Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Google стал помечать креативы, созданные ИИ
Google объявил, что начнёт помечать рекламные креативы, созданные нейросетями. Причина — ИИ-баннеры и видео стали слишком похожи на настоящие.
Формат и заметность маркировки будут зависеть от законов конкретного региона: где-то предупреждение появится прямо на креативе, где-то — в его информации.
Что это значит для арбитражников и когда правила …
➡️ Читайте на сайте: https://aff.top/blog/google-stal-pomechat-kreativy-sozdannye-ii
🧠 Ещё больше инсайтов → в канале AFF.top
Google объявил, что начнёт помечать рекламные креативы, созданные нейросетями. Причина — ИИ-баннеры и видео стали слишком похожи на настоящие.
Формат и заметность маркировки будут зависеть от законов конкретного региона: где-то предупреждение появится прямо на креативе, где-то — в его информации.
Что это значит для арбитражников и когда правила …
➡️ Читайте на сайте: https://aff.top/blog/google-stal-pomechat-kreativy-sozdannye-ii
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Компания Meta выпустила Muse Spark 1.1
Meta выпустила Muse Spark 1.1 почти одновременно с новой ChatGPT-5.6. Это мультимодальный агент, который сам дробит задачу на подзадачи и распределяет их между субагентами.
Стоимость тоже заметно ниже топовых западных моделей: $1.25 за миллион входных токенов и $4.25 за миллион выходных.
Но главный вопрос — насколько она реально сильна на фоне к…
➡️ Читайте на сайте: https://aff.top/blog/kompaniia-meta-vypustila-muse-spark-1-1
🧠 Ещё больше инсайтов → в канале AFF.top
Meta выпустила Muse Spark 1.1 почти одновременно с новой ChatGPT-5.6. Это мультимодальный агент, который сам дробит задачу на подзадачи и распределяет их между субагентами.
Стоимость тоже заметно ниже топовых западных моделей: $1.25 за миллион входных токенов и $4.25 за миллион выходных.
Но главный вопрос — насколько она реально сильна на фоне к…
➡️ Читайте на сайте: https://aff.top/blog/kompaniia-meta-vypustila-muse-spark-1-1
🧠 Ещё больше инсайтов → в канале AFF.top
What do you do when you can't randomize at the user level?
The question: cookie loss and privacy walls are making clean user-level holdouts harder every quarter. When you can't split individuals, can you still run a credible causal test?
What the methodology says: yes — geo experiments randomize at the market level. You turn a channel on in some metro areas and off (or down) in matched others, then measure the difference in outcomes. The credibility comes from how you pick the controls. Naive geo tests fail because markets aren't comparable. The serious version uses synthetic control — constructing a weighted blend of untreated markets that closely tracks the treated market's pre-period behavior, so the post-period gap is a clean estimate of lift.
What the research shows: methods like 'GeoLift' and Bayesian structural time-series ('CausalImpact') formalize this. Validation studies find geo experiments recover incrementality estimates close to user-level RCTs when the pre-period match is tight and enough geos are used — but the estimates get unstable fast with few markets or a short baseline.
The nuance: geo tests measure aggregate market lift, not individual causation. They also can't isolate creative or audience effects within a channel — only the channel's net market impact. And spillover between adjacent markets biases results toward zero.
What to actually do: require at least a few months of stable pre-period data and enough matched markets for a synthetic control to converge. Pre-register the analysis window before you peek.
Bottom line for practitioners: when you lose the individual, randomize the market — but a geo test is only as honest as the counterfactual you build for it.
The question: cookie loss and privacy walls are making clean user-level holdouts harder every quarter. When you can't split individuals, can you still run a credible causal test?
What the methodology says: yes — geo experiments randomize at the market level. You turn a channel on in some metro areas and off (or down) in matched others, then measure the difference in outcomes. The credibility comes from how you pick the controls. Naive geo tests fail because markets aren't comparable. The serious version uses synthetic control — constructing a weighted blend of untreated markets that closely tracks the treated market's pre-period behavior, so the post-period gap is a clean estimate of lift.
What the research shows: methods like 'GeoLift' and Bayesian structural time-series ('CausalImpact') formalize this. Validation studies find geo experiments recover incrementality estimates close to user-level RCTs when the pre-period match is tight and enough geos are used — but the estimates get unstable fast with few markets or a short baseline.
The nuance: geo tests measure aggregate market lift, not individual causation. They also can't isolate creative or audience effects within a channel — only the channel's net market impact. And spillover between adjacent markets biases results toward zero.
What to actually do: require at least a few months of stable pre-period data and enough matched markets for a synthetic control to converge. Pre-register the analysis window before you peek.
Bottom line for practitioners: when you lose the individual, randomize the market — but a geo test is only as honest as the counterfactual you build for it.
How much credit does an ad nobody clicked deserve?
The question: view-through conversions — credit assigned because a user saw an impression and later converted without ever clicking — are one of the most disputed line items in attribution. Are they real causal influence or sophisticated double-counting?
What the evidence says: the honest answer is 'it depends, and usually less than reported.' Controlled studies of display and video repeatedly find view-through credit is heavily contaminated by two effects. First, targeting endogeneity: ad systems show impressions to users already likely to convert, so the impression correlates with conversion without causing it. Second, the standard view-through windows (24 hours to 30 days) are wide enough to sweep in conversions that would have happened regardless. Incrementality experiments often find true view-through lift is a fraction of the platform-reported figure.
The nuance: it's not zero. Brand and upper-funnel video genuinely move users who never click — that's the whole point of awareness media. Dismissing all view-through as fraud is as wrong as accepting it at face value. The error is in magnitude and in attributing it without a counterfactual.
What to actually do: never include view-through credit in a model that drives budget decisions unless it's been calibrated against a holdout. Shorten windows aggressively as a sanity check — if 80% of view-through credit lives in the 7-30 day tail, most of it is correlation.
Bottom line for practitioners: a view is the weakest causal claim in attribution. Demand an experiment before you pay it like a click.
The question: view-through conversions — credit assigned because a user saw an impression and later converted without ever clicking — are one of the most disputed line items in attribution. Are they real causal influence or sophisticated double-counting?
What the evidence says: the honest answer is 'it depends, and usually less than reported.' Controlled studies of display and video repeatedly find view-through credit is heavily contaminated by two effects. First, targeting endogeneity: ad systems show impressions to users already likely to convert, so the impression correlates with conversion without causing it. Second, the standard view-through windows (24 hours to 30 days) are wide enough to sweep in conversions that would have happened regardless. Incrementality experiments often find true view-through lift is a fraction of the platform-reported figure.
The nuance: it's not zero. Brand and upper-funnel video genuinely move users who never click — that's the whole point of awareness media. Dismissing all view-through as fraud is as wrong as accepting it at face value. The error is in magnitude and in attributing it without a counterfactual.
What to actually do: never include view-through credit in a model that drives budget decisions unless it's been calibrated against a holdout. Shorten windows aggressively as a sanity check — if 80% of view-through credit lives in the 7-30 day tail, most of it is correlation.
Bottom line for practitioners: a view is the weakest causal claim in attribution. Demand an experiment before you pay it like a click.
Why is the attribution window the most consequential decision nobody debates?
The question: every attribution system has a lookback window — 7 days, 30, 90 — defining how far back a touch can claim credit. It's usually set once and forgotten. Yet it silently determines which channels win.
What the data shows: window length is not a neutral technical setting; it's a thumb on the scale. Lengthen the window and you mechanically shift credit toward early, upper-funnel touches that would otherwise fall outside the lookback. Shorten it and you concentrate credit on closers. Studies of path-length distributions find that for considered purchases, a meaningful share of influential first-touches sit beyond a 30-day window entirely — so a 30-day setting structurally undercounts top-of-funnel by construction.
The deeper issue: the 'right' window is a function of your actual purchase cycle, which you can measure — the distribution of time-from-first-touch-to-conversion. Most teams instead inherit the platform default and never check whether it matches their funnel. A 7-day window on a 60-day B2B cycle isn't aggressive, it's broken.
The nuance: longer isn't simply better. Wide windows inflate spurious credit by absorbing coincidental touches — the same over-crediting problem as long view-through windows.
What to actually do: plot your time-to-conversion distribution and set the window to capture the bulk of it (say, the 90th percentile), not a round number someone picked. Re-examine it when your funnel changes.
Bottom line for practitioners: the attribution window is a hidden weighting parameter. Choosing it by default is choosing your conclusions by default.
The question: every attribution system has a lookback window — 7 days, 30, 90 — defining how far back a touch can claim credit. It's usually set once and forgotten. Yet it silently determines which channels win.
What the data shows: window length is not a neutral technical setting; it's a thumb on the scale. Lengthen the window and you mechanically shift credit toward early, upper-funnel touches that would otherwise fall outside the lookback. Shorten it and you concentrate credit on closers. Studies of path-length distributions find that for considered purchases, a meaningful share of influential first-touches sit beyond a 30-day window entirely — so a 30-day setting structurally undercounts top-of-funnel by construction.
The deeper issue: the 'right' window is a function of your actual purchase cycle, which you can measure — the distribution of time-from-first-touch-to-conversion. Most teams instead inherit the platform default and never check whether it matches their funnel. A 7-day window on a 60-day B2B cycle isn't aggressive, it's broken.
The nuance: longer isn't simply better. Wide windows inflate spurious credit by absorbing coincidental touches — the same over-crediting problem as long view-through windows.
What to actually do: plot your time-to-conversion distribution and set the window to capture the bulk of it (say, the 90th percentile), not a round number someone picked. Re-examine it when your funnel changes.
Bottom line for practitioners: the attribution window is a hidden weighting parameter. Choosing it by default is choosing your conclusions by default.
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Российские букмекеры увеличили закуп трафика с мобильных приложений
Российские букмекеры в 1 квартале 2026 года заметно нарастили закупку трафика из мобильных приложений. На фоне ужесточения регулирования они смещают бюджеты в новые каналы, где ещё есть живой трафик.
По данным UMG, доля in-app-рекламы выросла с 3-4% до 5-6% при объёме рынка около 10 млрд рублей. Но это может быть только начало — в блоге разбираем…
➡️ Читайте на сайте: https://aff.top/blog/rossiiskie-bukmekery-uvelichili-zakup-trafika-s-mobilnykh-prilozhenii
🧠 Ещё больше инсайтов → в канале AFF.top
Российские букмекеры в 1 квартале 2026 года заметно нарастили закупку трафика из мобильных приложений. На фоне ужесточения регулирования они смещают бюджеты в новые каналы, где ещё есть живой трафик.
По данным UMG, доля in-app-рекламы выросла с 3-4% до 5-6% при объёме рынка около 10 млрд рублей. Но это может быть только начало — в блоге разбираем…
➡️ Читайте на сайте: https://aff.top/blog/rossiiskie-bukmekery-uvelichili-zakup-trafika-s-mobilnykh-prilozhenii
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Mia Khalifa стала амбпссадором 1win
Mia Khalifa снова засветилась рядом с 1win: в Instagram она показала цепочку с логотипом бренда и статус «VIP 1win».
Параллельно всплыла история с ее ставкой на Испанию на ЧМ — по данным поста, выигрыш мог составить 200к$.
Официального подтверждения амбассадорства пока нет, но для арбитражников это уже повод для новых креативов. Что именно здесь…
➡️ Читайте на сайте: https://aff.top/blog/mia-khalifa-stala-ambpssadorom-1win
🧠 Ещё больше инсайтов → в канале AFF.top
Mia Khalifa снова засветилась рядом с 1win: в Instagram она показала цепочку с логотипом бренда и статус «VIP 1win».
Параллельно всплыла история с ее ставкой на Испанию на ЧМ — по данным поста, выигрыш мог составить 200к$.
Официального подтверждения амбассадорства пока нет, но для арбитражников это уже повод для новых креативов. Что именно здесь…
➡️ Читайте на сайте: https://aff.top/blog/mia-khalifa-stala-ambpssadorom-1win
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from КРАВЧЕНКО
Дорогие коллеги и партнеры,
Наш маршрут конференций за последние недели, получился особенно насыщенным.
Со стендами PoshFriends мы побывали на MAC и GGate, а затем продолжили встречи уже в полях iGB Live в Лондоне.
В Ереване увиделись с любимыми SEO-командами, попробовали местные вина, обменялись новостями и зарядились энергией УБТ-команд.
В Тбилиси обсуждали тренды, новые связки и совместные планы, встречались с действующими партнерами и знакомились с новыми. А за настроение на стенде отвечала Черемша, которая чуть не стала маскотом одного из наших продуктов. С этой задачей, кажется, справилась лучше всех.
В Лондоне все было уже по-деловому. Провели серию встреч с топ-партнерами, обсудили Японию, бурж и новые точки роста. География интересов растет, планы становятся амбициознее. Воротники, как выяснилось, нагладили не зря.
Спасибо всем, с кем удалось увидеться на этом маршруте. За открытые разговоры, новые идеи, доверие и планы, которые постепенно превращаются в реальные проекты.
Конференционный сезон продолжается. Скоро увидимся снова.
Всегда ваши, Команда Posh Friends 🤝
Наш маршрут конференций за последние недели, получился особенно насыщенным.
Со стендами PoshFriends мы побывали на MAC и GGate, а затем продолжили встречи уже в полях iGB Live в Лондоне.
В Ереване увиделись с любимыми SEO-командами, попробовали местные вина, обменялись новостями и зарядились энергией УБТ-команд.
В Тбилиси обсуждали тренды, новые связки и совместные планы, встречались с действующими партнерами и знакомились с новыми. А за настроение на стенде отвечала Черемша, которая чуть не стала маскотом одного из наших продуктов. С этой задачей, кажется, справилась лучше всех.
В Лондоне все было уже по-деловому. Провели серию встреч с топ-партнерами, обсудили Японию, бурж и новые точки роста. География интересов растет, планы становятся амбициознее. Воротники, как выяснилось, нагладили не зря.
Спасибо всем, с кем удалось увидеться на этом маршруте. За открытые разговоры, новые идеи, доверие и планы, которые постепенно превращаются в реальные проекты.
Конференционный сезон продолжается. Скоро увидимся снова.
Всегда ваши, Команда Posh Friends 🤝
Does retargeting work, or does it just take credit for people coming back anyway?
The question: retargeting reliably shows spectacular last-click and even data-driven ROIs. It also shows ads to people who already visited your site — i.e., people already predisposed to return. How do you separate the ad's effect from the user's pre-existing intent?
What the research found: this is the textbook case of selection bias in attribution, and it's been tested rigorously. Multiple large-scale incrementality experiments on retargeting found that observational ROI dramatically overstates causal lift — in several published studies, true incremental conversions were a small fraction of attributed ones, and in some segments the incremental effect was statistically indistinguishable from zero. The retargeting wasn't causing the return; it was advertising to people who were returning regardless, then claiming them.
The nuance: retargeting absolutely can be incremental — for cart-abandoners with a real nudge, for long consideration cycles, for re-activating lapsed users. The point is not 'retargeting is fake.' It's that observational attribution cannot tell you which case you're in, because the same data pattern (saw ad, converted) appears whether the ad caused it or not. That's the definition of confounding.
What to actually do: run a holdout where a random slice of your retargeting pool gets no ads. Measure the conversion gap. Budget to the gap, not to the attributed total.
Bottom line for practitioners: retargeting is where correlation and causation diverge most violently. Without a holdout, you are paying to advertise to people who already decided.
The question: retargeting reliably shows spectacular last-click and even data-driven ROIs. It also shows ads to people who already visited your site — i.e., people already predisposed to return. How do you separate the ad's effect from the user's pre-existing intent?
What the research found: this is the textbook case of selection bias in attribution, and it's been tested rigorously. Multiple large-scale incrementality experiments on retargeting found that observational ROI dramatically overstates causal lift — in several published studies, true incremental conversions were a small fraction of attributed ones, and in some segments the incremental effect was statistically indistinguishable from zero. The retargeting wasn't causing the return; it was advertising to people who were returning regardless, then claiming them.
The nuance: retargeting absolutely can be incremental — for cart-abandoners with a real nudge, for long consideration cycles, for re-activating lapsed users. The point is not 'retargeting is fake.' It's that observational attribution cannot tell you which case you're in, because the same data pattern (saw ad, converted) appears whether the ad caused it or not. That's the definition of confounding.
What to actually do: run a holdout where a random slice of your retargeting pool gets no ads. Measure the conversion gap. Budget to the gap, not to the attributed total.
Bottom line for practitioners: retargeting is where correlation and causation diverge most violently. Without a holdout, you are paying to advertise to people who already decided.
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Короткий домен Telegram перестал работать
Telegram лишился домена t.me: он разделегирован и больше не работает на уровне регистратора. Платформа срочно переезжает на telegram.me, а владельцам крупных каналов стоит обновить публичные ссылки. Сроки восстановления неизвестны, и есть риск, что t.me не вернётся вовсе на фоне давления на Telegram.
➡️ Читайте на сайте: https://aff.top/blog/korotkii-domen-telegram-perestal-rabotat
🧠 Ещё больше инсайтов → в канале AFF.top
Telegram лишился домена t.me: он разделегирован и больше не работает на уровне регистратора. Платформа срочно переезжает на telegram.me, а владельцам крупных каналов стоит обновить публичные ссылки. Сроки восстановления неизвестны, и есть риск, что t.me не вернётся вовсе на фоне давления на Telegram.
➡️ Читайте на сайте: https://aff.top/blog/korotkii-domen-telegram-perestal-rabotat
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Youtube тестирует поиск с AI
YouTube начал тестировать Ask YouTube — поиск с ИИ, где можно задавать вопросы обычным языком и получать не список ссылок, а готовую подборку видео и фрагментов.
Фича уже доступна в США и работает на сложные запросы: если нужно, ИИ уточняет вопрос и подсказывает следующий шаг.
Что это значит для поиска на YouTube и когда новинка дойдёт до других…
➡️ Читайте на сайте: https://aff.top/blog/youtube-testiruet-poisk-s-ai
🧠 Ещё больше инсайтов → в канале AFF.top
YouTube начал тестировать Ask YouTube — поиск с ИИ, где можно задавать вопросы обычным языком и получать не список ссылок, а готовую подборку видео и фрагментов.
Фича уже доступна в США и работает на сложные запросы: если нужно, ИИ уточняет вопрос и подсказывает следующий шаг.
Что это значит для поиска на YouTube и когда новинка дойдёт до других…
➡️ Читайте на сайте: https://aff.top/blog/youtube-testiruet-poisk-s-ai
🧠 Ещё больше инсайтов → в канале AFF.top
Is the 'naive' linear attribution model actually the most honest one?
The question: linear attribution splits credit equally across every touch in the path. It's dismissed as simplistic — the model you use when you've given up. But there's a contrarian case that it's the most intellectually honest default. Is it?
The argument for it: every weighted model — U-shaped, time-decay, even data-driven — encodes a claim about which touches mattered more. Linear makes the opposite, deliberately agnostic claim: we don't know, so we won't pretend. In the absence of a causal experiment, any non-uniform weighting is a hypothesis dressed as a result. Linear refuses to invent precision it doesn't have.
What the analysis shows: studies comparing models find linear is rarely the best fit to conversion data — but it's also rarely catastrophically wrong, and crucially it has no systematic directional bias toward openers or closers. Time-decay over-credits the bottom; first-touch over-credits the top; linear sits in the middle by construction. Its errors are diffuse rather than skewed.
The nuance: 'no bias' isn't 'accurate.' Linear is provably wrong whenever some touches genuinely matter more (they usually do). Its honesty is the honesty of a shrug — fine as a baseline, weak as a decision tool.
What to actually do: use linear as your reference point. When a fancier model disagrees with linear, ask whether the difference is evidence-backed or just the model's built-in prior talking.
Bottom line for practitioners: linear attribution is the null hypothesis of crediting. Don't ship it — but make every other model beat it on evidence, not on vibes.
The question: linear attribution splits credit equally across every touch in the path. It's dismissed as simplistic — the model you use when you've given up. But there's a contrarian case that it's the most intellectually honest default. Is it?
The argument for it: every weighted model — U-shaped, time-decay, even data-driven — encodes a claim about which touches mattered more. Linear makes the opposite, deliberately agnostic claim: we don't know, so we won't pretend. In the absence of a causal experiment, any non-uniform weighting is a hypothesis dressed as a result. Linear refuses to invent precision it doesn't have.
What the analysis shows: studies comparing models find linear is rarely the best fit to conversion data — but it's also rarely catastrophically wrong, and crucially it has no systematic directional bias toward openers or closers. Time-decay over-credits the bottom; first-touch over-credits the top; linear sits in the middle by construction. Its errors are diffuse rather than skewed.
The nuance: 'no bias' isn't 'accurate.' Linear is provably wrong whenever some touches genuinely matter more (they usually do). Its honesty is the honesty of a shrug — fine as a baseline, weak as a decision tool.
What to actually do: use linear as your reference point. When a fancier model disagrees with linear, ask whether the difference is evidence-backed or just the model's built-in prior talking.
Bottom line for practitioners: linear attribution is the null hypothesis of crediting. Don't ship it — but make every other model beat it on evidence, not on vibes.
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Z.ai анонсировала новую GLM-5.5
Z.ai готовит релиз флагманской GLM-5.5: модель обещают показать в августе 2026 года.
Главная интрига — рост до 1 трлн параметров при том же контекстном окне в 1 млн токенов. Новинка снова будет заточена под код и агентные задачи.
Почему версия сразу 5.5, без 5.3 и 5.4, и что это может означать для рынка — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/z-ai-anonsirovala-novuiu-glm-5-5
🧠 Ещё больше инсайтов → в канале AFF.top
Z.ai готовит релиз флагманской GLM-5.5: модель обещают показать в августе 2026 года.
Главная интрига — рост до 1 трлн параметров при том же контекстном окне в 1 млн токенов. Новинка снова будет заточена под код и агентные задачи.
Почему версия сразу 5.5, без 5.3 и 5.4, и что это может означать для рынка — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/z-ai-anonsirovala-novuiu-glm-5-5
🧠 Ещё больше инсайтов → в канале AFF.top
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Telegram запустил собственный сервер для ботов
Telegram запустил собственный сервер для ботов и мани-приложений: теперь backend можно размещать прямо внутри инфраструктуры мессенджера.
Сервер работает на JavaScript/TypeScript, через вебхуки, и позволяет подключать SQL-базу для сбора контактов без посредников.
Пока неясны цена и ограничения — что именно уже можно тестировать, а где скрыт подв…
➡️ Читайте на сайте: https://aff.top/blog/telegram-zapustil-sobstvennyi-server-dlia-botov
🧠 Ещё больше инсайтов → в канале AFF.top
Telegram запустил собственный сервер для ботов и мани-приложений: теперь backend можно размещать прямо внутри инфраструктуры мессенджера.
Сервер работает на JavaScript/TypeScript, через вебхуки, и позволяет подключать SQL-базу для сбора контактов без посредников.
Пока неясны цена и ограничения — что именно уже можно тестировать, а где скрыт подв…
➡️ Читайте на сайте: https://aff.top/blog/telegram-zapustil-sobstvennyi-server-dlia-botov
🧠 Ещё больше инсайтов → в канале AFF.top
Is your incrementality win real, or did you just test during a novelty spike?
The question: you run a clean lift test on a new channel or creative, see strong incremental conversions, and scale. Three months later the lift has evaporated. Was the experiment wrong, or did the world change?
What the literature describes: this is the novelty effect (and its cousin, ad fatigue) — well-documented in both UX and advertising research. Early exposure to something new generates a temporary response that decays as the audience habituates. A short incrementality test run at launch captures the peak, not the steady state. The estimate isn't biased by bad design; it's biased by timing — you measured a transient.
The deeper problem: this isn't correlation-vs-causation. The causal effect was real during the test. The error is assuming a point-in-time causal estimate is stationary — that what lifted conversions in week one will lift them identically in month six. Most attribution and incrementality work implicitly assumes stationarity, and audiences are anything but stationary.
The nuance: the inverse also happens — some effects build over time (brand, retargeting saturation), so a too-early test understates them. The direction depends on the channel.
What to actually do: run lift tests over a duration long enough to see the decay curve flatten, not just clear significance. Re-test scaled channels periodically — incrementality is a depreciating asset, not a constant.
Bottom line for practitioners: a single incrementality number is a snapshot of a moving target. Measure the curve, not the peak, before you bet the budget.
The question: you run a clean lift test on a new channel or creative, see strong incremental conversions, and scale. Three months later the lift has evaporated. Was the experiment wrong, or did the world change?
What the literature describes: this is the novelty effect (and its cousin, ad fatigue) — well-documented in both UX and advertising research. Early exposure to something new generates a temporary response that decays as the audience habituates. A short incrementality test run at launch captures the peak, not the steady state. The estimate isn't biased by bad design; it's biased by timing — you measured a transient.
The deeper problem: this isn't correlation-vs-causation. The causal effect was real during the test. The error is assuming a point-in-time causal estimate is stationary — that what lifted conversions in week one will lift them identically in month six. Most attribution and incrementality work implicitly assumes stationarity, and audiences are anything but stationary.
The nuance: the inverse also happens — some effects build over time (brand, retargeting saturation), so a too-early test understates them. The direction depends on the channel.
What to actually do: run lift tests over a duration long enough to see the decay curve flatten, not just clear significance. Re-test scaled channels periodically — incrementality is a depreciating asset, not a constant.
Bottom line for practitioners: a single incrementality number is a snapshot of a moving target. Measure the curve, not the peak, before you bet the budget.
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Google картинки станут конкурентом Pinterest
Google Картинки начали превращать в полноценную платформу с персональной лентой по прошлым запросам — по сути, в аналог Pinterest.
Во вкладке For you уже тестируют подборки, а ещё обещают коллекции и генерацию изображений во встроенной Nano Banana.
Как это будет работать и когда новинка дойдёт до других стран — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/google-kartinki-stanut-konkurentom-pinterest
🧠 Ещё больше инсайтов → в канале AFF.top
Google Картинки начали превращать в полноценную платформу с персональной лентой по прошлым запросам — по сути, в аналог Pinterest.
Во вкладке For you уже тестируют подборки, а ещё обещают коллекции и генерацию изображений во встроенной Nano Banana.
Как это будет работать и когда новинка дойдёт до других стран — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/google-kartinki-stanut-konkurentom-pinterest
🧠 Ещё больше инсайтов → в канале AFF.top
Are your newest campaigns being unfairly penalized by math?
The question: you compare last week's campaign to last quarter's and the new one looks worse. But conversions take time to land. Could the comparison itself be structurally rigged against anything recent?
What's happening: yes — this is conversion lag bias (sometimes called the delayed-feedback or attribution-maturation problem). Conversions arrive on a distribution: some same-day, many over the following days and weeks. Any campaign you measure 'now' has only collected its early conversions; its long-tail hasn't matured. Compared against a fully-matured older campaign, the recent one looks artificially weak — not because it performed worse, but because it isn't done converting.
What the research formalizes: delayed-feedback models in computational advertising treat each conversion's delay as a random variable and model the survival curve, so a campaign's expected final performance can be estimated before all conversions land. Without that correction, optimization systems that judge campaigns too early will systematically kill long-cycle channels prematurely — the channels whose conversions arrive slowest get cut before they can prove themselves.
The nuance: the bias is worst for considered purchases and B2B, negligible for impulse e-commerce. It also interacts nastily with the attribution window — a short window plus lag bias double-penalizes long cycles.
What to actually do: never compare a fresh campaign's raw conversions to a matured one's. Either wait for maturation or model the lag curve and compare projected finals. Build a 'cohort maturity' check into reporting.
Bottom line for practitioners: recent campaigns aren't underperforming, they're under-aged. Compare like-aged cohorts or you'll execute your slowest-converting channels for the crime of being measured too soon.
The question: you compare last week's campaign to last quarter's and the new one looks worse. But conversions take time to land. Could the comparison itself be structurally rigged against anything recent?
What's happening: yes — this is conversion lag bias (sometimes called the delayed-feedback or attribution-maturation problem). Conversions arrive on a distribution: some same-day, many over the following days and weeks. Any campaign you measure 'now' has only collected its early conversions; its long-tail hasn't matured. Compared against a fully-matured older campaign, the recent one looks artificially weak — not because it performed worse, but because it isn't done converting.
What the research formalizes: delayed-feedback models in computational advertising treat each conversion's delay as a random variable and model the survival curve, so a campaign's expected final performance can be estimated before all conversions land. Without that correction, optimization systems that judge campaigns too early will systematically kill long-cycle channels prematurely — the channels whose conversions arrive slowest get cut before they can prove themselves.
The nuance: the bias is worst for considered purchases and B2B, negligible for impulse e-commerce. It also interacts nastily with the attribution window — a short window plus lag bias double-penalizes long cycles.
What to actually do: never compare a fresh campaign's raw conversions to a matured one's. Either wait for maturation or model the lag curve and compare projected finals. Build a 'cohort maturity' check into reporting.
Bottom line for practitioners: recent campaigns aren't underperforming, they're under-aged. Compare like-aged cohorts or you'll execute your slowest-converting channels for the crime of being measured too soon.
Forwarded from AFF.TOP
This media is not supported in your browser
VIEW IN TELEGRAM
Россияне не смогут покупать стейблкоины
Россиянам могут закрыть доступ к покупке стейблкоинов: в новой версии закона их приравняли к иностранным активам.
Купить такие токены смогут только квалифицированные инвесторы — например, с активами от 24 млн рублей или доходом от 12 млн в год.
Что это значит для обычных пользователей и когда правило заработает — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/rossiiane-ne-smogut-pokupat-steiblkoiny
🧠 Ещё больше инсайтов → в канале AFF.top
Россиянам могут закрыть доступ к покупке стейблкоинов: в новой версии закона их приравняли к иностранным активам.
Купить такие токены смогут только квалифицированные инвесторы — например, с активами от 24 млн рублей или доходом от 12 млн в год.
Что это значит для обычных пользователей и когда правило заработает — в блоге.
➡️ Читайте на сайте: https://aff.top/blog/rossiiane-ne-smogut-pokupat-steiblkoiny
🧠 Ещё больше инсайтов → в канале AFF.top
23-24 июля встречаемся в Лимассоле! 🔥
Команда AdsCard врывается на Conversion Conf в статусе HOOKAH LOUNGE SPONSOR! Мы готовим для вас идеальное пространство для неформального общения и обсуждения серьезных дел.
Ищете платежное решение, которое не подведет в самый ответственный момент? Хотите масштабировать свои рекламные кампании без головной боли? Давайте обсудим это в расслабленной атмосфере.
Что ждет вас в нашей лаунж-зоне?
0️⃣ Поделимся инсайдами и свежими кейсами по заливу с наших карт на самых требовательных источниках.
0️⃣ Обсудим наши эксклюзивные условия для команд и расскажем, как получить максимум от нашего сервиса.
0️⃣ Познакомим с топами индустрии, угостим дымным кальяном и просто отлично проведем время.
Присоединяйтесь к нам, чтобы совместить приятное с полезным: качественный нетворкинг и эффективные платежные решения.
📍 Где искать: Parklane Hotel, HOOKAH LOUNGE от AdsCard
Ждем всех на Conversion Conf для незабываемого ивента и крутых знакомств! До встречи! 😎
Команда AdsCard врывается на Conversion Conf в статусе HOOKAH LOUNGE SPONSOR! Мы готовим для вас идеальное пространство для неформального общения и обсуждения серьезных дел.
Ищете платежное решение, которое не подведет в самый ответственный момент? Хотите масштабировать свои рекламные кампании без головной боли? Давайте обсудим это в расслабленной атмосфере.
Что ждет вас в нашей лаунж-зоне?
Присоединяйтесь к нам, чтобы совместить приятное с полезным: качественный нетворкинг и эффективные платежные решения.
📍 Где искать: Parklane Hotel, HOOKAH LOUNGE от AdsCard
Ждем всех на Conversion Conf для незабываемого ивента и крутых знакомств! До встречи! 😎
Please open Telegram to view this post
VIEW IN TELEGRAM
Can a channel improve in every segment yet look worse overall?
The question: a channel's conversion rate rose in mobile, rose in desktop, rose in tablet — but its blended rate fell. Your reporting says the channel got worse. Every segment says it got better. Who's lying?
What's at work: nobody is. This is Simpson's paradox — when a trend present in subgroups reverses or vanishes after aggregation, because the mix of subgroups shifted. If the channel's traffic moved toward a lower-converting segment (say, more cold prospecting, less warm retargeting), the blended average can drop even as every component rises. The aggregate is a weighted average, and the weights moved.
Why it matters for attribution specifically: attribution reports live and die on aggregated rates — conversions per channel, ROI per source. Any time the underlying composition of a channel's traffic changes (new geos, new placements, new audience mix), the headline number can move in the opposite direction from reality. Acting on the aggregate without checking the mix is how teams cut channels that are genuinely improving.
The nuance: the paradox cuts both ways — a channel can look like it's improving in aggregate purely because its mix shifted toward easy converters, masking a real decline in every segment.
What to actually do: never read a channel's blended rate without a composition check. Decompose into stable segments and look at within-segment trends and the mix shift separately. If they disagree, trust the segments and investigate the mix.
Bottom line for practitioners: aggregated attribution numbers can flip the truth. When the total contradicts every segment, the total is an artifact of the mix — not a finding.
The question: a channel's conversion rate rose in mobile, rose in desktop, rose in tablet — but its blended rate fell. Your reporting says the channel got worse. Every segment says it got better. Who's lying?
What's at work: nobody is. This is Simpson's paradox — when a trend present in subgroups reverses or vanishes after aggregation, because the mix of subgroups shifted. If the channel's traffic moved toward a lower-converting segment (say, more cold prospecting, less warm retargeting), the blended average can drop even as every component rises. The aggregate is a weighted average, and the weights moved.
Why it matters for attribution specifically: attribution reports live and die on aggregated rates — conversions per channel, ROI per source. Any time the underlying composition of a channel's traffic changes (new geos, new placements, new audience mix), the headline number can move in the opposite direction from reality. Acting on the aggregate without checking the mix is how teams cut channels that are genuinely improving.
The nuance: the paradox cuts both ways — a channel can look like it's improving in aggregate purely because its mix shifted toward easy converters, masking a real decline in every segment.
What to actually do: never read a channel's blended rate without a composition check. Decompose into stable segments and look at within-segment trends and the mix shift separately. If they disagree, trust the segments and investigate the mix.
Bottom line for practitioners: aggregated attribution numbers can flip the truth. When the total contradicts every segment, the total is an artifact of the mix — not a finding.
Pairs well with this channel
@TheOpsPlaybook — Battle-tested SOPs, checklists, and frameworks for running digital operations —… Quietly one of the better feeds in the space.
@TheOpsPlaybook — Battle-tested SOPs, checklists, and frameworks for running digital operations —… Quietly one of the better feeds in the space.