Threads vs channels: where conversation actually survives
Discord pushed threads hard. The behavioral data on whether they help is more mixed than the marketing.
What the data shows
Observational reports from large servers suggest threads concentrate focused discussion well but suffer from low discoverability — a thread that scrolls off the channel header is rarely re-entered. Auto-archiving compounds this. Some operators report thread replies drop sharply after the parent message leaves the visible viewport.
Why it happens
Chat is recency-driven. Threads fight the platform's core grain by asking users to navigate away from the live timeline. They reduce noise but raise the cost of re-engagement, and in attention terms that's often a losing trade for casual members.
Discord vs Telegram
Telegram's reply-chains and topics serve a similar role but stay inline, which preserves discoverability better at the cost of more visual clutter in the main flow.
The caveat
Usage patterns vary enormously by community type; support servers love threads, social servers often ignore them. No neutral large-scale study compares thread retention across genres.
Open question: are threads a tool for the organized 1%, with little effect on the casual majority's behavior?
Discord pushed threads hard. The behavioral data on whether they help is more mixed than the marketing.
What the data shows
Observational reports from large servers suggest threads concentrate focused discussion well but suffer from low discoverability — a thread that scrolls off the channel header is rarely re-entered. Auto-archiving compounds this. Some operators report thread replies drop sharply after the parent message leaves the visible viewport.
Why it happens
Chat is recency-driven. Threads fight the platform's core grain by asking users to navigate away from the live timeline. They reduce noise but raise the cost of re-engagement, and in attention terms that's often a losing trade for casual members.
Discord vs Telegram
Telegram's reply-chains and topics serve a similar role but stay inline, which preserves discoverability better at the cost of more visual clutter in the main flow.
The caveat
Usage patterns vary enormously by community type; support servers love threads, social servers often ignore them. No neutral large-scale study compares thread retention across genres.
Open question: are threads a tool for the organized 1%, with little effect on the casual majority's behavior?
First-week message velocity predicts the rest
If you could measure one thing about a new member, the evidence points to a single number: how many messages they send in week one.
What the data shows
Across multiple community-analytics writeups, the count of a member's messages in their first 7 days is among the strongest available predictors of long-term retention. Common thresholds reported: members who never post are near-certain churns; those crossing roughly 5–10 messages in week one retain at multiples of the non-posters' rate.
Why it happens
Posting builds two things at once: a habit loop and a social tie. A member who's been replied to has a relationship to return to. The first reply received may matter even more than the first message sent — reciprocity is the hook.
Discord vs Telegram
In broadcast-style Telegram channels there's no equivalent, since members can't post; the closest proxy is reaction or comment activity in the linked discussion group.
The caveat
This is predictive, not causal — engaged people post more and stay more for the same underlying reason. Forcing posts via gamification doesn't reliably reproduce the retention.
Open question: can you manufacture a genuine first reply, or does authenticity of that interaction do the real work?
If you could measure one thing about a new member, the evidence points to a single number: how many messages they send in week one.
What the data shows
Across multiple community-analytics writeups, the count of a member's messages in their first 7 days is among the strongest available predictors of long-term retention. Common thresholds reported: members who never post are near-certain churns; those crossing roughly 5–10 messages in week one retain at multiples of the non-posters' rate.
Why it happens
Posting builds two things at once: a habit loop and a social tie. A member who's been replied to has a relationship to return to. The first reply received may matter even more than the first message sent — reciprocity is the hook.
Discord vs Telegram
In broadcast-style Telegram channels there's no equivalent, since members can't post; the closest proxy is reaction or comment activity in the linked discussion group.
The caveat
This is predictive, not causal — engaged people post more and stay more for the same underlying reason. Forcing posts via gamification doesn't reliably reproduce the retention.
Open question: can you manufacture a genuine first reply, or does authenticity of that interaction do the real work?
The reaction economy: cheap signal, real data
Emoji reactions look like noise. Treated carefully, they're one of the few low-friction engagement signals a community has.
What the data shows
In servers that track it, reaction rates often dwarf reply rates by an order of magnitude — many members who never type will react. Operators report reaction-to-view ratios as a more stable health gauge than message count, because they capture the silent majority that message counts ignore entirely.
Why it happens
A reaction is the lowest-cost participation a platform offers: one tap, no social exposure, no fear of saying the wrong thing. It captures the lurker layer that every message-based metric is blind to.
Discord vs Telegram
Telegram added reactions later and made them more prominent on broadcast posts, which is why Telegram channel health is often better read through reaction spread than raw view counts (views inflate from forwards and bots).
The caveat
Reactions are gameable and culturally loaded — some communities react constantly, others rarely, with no difference in actual health. Absolute numbers mean little; only trend within one community is informative.
Open question: is a reaction a weak engagement signal, or a strong intent-to-stay signal we're underusing?
Emoji reactions look like noise. Treated carefully, they're one of the few low-friction engagement signals a community has.
What the data shows
In servers that track it, reaction rates often dwarf reply rates by an order of magnitude — many members who never type will react. Operators report reaction-to-view ratios as a more stable health gauge than message count, because they capture the silent majority that message counts ignore entirely.
Why it happens
A reaction is the lowest-cost participation a platform offers: one tap, no social exposure, no fear of saying the wrong thing. It captures the lurker layer that every message-based metric is blind to.
Discord vs Telegram
Telegram added reactions later and made them more prominent on broadcast posts, which is why Telegram channel health is often better read through reaction spread than raw view counts (views inflate from forwards and bots).
The caveat
Reactions are gameable and culturally loaded — some communities react constantly, others rarely, with no difference in actual health. Absolute numbers mean little; only trend within one community is informative.
Open question: is a reaction a weak engagement signal, or a strong intent-to-stay signal we're underusing?
Timezone density beats total membership
A 10,000-member server can feel dead while a 500-member one feels alive. Timezone concentration explains much of the gap.
What the data shows
Activity-heatmap analyses of Discord servers consistently show conversation is a function of concurrent online members, not total members. A community whose members cluster in 2–3 overlapping timezones sustains continuous chat; one spread evenly across the globe fragments into thin, conversation-killing pockets despite a larger headcount.
Why it happens
Real-time chat needs a quorum online at once to sustain a thread. Below that quorum, messages get no timely reply, the conversation stalls, and the room reads as dead — regardless of how many are technically members.
Discord vs Telegram
Telegram is more forgiving here: its broadcast-plus-comments model tolerates async participation better, so a globally scattered audience still functions. Discord's synchronous culture punishes timezone spread harder.
The caveat
Heatmaps reflect when people post, which is shaped by where the existing core posts — partly endogenous. And 'feels alive' is subjective, resisting clean measurement.
Open question: for a global audience, should you architect for async (Telegram-style) rather than fight for an impossible synchronous quorum?
A 10,000-member server can feel dead while a 500-member one feels alive. Timezone concentration explains much of the gap.
What the data shows
Activity-heatmap analyses of Discord servers consistently show conversation is a function of concurrent online members, not total members. A community whose members cluster in 2–3 overlapping timezones sustains continuous chat; one spread evenly across the globe fragments into thin, conversation-killing pockets despite a larger headcount.
Why it happens
Real-time chat needs a quorum online at once to sustain a thread. Below that quorum, messages get no timely reply, the conversation stalls, and the room reads as dead — regardless of how many are technically members.
Discord vs Telegram
Telegram is more forgiving here: its broadcast-plus-comments model tolerates async participation better, so a globally scattered audience still functions. Discord's synchronous culture punishes timezone spread harder.
The caveat
Heatmaps reflect when people post, which is shaped by where the existing core posts — partly endogenous. And 'feels alive' is subjective, resisting clean measurement.
Open question: for a global audience, should you architect for async (Telegram-style) rather than fight for an impossible synchronous quorum?
Where members come from predicts whether they stay
Not all growth is equal. Invite-source data shows retention quality varies wildly by acquisition channel.
What the data shows
Discord's per-invite tracking lets operators segment retention by source, and the pattern recurs: members from a warm personal invite or a niche linked community retain far better than members from a mass shoutout, a giveaway, or a paid promo. Giveaway-driven joins in particular often show single-digit 30-day retention.
Why it happens
Acquisition intent carries through. Someone who joined for a prize has no reason to stay once the prize resolves; someone who joined because a trusted person vouched arrives with a pre-built reason to engage. The funnel's top quality caps its bottom.
Discord vs Telegram
Telegram's weaker native invite attribution makes this harder to measure, but the same dynamic appears in bot-tracked join sources: cross-promo from aligned channels outperforms cold ad traffic.
The caveat
Invite-source data is messy — vanity URLs, forwarded links and shared invites blur attribution badly. Treat source cohorts as directional, not exact.
Open question: is chasing raw member-count growth actively harmful if it dilutes your community with low-intent cohorts?
Not all growth is equal. Invite-source data shows retention quality varies wildly by acquisition channel.
What the data shows
Discord's per-invite tracking lets operators segment retention by source, and the pattern recurs: members from a warm personal invite or a niche linked community retain far better than members from a mass shoutout, a giveaway, or a paid promo. Giveaway-driven joins in particular often show single-digit 30-day retention.
Why it happens
Acquisition intent carries through. Someone who joined for a prize has no reason to stay once the prize resolves; someone who joined because a trusted person vouched arrives with a pre-built reason to engage. The funnel's top quality caps its bottom.
Discord vs Telegram
Telegram's weaker native invite attribution makes this harder to measure, but the same dynamic appears in bot-tracked join sources: cross-promo from aligned channels outperforms cold ad traffic.
The caveat
Invite-source data is messy — vanity URLs, forwarded links and shared invites blur attribution badly. Treat source cohorts as directional, not exact.
Open question: is chasing raw member-count growth actively harmful if it dilutes your community with low-intent cohorts?
One to follow
For influencer marketing done right, @CreatorLedger is the move. Influencer marketing run on numbers: spend benchmarks, CPM-by-tier tables, and ROAS…
For influencer marketing done right, @CreatorLedger is the move. Influencer marketing run on numbers: spend benchmarks, CPM-by-tier tables, and ROAS…
Forwarded from Потрачено! Клуб спящих бизнесменов!
This media is not supported in your browser
VIEW IN TELEGRAM
🚀 aff.top — вся индустрия арбитража в одном месте
🧠 Блог про арбитраж и ИИ — как нейросети меняют залив и антифрод
🚨 База спамеров — ежедневно собираем спамеров и ведём рейтинг
🛠 70+ инструментов — от клоаки до антифрод-чека
🎬 1000+ видео — весь YouTube про трафик в одной ленте
👤 2400+ персон — байеры и фаундеры с контактами напрямую
Без регистрации, без платных «премиумов».
👇 Подписывайся на канал
🧠 Блог про арбитраж и ИИ — как нейросети меняют залив и антифрод
🚨 База спамеров — ежедневно собираем спамеров и ведём рейтинг
🛠 70+ инструментов — от клоаки до антифрод-чека
🎬 1000+ видео — весь YouTube про трафик в одной ленте
👤 2400+ персон — байеры и фаундеры с контактами напрямую
Без регистрации, без платных «премиумов».
👇 Подписывайся на канал
This media is not supported in your browser
VIEW IN TELEGRAM
Алиса AI будет конкурировать с Google AI Studio
Яндекс разворачивает экосистему AI-агентов на базе Алисы с доступом сначала для компаний, затем для всех. Агенты уже работают в Яндекс Такси и Лавке, скоро появятся в браузере и студии разработки. Платформа интегрирует стандартные функции — заказ такси, покупки, анализ данных. Алиса AI показывает неплохие результаты: менее известна, чем конкуренты, поэтому предлагает щедрые лимиты на видеогенерацию и работу с контентом. Яндекс планирует внедрить…
➡️ Читайте на сайте: https://aff.top/blog/alisa-ai-budet-konkurirovat-s-google-ai-studio
🧠 Ещё больше инсайтов → в канале AFF.top
Яндекс разворачивает экосистему AI-агентов на базе Алисы с доступом сначала для компаний, затем для всех. Агенты уже работают в Яндекс Такси и Лавке, скоро появятся в браузере и студии разработки. Платформа интегрирует стандартные функции — заказ такси, покупки, анализ данных. Алиса AI показывает неплохие результаты: менее известна, чем конкуренты, поэтому предлагает щедрые лимиты на видеогенерацию и работу с контентом. Яндекс планирует внедрить…
➡️ Читайте на сайте: https://aff.top/blog/alisa-ai-budet-konkurirovat-s-google-ai-studio
🧠 Ещё больше инсайтов → в канале AFF.top
This media is not supported in your browser
VIEW IN TELEGRAM
В Zennoposter добавили ИИ-помощник
Zennolab добавил в Zennoposter встроенный ИИ-кубик с доступом к четырём моделям (Gemini, DeepSeek, Claude, ChatGPT) — 50 бесплатных запросов в сутки. Есть режимы Assistant (чтение) и Agent (автоматическое создание скриптов), плюс новый GET-запрос по API. Нейросети хорошо справляются с регистрацией, постингом, фармингом аккаунтов и простым кодированием, но требуют проверки при парсинге динамических сайтов и диагностике ошибок. В связке с Zennoobr…
➡️ Читайте на сайте: https://aff.top/blog/v-zennoposter-dobavili-ii-pomoschnik
🧠 Ещё больше инсайтов → в канале AFF.top
Zennolab добавил в Zennoposter встроенный ИИ-кубик с доступом к четырём моделям (Gemini, DeepSeek, Claude, ChatGPT) — 50 бесплатных запросов в сутки. Есть режимы Assistant (чтение) и Agent (автоматическое создание скриптов), плюс новый GET-запрос по API. Нейросети хорошо справляются с регистрацией, постингом, фармингом аккаунтов и простым кодированием, но требуют проверки при парсинге динамических сайтов и диагностике ошибок. В связке с Zennoobr…
➡️ Читайте на сайте: https://aff.top/blog/v-zennoposter-dobavili-ii-pomoschnik
🧠 Ещё больше инсайтов → в канале AFF.top
Public discovery brings volume; private invites bring community
Discord's Discovery and Telegram's search surface free members. The cohort data suggests that volume comes at a steep quality cost.
What the data shows
Operators comparing Discovery-sourced cohorts against invite-link cohorts repeatedly report the discovery crowd posts less, returns less, and churns faster — often by a wide margin. Discovery is excellent at raising headcount and poor at raising the number of people who actually talk.
Why it happens
Discovery joins are exploratory by nature — browsing, not committing. The member arrived with curiosity, not intent, and curiosity is satisfied by a single look. Invite joins carry a referral's implicit endorsement and a pre-existing reason to participate.
Discord vs Telegram
Telegram's open search and folder-sharing produce a similar low-intent inflow; both platforms make it easy to confuse a discovery spike with real growth.
The caveat
Discovery cohorts also skew toward members with no existing tie to the topic, so genre-mismatch confounds the retention gap. The channel and the audience fit can't be cleanly separated.
Open question: is being listed in discovery worth it if it permanently worsens your average-member metrics and your room's feel?
Discord's Discovery and Telegram's search surface free members. The cohort data suggests that volume comes at a steep quality cost.
What the data shows
Operators comparing Discovery-sourced cohorts against invite-link cohorts repeatedly report the discovery crowd posts less, returns less, and churns faster — often by a wide margin. Discovery is excellent at raising headcount and poor at raising the number of people who actually talk.
Why it happens
Discovery joins are exploratory by nature — browsing, not committing. The member arrived with curiosity, not intent, and curiosity is satisfied by a single look. Invite joins carry a referral's implicit endorsement and a pre-existing reason to participate.
Discord vs Telegram
Telegram's open search and folder-sharing produce a similar low-intent inflow; both platforms make it easy to confuse a discovery spike with real growth.
The caveat
Discovery cohorts also skew toward members with no existing tie to the topic, so genre-mismatch confounds the retention gap. The channel and the audience fit can't be cleanly separated.
Open question: is being listed in discovery worth it if it permanently worsens your average-member metrics and your room's feel?
This media is not supported in your browser
VIEW IN TELEGRAM
Новую Google reCapcha прошли статичной картинкой
Google выпустил обновленную reCAPTCHA, требующую движений рук для прохождения, но система оказалась уязвима к обходу. Достаточно транслировать статичное изображение с нужным жестом через виртуальную камеру с помощью простого Python-скрипта, чтобы нейросеть пропустила пользователя. Это создает серьёзный риск для сайтов: защита от ботов, позиционировавшаяся как прорыв, на деле не работает. Баг остается актуальным и позволяет спамерам легко автомат…
➡️ Читайте на сайте: https://aff.top/blog/novuiu-google-recapcha-proshli-statichnoi-kartinkoi
🧠 Ещё больше инсайтов → в канале AFF.top
Google выпустил обновленную reCAPTCHA, требующую движений рук для прохождения, но система оказалась уязвима к обходу. Достаточно транслировать статичное изображение с нужным жестом через виртуальную камеру с помощью простого Python-скрипта, чтобы нейросеть пропустила пользователя. Это создает серьёзный риск для сайтов: защита от ботов, позиционировавшаяся как прорыв, на деле не работает. Баг остается актуальным и позволяет спамерам легко автомат…
➡️ Читайте на сайте: https://aff.top/blog/novuiu-google-recapcha-proshli-statichnoi-kartinkoi
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
DeepSeek представит последнюю версию v4
DeepSeek выпустит v4 в середине июля с новой моделью ценообразования API: токены подорожают в 2 раза в часы пиковой нагрузки (09:00–12:00 и 14:00–18:00 по пекинскому времени). Компания планирует уведомлять пользователей по почте за 24 часа до изменения тарифов. Проблема с ошибками «server busy» останется, но обойдётся дороже — это может существенно повлиять на экономику проектов, которые активно используют API DeepSeek для автоматизации и масшта…
➡️ Читайте на сайте: https://aff.top/blog/deepseek-predstavit-posledniuiu-versiiu-v4
🧠 Ещё больше инсайтов → в канале AFF.top
DeepSeek выпустит v4 в середине июля с новой моделью ценообразования API: токены подорожают в 2 раза в часы пиковой нагрузки (09:00–12:00 и 14:00–18:00 по пекинскому времени). Компания планирует уведомлять пользователей по почте за 24 часа до изменения тарифов. Проблема с ошибками «server busy» останется, но обойдётся дороже — это может существенно повлиять на экономику проектов, которые активно используют API DeepSeek для автоматизации и масшта…
➡️ Читайте на сайте: https://aff.top/blog/deepseek-predstavit-posledniuiu-versiiu-v4
🧠 Ещё больше инсайтов → в канале AFF.top
Every community has a bus factor
Borrowed from software, 'bus factor' asks how many key people you'd have to lose before the project collapses. Communities have one too, and it's usually alarmingly low.
What the data shows
Message-attribution analyses of active servers routinely find that a tiny core — often under 10 people — produces the majority of substantive conversation, even in communities of thousands. When network analysis is applied, conversation frequently routes through a handful of high-centrality hubs. Lose two or three and activity can fall off a cliff.
Why it happens
Conversation is a reciprocal network, not a crowd. A few people initiate the threads everyone else replies to. They're load-bearing in a way headcount completely hides — the org chart of a community lives in who-replies-to-whom, not in the member list.
Discord vs Telegram
Telegram broadcast channels concentrate this to the extreme — bus factor is often one (the operator). Discord groups spread it slightly across active regulars.
The caveat
Centrality measured over short windows overstates fragility; some hubs are replaceable over time as others step up. The network adapts, sometimes.
Open question: can you deliberately lower your bus factor by cultivating second-tier hubs, or does conversational gravity resist being engineered?
Borrowed from software, 'bus factor' asks how many key people you'd have to lose before the project collapses. Communities have one too, and it's usually alarmingly low.
What the data shows
Message-attribution analyses of active servers routinely find that a tiny core — often under 10 people — produces the majority of substantive conversation, even in communities of thousands. When network analysis is applied, conversation frequently routes through a handful of high-centrality hubs. Lose two or three and activity can fall off a cliff.
Why it happens
Conversation is a reciprocal network, not a crowd. A few people initiate the threads everyone else replies to. They're load-bearing in a way headcount completely hides — the org chart of a community lives in who-replies-to-whom, not in the member list.
Discord vs Telegram
Telegram broadcast channels concentrate this to the extreme — bus factor is often one (the operator). Discord groups spread it slightly across active regulars.
The caveat
Centrality measured over short windows overstates fragility; some hubs are replaceable over time as others step up. The network adapts, sometimes.
Open question: can you deliberately lower your bus factor by cultivating second-tier hubs, or does conversational gravity resist being engineered?
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Anthropic выпустили Sonnet 5
30 июня вышла Claude Sonnet 5 — новая версия позиционируется как самая агентная в линейке и приближается к флагманской Opus 4.8. Модель лучше справляется со сложными многоуровневыми задачами, устойчива к вредоносным запросам и не генерирует эксплойты. Sonnet 5 доступна на Free-тарифе, но тестирование показало скромные улучшения: хотя работает лучше Sonnet 4.6, её обгоняют конкуренты, включая китайские модели, которые дешевле через API при лучшей…
➡️ Читайте на сайте: https://aff.top/blog/anthropic-vypustili-sonnet-5
🧠 Ещё больше инсайтов → в канале AFF.top
30 июня вышла Claude Sonnet 5 — новая версия позиционируется как самая агентная в линейке и приближается к флагманской Opus 4.8. Модель лучше справляется со сложными многоуровневыми задачами, устойчива к вредоносным запросам и не генерирует эксплойты. Sonnet 5 доступна на Free-тарифе, но тестирование показало скромные улучшения: хотя работает лучше Sonnet 4.6, её обгоняют конкуренты, включая китайские модели, которые дешевле через API при лучшей…
➡️ Читайте на сайте: https://aff.top/blog/anthropic-vypustili-sonnet-5
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Clickstar прекращает работу
Clickstar закрывается. Легендарная пуш-сеть прекращает закуп трафика с 1 августа, полная остановка — 20 августа.
Сетка работала почти 8 лет и была одним из лучших источников качественного трафика на Россию и СНГ. Сейчас пуш-трафик стал слишком ботовым из-за гугловских банов на скрипты сбора.
Что это означает для арбитражников — разбираемся в ста…
➡️ Читайте на сайте: https://aff.top/blog/clickstar-prekraschaet-rabotu
🧠 Ещё больше инсайтов → в канале AFF.top
Clickstar закрывается. Легендарная пуш-сеть прекращает закуп трафика с 1 августа, полная остановка — 20 августа.
Сетка работала почти 8 лет и была одним из лучших источников качественного трафика на Россию и СНГ. Сейчас пуш-трафик стал слишком ботовым из-за гугловских банов на скрипты сбора.
Что это означает для арбитражников — разбираемся в ста…
➡️ Читайте на сайте: https://aff.top/blog/clickstar-prekraschaet-rabotu
🧠 Ещё больше инсайтов → в канале AFF.top
Why events spike activity and don't move retention
AMAs, game nights and giveaways reliably spike the activity graph. The cohort follow-up data is sobering about whether any of it sticks.
What the data shows
Operators who track post-event cohorts tend to find a sharp activity spike during the event, a smaller echo for a day or two, then a return almost exactly to the pre-event baseline. Members acquired during an event-driven surge churn faster than baseline members. The graph looks like growth; the cohort table says otherwise.
Why it happens
Events pull forward existing engagement and attract event-seekers rather than community-seekers. The spike is largely the same regulars being more active at once, plus a low-intent inflow that leaves when the event ends. Activity and retention are different axes, and events move only the first.
Discord vs Telegram
Telegram's broadcast events (live streams, AMAs in comments) show the same shape — a view spike, then reversion — with even thinner conversion to durable participation.
The caveat
This varies by event type; recurring rituals may build habit where one-off spectacles don't. Few operators run the controlled comparison needed to tell them apart.
Open question: are events worth running for retention at all, or only for morale and content — and is that enough to justify them?
AMAs, game nights and giveaways reliably spike the activity graph. The cohort follow-up data is sobering about whether any of it sticks.
What the data shows
Operators who track post-event cohorts tend to find a sharp activity spike during the event, a smaller echo for a day or two, then a return almost exactly to the pre-event baseline. Members acquired during an event-driven surge churn faster than baseline members. The graph looks like growth; the cohort table says otherwise.
Why it happens
Events pull forward existing engagement and attract event-seekers rather than community-seekers. The spike is largely the same regulars being more active at once, plus a low-intent inflow that leaves when the event ends. Activity and retention are different axes, and events move only the first.
Discord vs Telegram
Telegram's broadcast events (live streams, AMAs in comments) show the same shape — a view spike, then reversion — with even thinner conversion to durable participation.
The caveat
This varies by event type; recurring rituals may build habit where one-off spectacles don't. Few operators run the controlled comparison needed to tell them apart.
Open question: are events worth running for retention at all, or only for morale and content — and is that enough to justify them?