Marketero · Entrevistas
Dynamic Creative & Machine Learning
25 ideas de marketing sacadas de «Dynamic Creative & Machine Learning», de Professor Charley T. Sobre todo de medición y experimentos, canales de…
- 25 ideas despiezadas
- Fuente Professor Charley T
- Dura 54 min
- Publicada en 2023
Ver la entrevista entera en YouTube
De qué habla
Las 25 ideas de esta entrevista se reparten así entre las etapas del marketing. El reparto es lo que distingue una entrevista de otra: dice en qué se detuvo de verdad quien hablaba.
Las 25 ideas, con su cita y su minuto
Cada una está sacada de un punto concreto de la entrevista. El enlace del titular lleva a
la página de esa idea, donde se ve quién más la dice por separado; el del pie salta al
segundo exacto del vídeo para que la compruebes tú.
Canales de adquisiciónNo se debe descartar un anuncio por considerarlo estéticamente simple o poco refinado; en publicidad digital, imágenes improvisadas y de baja producción a menudo superan sistemáticamente a campañas de alta gama o colaboraciones costosas con influencers.
a beta Kardashian 100 grand to Shell the shake I think we made 15K on chest the best ad we had running at that was a stupid picture that the receptionist took in the kitchen making a shake with an iPhone and it was like and we just throw it up as an Instagram creative agency and this stupid picture it was literally that cheesy picture of the the company people ins in spend again it would just Crush everything in my being was like this is the worst ad ever like the copy was didn't read it was just like never be attached or have dogmas because the weird can win and you could you as a choices do…
Escúchalo en el minuto 37:06
Medición y experimentosLa decisión de separar un producto en embudos y campañas distintos para diferentes perfiles de cliente no debe basarse solo en el ángulo creativo, sino en si difieren en sus unit economics (CPA, LTV y comportamiento de retención). Si las métricas económicas coinciden, se gestionan bajo un único objetivo de negocio; si difieren, se separan según las necesidades de flujo de caja e inventario.
first thing I'm going to do is a PSM analysis on those audiences on and on that beauty buyer do they behave the same as the weight loss buyer or the you know right do they behave the same do they have the same second practice right LTV CPA do the unit economics of that type of buyer match and if so then yes it's one business objective one campaign if the unit economics about attention are different then I will separate them out with different business objectives right but like if you have different unit economics for customer Journeys I want to identify this Union economics and then…
Escúchalo en el minuto 42:13
Briefs creativos basados en datos de control y objetivos de negocio específicos
Estrategia de contenidosEn lugar de encargar creatividad abierta a agencias o creadores, se deben elaborar briefs hiperespecíficos basados en las métricas de control existentes (frecuencia, CPM por etapa del embudo y lift incremental) para evaluar al equipo creativo según su tasa de éxito al resolver un problema puntual de rentabilidad.
say now let's take it to the nth degree I've got a very simple campaign when I got four five six control ads and I know the frequency and the CPM by day on these ads they know how they're affecting my Omni Channel marketing and I've got the incremental lift nailed down I can go to my creative team and say I need a video ad that will Target mid funnel that's going to have a better CPM or CPA than something else here's the thing it has to be now my creative shop but now that creative team is being evaluated on their hit rate of taking the extraordinarily specific creative brief of the business…
Escúchalo en el minuto 23:37
Medición y experimentosUn anuncio de Facebook con un costo por adquisición (CPA) aparentemente malo en plataforma puede estar alimentando el volumen de búsqueda de marca y la captación de correo. Facebook funciona como un amplificador de atención para todo el modelo de negocio, por lo que apagar anuncios basándose solo en métricas internas puede destruir ingresos globales.
what is that as a bad CPA the answer might be no like I was just bad ad gets all the spend and as soon as they turn it off their search volume goes down by like 20 their email on take rate moves down by like five or ten as soon as they turn that bad ad off they lose a ton of money source of Revenue Facebook is a business model amplification device you are buying attention and your job as a business operator is to monetize that attention the most
Escúchalo en el minuto 13:24
Evitar campañas de testeo aisladas frente a probar directo en campañas CBO escaladas
Medición y experimentosTestear anuncios en campañas separadas con presupuesto bajo genera un alto costo de oportunidad y retrasos de semanas. Introducir nuevos creativos directamente en una campaña CBO escalada contra los anuncios ganadores permite comprobar de inmediato si el algoritmo les asigna presupuesto y si realmente tienen la capacidad de rendir y escalar.
great so it took you three weeks to identify a winner that could have been a banging on day three cost is massive here is that instead of testing it separately proven if you were to directly test that against the the big Winners right off of like like you just said three weeks stick this in another ad set in a CBO that's scaled up to 10 20K a day and if it does get the spend and it is converting I already know that it has the the legs to compete at that level
Escúchalo en el minuto 11:52
Canales de adquisiciónSi un anuncio nuevo no recibe presupuesto de manera orgánica dentro de una campaña abierta en Meta, no se debe forzar el gasto creando conjuntos de anuncios aislados. El algoritmo no le asigna impresiones porque no retiene la atención de los usuarios; forzar gasto genera competencia interna, encarece los CPMs y acelera la fatiga prematura de los creativos.
things to be true number one if it doesn't earn spend it's not good so if that ad doesn't get spent this is a liability to our business costs us money we're not going to let if we do the CPM is going to be really so if that ad doesn't get delivery it's a clear sign that for whatever reason do is not trust Facebook at all and in a month in two months you have an ad account with a bunch of isolated killers that are all dying off faster and faster and faster because you set up depreciation in all those assets speaking your cpms are going up too that all work in isolation competing against each…
Escúchalo en el minuto 15:07
Sistema de testing creativo «destronar al campeón»
Medición y experimentosPara gestionar la experimentación de anuncios y alinear al equipo creativo con el comprador de medios, se aprueba un concepto central, se producen cuatro iteraciones y se ponen a competir contra el mejor anuncio activo actual («el campeón»). La sistematización elimina el sesgo emocional y da claridad objetiva sobre el rendimiento.
um gamify it and so essentially you have your creatives and every week we'd run a test and to Charlie's point we'd throw it into the mix and you try and Dethrone essentially we would green light one core concept the creative agency would give us four iterations of that core concept and then we run all those systemization portrayed a feeling of fairness where to your point like oh my creatives were getting a shot did this beat the the champion creative and a lot and so for me that that systemization of getting in that Cadence of the creative of like here's our core concept you need to develop…
Escúchalo en el minuto 30:36
Riesgo de validar creatividades ganadoras con volumen insignificante de conversiones
Medición y experimentosEvaluar la efectividad de un anuncio con solo 3 a 10 ventas en una campaña de prueba de bajo presupuesto genera datos de baja integridad que no predicen el rendimiento al escalar a presupuestos mayores de miles de dólares.
mistaken opinion that you're testing your testing campaign is seeing what lottery ticket do you want to actually and your test doesn't actually start spend any meaningful money until it hits okay great your winner had what three got lucky the most is that anything at 50 bucks a day 500 a day that you want to spend five thousand for me that's a very low Integrity data set built on on non-actionable set of data that I'm then risking the stability
Escúchalo en el minuto 6:09
Investigación de mercado y clienteAdquirir tipos de clientes que tienen un CPA aparentemente más bajo pero un LTV deficiente introduce una enorme complejidad operativa (duplicación de inventario, flujos de email separados, landing pages distintas y mayor carga de soporte), multiplicando los costes postventa y mermando el beneficio neto.
customer has a lower CPA and the LTV is it cost me 20 cents to do it maybe the hair buyer I lose 20 cents of opportunity cost every time I buy that customer and it hurts my inventory because now I gotta stock two different offers my up I gotta now test two completely different email flows two completely different weight landing page tests I have to do with different types of customer service like the post purchase costs aren't double they're like Templar quadruple because of the complexity I'm making my business way more difficult so that it can make less money
Escúchalo en el minuto 45:25
Medición y experimentosEn lugar de juzgar los anuncios como 'buenos' o 'malos' de forma emocional, los compradores de medios deben operar bajo el método científico comprobando si una hipótesis fue validada o refutada, utilizando los anuncios no ganadores para extraer aprendizajes concretos que ajusten los próximos briefs creativos.
the best media buyers I know don't think of things in good or bad they think of things in the scientific method here's my thesis here's where I think it's going to end up and then you measure and here's where it ended up because if you totally amoral it's just is this the system I'm running and is this system producing the results I want okay this experiment did not that's great there's some learnings that I can integrate so now I have an even more tighter brief to then give my creative like never use shift your framing from successful or unsuccessful versus thesis verified or thesis…
Escúchalo en el minuto 32:57
Evitar la falacia del costo hundido al testear creatividades en paid media
Medición y experimentosForzar presupuesto publicitario para intentar validar o salvar un anuncio solo porque su producción fue costosa constituye una trampa de costo hundido; en plataformas como Facebook Ads, los creativos raramente remontan si no muestran tracción inicial en la subasta.
is good money behind bad money if they created an ad that is not performing do you really want to spend ten thousand dollars to validate that this ad does like there's a economic thingy to get nerdy right on the sun costs like that or marketing leaders is capital put good money behind bad money so that very rarely do you have this come from behind like if anything Facebook is not a come from behind story like you win or
Escúchalo en el minuto 25:40
Canales de adquisiciónLos anuncios con formatos simples y directos creados rápidamente en herramientas accesibles pueden superar el rendimiento de piezas audiovisuales de alto costo de producción, permitiendo reasignar el presupuesto ahorrado a mayor inversión en pauta publicitaria o creadores.
um we were building ads in canva that were destroying these high production ads and so it got to a place of like dude we don't need to spend like twenty thousand dollars to run this ad account on creatives let's spend it more on ads or let's throw events or to Charlie's point let's hire influencers and so I I
Escúchalo en el minuto 26:19
Medición y experimentosEnseñar al equipo creativo a interpretar los reportes de rendimiento y distribución de presupuesto publicitario crea un incentivo directo y colaborativo para diseñar creativos capaces de destronar al anuncio con mejor rendimiento.
in my opinion a good creative director will look at Dynamic creative as a tool to help them be better and better at their job you can really start to but I let very much I thought the creative director how to pull reports in and I literally was like these are the ads that are spending these ones that like this is what's working this is what isn't and then we would have conversations of like how do I get how can I work with you to Dethrone as as best interest to protect help you Dethrone the best one so what do we need relationship but that person is hella motivated to make it happen and to be
Escúchalo en el minuto 35:07
Medición y experimentosAislar variables para probar anuncios funciona en medios con CPM fijo e inventario constante, pero queda obsoleto en Facebook Ads porque la plataforma opera bajo un modelo de CPM optimizado (oCPM) orientado a la experiencia del usuario final.
say you're buying uh CPM basis for the way to go because the CPM is a constant cost and because you don't have you're just testing to see what gets you absolutely makes sense to isolate your the problem is at its core Facebook is it is an optimized CPM environment optimizing for the end user experience testing method for something that's obsolete that isn't built for how the code is written doesn't really make a
Escúchalo en el minuto 5:01
Canales de adquisiciónEl modelo de negocio de Meta se basa en retener la atención del usuario en la plataforma el mayor tiempo posible, por lo que premia anuncios que entretienen y prolongan la estancia. En cambio, Google y Amazon buscan resolver la intención de búsqueda rápidamente para que el usuario concrete la acción y salga del buscador.
what they is they retain attention for as long as possible Google and Amazon their business model is to get you off of Google and off of Amazon as quickly as possible so the business objective of how to be good at Google and how to test for that is how do I give somebody what they want so they leave Facebook is how we get somebody what they want so they'll stay
Escúchalo en el minuto 17:15
Medición y experimentosForzar presupuesto y entrega en anuncios dentro de entornos de prueba aislados genera fatiga publicitaria artificial en esos activos y perjudica el entorno de control general, resultando en un sistema que no escala más allá de las primeras etapas.
there so you're setting up a system that ultimately can't scale so even if you're is your other environment gets directly impacted by it and do I want one ad that works well that I know because I'm forcing delivery I'm going to see ad fatigue which is a choice and I'm creating ad fatigue on this asset while also depreciating my control environment what I don't see happen is you do that and then you have three ads that you're forcing in isolation your entire accounts just basically those three ads and it works forever and you'd ever have to test again like that model doesn't work past step two…
Escúchalo en el minuto 20:05
Sincronización estacional de captación según el ciclo de flujo de caja del cliente
Canales de adquisiciónAlinea la adquisición estacional con el tipo de retorno: aprovecha Q4 para adquirir clientes de alto LTV a largo plazo y Q1 para perfiles de compra rápida y flujo de caja inmediato, de modo que durante Q2 y Q3 los ingresos dependan del CRM y las recompras mientras se optimiza el modelo de negocio.
so now I need to understand what's my business objective so if it's quarter form I'm going after all the the long money I can like I'm going after the the older woman that's gonna spend more money low price you're accessing a higher LTB at a larger margin so in that case I'm gonna go up to long run in Q4 q1 I'm gonna warped by the short money so with for me the long money from Q4 hits and the short money from q1 hits and so now Q2 and Q3 is all CRM and I can work on my business model what can I do better as a business so that in Q3 and Q4 it's again I can bolt even higher I've done
Escúchalo en el minuto 43:20
Canales de adquisiciónLanzar demasiados tests creativos simultáneos atrae tipos de atención dispares que invalidan la optimización post-clic en la página de aterrizaje y aumentan los costos generales en la plataforma al quemar presupuesto en anuncios perdedores.
is because you're launching so many tests how many different types of attention are you bringing into your like any landing page test that you run as soon as you change the ad of the targeting basically begins becomes obsolete so you're making Every After The Click optimization effectively null majority of money developing losers that raises your cost of advertising across getting more expensive is that you're creating more harm than reducing your quality of the relationship the platform
Escúchalo en el minuto 7:14
Canales de adquisiciónLanzar demasiados tests creativos simultáneos atrae tipos de atención dispares que invalidan la optimización post-clic en la página de aterrizaje y aumentan los costos generales en la plataforma al quemar presupuesto en anuncios perdedores.
is because you're launching so many tests how many different types of attention are you bringing into your like any landing page test that you run as soon as you change the ad of the targeting basically begins becomes obsolete so you're making Every After The Click optimization effectively null majority of money developing losers that raises your cost of advertising across getting more expensive is that you're creating more harm than reducing your quality of the relationship the platform
Escúchalo en el minuto 7:14
Canales de adquisiciónProbar creatividades en una campaña aislada con bajo presupuesto para luego trasladar los anuncios ganadores a una campaña de escalado suele fallar en Meta Ads, ya que los anuncios que funcionan en entornos controlados de bajo gasto a menudo no logran competir con los anuncios consolidados a presupuestos elevados.
you have a testing campaign and a scaling campaign right so the theory week you set up a test campaign for seven days hey you got a winner cool let's toss it into this to the scaling account structure right so say for example you know you're testing four or five creatives in this in its own environment at a low budget you move it into this big campaign and it's and it's this like the big boys can this new ad compete with the big boys or does it completely mess up the entire structure
Escúchalo en el minuto 2:45
Canales de adquisiciónProbar creatividades en una campaña aislada con bajo presupuesto para luego trasladar los anuncios ganadores a una campaña de escalado suele fallar en Meta Ads, ya que los anuncios que funcionan en entornos controlados de bajo gasto a menudo no logran competir con los anuncios consolidados a presupuestos elevados.
you have a testing campaign and a scaling campaign right so the theory week you set up a test campaign for seven days hey you got a winner cool let's toss it into this to the scaling account structure right so say for example you know you're testing four or five creatives in this in its own environment at a low budget you move it into this big campaign and it's and it's this like the big boys can this new ad compete with the big boys or does it completely mess up the entire structure
Escúchalo en el minuto 2:45
Canales de adquisiciónEn la marca Underoutfit, un anuncio básico en formato slideshow montado directamente dentro de Meta con 5 o 6 fotos de una sesión fotográfica logró absorber casi 2 millones de dólares en inversión publicitaria escalable.
like I mean it under outfit we had an ad that was quite literally from a photo shoot I built it in the slideshow I think Felix built it in a slide show in Facebook's ad unit like just the here's an ad here's five six images put it in a little video that one ad spent like nearly two million dollars
Escúchalo en el minuto 24:42
Independencia respecto a las plataformas publicitarias
Oferta y precioSi un negocio necesita que Facebook Ads funcione perfectamente para sobrevivir, significa que su modelo de negocio subyacente es deficiente; el valor real y la rentabilidad deben residir en el modelo y no en la optimización del canal de adquisición frontal.
the front end of the business because it means they're spending the most of their time on what actually makes money and be fair if you need Facebook to work to be in business you don't have a good business
Escúchalo en el minuto 47:46
Retención y relación con el clienteEnfocar el esfuerzo en mejorar la segunda compra, el CRM y el modelo de negocio permitió escalar Underoutfit de 50.000 dólares al mes a 1 millón semanal, y 310 Nutrition de 18 a 95 millones de facturación en menos de dos años.
want short money Q4 is Long Money Q2 Q3 I'm letting the CRM come in I'm gonna improve the business model and if I can continue that flywheel the last half exponential growth I mean under Alpha went from 50k a month to a million a week 310 nutritional from 18 million to 95. and we're talking like less than two years to make that happen
Escúchalo en el minuto 47:09
Tasa histórica de éxito al intentar superar un anuncio campeón
Medición y experimentosEn un programa de experimentación publicitaria continua a lo largo de casi dos años, solo 4 de aproximadamente 50 a 60 pruebas creativas lograron superar el rendimiento del anuncio campeón existente.
dude we I was there for almost two years and I think like there was four times four times where it was dethrough on the champion out of I can't like 50 tests maybe 60 tests so it's what I I guess
Escúchalo en el minuto 32:38
Seguir leyendo
Esta es una de las 1872 entrevistas despiezadas. Puedes
ver todas, leer las ideas por tema o
empezar por la ruta guiada si no sabes por dónde.