), 1.3% were still voting in 2025 — comparable to the posting rate but representing a different kind of blockchain engagement entirely.

The takeaway: Splinterlands brought hundreds of thousands of users to the Hive blockchain, but very few to the Hive social platform. These are fundamentally different outcomes, and measuring one by the standards of the other is misleading.

CheckInWithXYZ: Early Returns

CheckInWithXYZ is too new (launched mid-2025) for definitive retention conclusions, but early data is available. Of 837 tracked users:

• 98.1% ever posted (the check-in itself is a post, so this is by design)
• 63.3% posted more than once
• 16.1% posted after 1 month
• 8.1% posted after 3 months

The biggest cohort came in April 2025 (188 users), with a notable spike in August 2025 (124 users) around the Cripto Latin Fest event. Early cohorts with enough time elapsed show ~15-22% posting after 3 months, which is competitive with other free account creation services. Whether the "Proof of Person" model and structured task sequence will produce better long-term retention remains to be seen.

Part 3: Key Insights and Takeaways

The Volume-Quality Tradeoff

The data reveals a fundamental tension in onboarding strategy. Services that make account creation easiest (Leo.voter's Lite Accounts, Splinterlands' game signups) create the most accounts but retain the fewest per capita. Services that invest more in each user (OCD's vetted onboarding, Aliento's community support) retain far more, but at much lower volumes.

The question for the ecosystem is: which approach produces more retained users in absolute terms? In 2023, OCD created 627 accounts and retained 146 at 6 months. OCDB created 11,011 accounts and retained 800. Leo.voter created 9,733 and retained 155. The high-volume approaches win on absolute numbers, but at vastly higher cost in wasted account creation tokens and ecosystem resources.

What Actually Works for Retention

The data consistently points to a few factors that predict retention:

1. Early rewards are the strongest lever. First-post payouts of $5+ correlate with 3x better retention than $0. This was true in my @demotruk/the-single-most-impactful-thing-you-can-do-for-hive-upvote-a-newbie" target="_blank" rel="noopener noreferrer">previous analysis, and it remains true now with broader data across onboarding methods.

2. Community connection matters. Users who write an #introduceyourself post retain 3-6x better. Users who receive attention from established curators (Curie, OCD) retain dramatically better. Whether this is selection effect or causal, the correlation is undeniable.

3. Paid accounts retain better than free ones. AnonSteem and BlockTrades users outperformed Steemit.com users at every retention window, across 8 years of data. Paying even $2-3 for an account selects for commitment.

4. Lite accounts don't solve retention. Removing the key management barrier brings in more sign-ups, but these users don't become content creators at meaningful rates. Leo.voter's 1.6% 6-month posting retention suggests that frictionless onboarding alone isn't enough.

5. Boom periods dilute quality. Every creator saw improved retention rates from 2021 to 2023 as volumes dropped. When the speculators leave, the signal-to-noise ratio improves.

Implications for Onboarding Projects

For projects building onboarding tools and programs today, the data suggests:

Invest in the first week, not just the first click. Account creation is a solved problem — we have millions of unused creation tokens. The bottleneck is what happens after creation. Early curation, community engagement, and structured guidance (like CheckInWithXYZ's "Journey of a Hivean" or OCD's nomination system) all correlate with better outcomes.

An intro post matters more than you'd think. Encouraging users to write a genuine #introduceyourself post seems to function as both a commitment device and a social connector. Projects that skip this step in favor of immediate transactional engagement may be leaving retention on the table.

Target quality over quantity. OCD's 627 carefully onboarded accounts in 2023 produced 146 retained users at 6 months — nearly as many as Leo.voter's 9,733 Lite Accounts produced (155). One approach requires 15x fewer resources.

The payout question is still central. If the single best predictor of retention is first-post earnings, then the most impactful thing any Hive stakeholder can do remains what it has always been: find a new user and upvote their content. All the UX improvements, lite accounts, and gamification in the world don't substitute for that first meaningful reward that tells a new user, "someone noticed, and what you did has value here."