Marketero

Marketero · Entrevistas

Low Facebook Ad ROAS after Andromeda? Watch this.

27 ideas de marketing sacadas de «Low Facebook Ad ROAS after Andromeda? Watch this.», de Professor Charley T. Sobre todo de canales de adquisición y…

Ver la entrevista entera en YouTube

De qué habla

Las 27 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 27 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.

Saturar un conjunto de anuncios con exceso de creativos diluye los datos y empeora el rendimiento

Canales de adquisición

Tener demasiados anuncios compitiendo en una misma campaña fragmenta el presupuesto y reduce el volumen de datos por creativo, lo que disminuye la confianza del algoritmo de Meta para asignar impresiones a usuarios listos para comprar y deteriora el ROAS.

Although I will give him that if you're still running hundreds of ads, It's because it's punishing people who spam the machine. are going to the ads where the machine has the highest confidence that that person's going to convert on it. But inherently, the more ads you have, the less spend any ad has, the less data any ad has, the less confident the machine is in serving your ad to somebody who's ready to buy. This entire account is essentially spending on three ads. But the rest, $0.02, $0.90, $2.75. They're confusing the machine, attracting low quality attention, making it harder to produce…

Escúchalo en el minuto 15:49

Estructura de test creativo 3:2:2 con formatos homogéneos y copies constantes

Medición y experimentos

En lugar de colocar 10 creatividades variadas dentro de un solo anuncio flexible (donde una combinación aleatoria suele acaparar entre el 75% y 90% del presupuesto desde el inicio), divide el test en anuncios 3:2:2 (3 creativos, 2 textos principales, 2 titulares) manteniendo el mismo formato (solo imágenes o solo vídeos) y usando el mismo copy entre pruebas para aislar científicamente qué elemento genera el impacto.

Instead of putting 10 in one, maybe make three 322 ads. Remember, that's three creative, two primary text, two headlines. Right now, what you're getting out of 10, first off, is random distribution in that 10. And whatever gets a good result early will dominate all of the spend. If you do the breakdown, you'll notice that 90, 80, 75% of the spend at a minimum is going to go to essentially one permutation. And that is exponentially more valuable, especially if you use the same copy in all three of those 322 tests. And that way you can see does the copy actually move the needle? If the same…

Escúchalo en el minuto 10:20

Estructura de campaña única y conjuntos de prueba en Meta Ads

Canales de adquisición

La estructura óptima de CBO en Meta Ads consiste en una campaña por objetivo de negocio con un conjunto de anuncios principal reservado exclusivamente para creativos ganadores demostrados, complementado por conjuntos adicionales de testeo diseñados para superar y reemplazar al peor anuncio del grupo titular evaluando el impacto en el rendimiento neto global.

First, it's one campaign per business objective. Also, your one ad set that they show here should only be made of your proven unicorn ads. And then you're going to have additional ad sets on top of that, specifically trying to let the CBO deliver spend someplace else to see if it improves the overall blended output Now, the reason that's important is because you should have one ad set with your proven The purpose of the other ad sets is to test to improve the weakest ad in your winner's ad set. Ultimately, what you're trying to do is put together an all-star team, and then you're testing to…

Escúchalo en el minuto 30:55

Saturar un conjunto de anuncios con exceso de creativos diluye los datos y empeora el rendimiento

Canales de adquisición

Tener demasiados anuncios compitiendo en una misma campaña fragmenta el presupuesto y reduce el volumen de datos por creativo, lo que disminuye la confianza del algoritmo de Meta para asignar impresiones a usuarios listos para comprar y deteriora el ROAS.

Although I will give him that if you're still running hundreds of ads, It's because it's punishing people who spam the machine. are going to the ads where the machine has the highest confidence that that person's going to convert on it. But inherently, the more ads you have, the less spend any ad has, the less data any ad has, the less confident the machine is in serving your ad to somebody who's ready to buy. This entire account is essentially spending on three ads. But the rest, $0.02, $0.90, $2.75. They're confusing the machine, attracting low quality attention, making it harder to produce…

Escúchalo en el minuto 15:49

Saturar un conjunto de anuncios con exceso de creativos diluye los datos y empeora el rendimiento

Canales de adquisición

Tener demasiados anuncios compitiendo en una misma campaña fragmenta el presupuesto y reduce el volumen de datos por creativo, lo que disminuye la confianza del algoritmo de Meta para asignar impresiones a usuarios listos para comprar y deteriora el ROAS.

Although I will give him that if you're still running hundreds of ads, It's because it's punishing people who spam the machine. are going to the ads where the machine has the highest confidence that that person's going to convert on it. But inherently, the more ads you have, the less spend any ad has, the less data any ad has, the less confident the machine is in serving your ad to somebody who's ready to buy. This entire account is essentially spending on three ads. But the rest, $0.02, $0.90, $2.75. They're confusing the machine, attracting low quality attention, making it harder to produce…

Escúchalo en el minuto 15:49

Saturar un conjunto de anuncios con exceso de creativos diluye los datos y empeora el rendimiento

Canales de adquisición

Tener demasiados anuncios compitiendo en una misma campaña fragmenta el presupuesto y reduce el volumen de datos por creativo, lo que disminuye la confianza del algoritmo de Meta para asignar impresiones a usuarios listos para comprar y deteriora el ROAS.

Although I will give him that if you're still running hundreds of ads, It's because it's punishing people who spam the machine. are going to the ads where the machine has the highest confidence that that person's going to convert on it. But inherently, the more ads you have, the less spend any ad has, the less data any ad has, the less confident the machine is in serving your ad to somebody who's ready to buy. This entire account is essentially spending on three ads. But the rest, $0.02, $0.90, $2.75. They're confusing the machine, attracting low quality attention, making it harder to produce…

Escúchalo en el minuto 15:49

Escalar atrayendo al mismo cliente ideal en lugar de dispersar creativos

Canales de adquisición

El crecimiento escalable en publicidad pagada se logra atrayendo en volumen al mismo perfil de cliente ideal mediante mensajes consistentes; lanzar cientos de creativos únicos y dispares fragmenta la audiencia y anula la efectividad de las etapas posteriores al clic (como CRO, email marketing y búsqueda).

And if all of them more or less reach the same two or three ideal customers, then instead of trying to get 1,000 people to buy once, you can get the same kind of person and reach thousands or 10,000 or hundreds of thousands of them. And that's how you scale a business. It's not by trying to get as many different people through the front door as possible. It's getting as many of the same person you can and then taking advantage of that. Because this creatives at scale, 100% unique creatives idea, essentially means that everything that happens after the click becomes compromised. If somebody…

Escúchalo en el minuto 5:03

Priorizar pocos creativos ganadores consolidados sobre el volumen masivo de anuncios

Canales de adquisición

Los sistemas modernos de entrega de anuncios optimizan mejor cuando disponen de pocos creativos cohesionados que acumulan datos de rendimiento a lo largo del tiempo, en lugar de saturar la cuenta con decenas de anuncios aleatorios donde se pierde la trazabilidad del impacto real en el negocio.

As long as your ads run for a while and have data and then also reinforce other ads that you have. As long as you have a cohesive customer journey through your ads, the machine is going to get better and better at showing them to the right person. It's not about taking a lot more dumb shots in the dark. It's about having a few really well-educated killers do all of the work for you. And the machine has never been better at serving that small group of killers to the absolute best people. And because so many of these ads look and feel the same, there's no way to say that one of them is the…

Escúchalo en el minuto 7:00

Priorizar el volumen total de adquisición sobre un ROAS aislado de último toque

Medición y experimentos

Para escalar un negocio es más rentable aceptar un ROAS publicitario menor (por ejemplo, 2x en lugar de 4x) a cambio de multiplicar el volumen de nuevos usuarios alcanzados, ya que este tráfico nutre todo el embudo e incrementa las conversiones en canales como email y búsqueda.

And scaling your business has nothing to do with trying to have your Facebook ad get credit for the last touch. It has everything to do with having your marketing funnel have way more opportunity. Please, please give me a 2x ROAS on five times more volume than a 4x ROAS. If I have a lower return on ad spend on my Facebook ads, and it means that I'm reaching 3, 4, 5 times as many people, and my search and my email go way up, and I'm reaching way more new customers, I'm going to make way, way more money.

Escúchalo en el minuto 16:39

Medición de lift incremental creativo mediante campaña única

Medición y experimentos

Para evaluar si una creatividad genera impacto financiero real sin sobrecargar el aprendizaje algorítmico, implementa una estructura de campaña única con un conjunto de control probado y uno o dos conjuntos de prueba con un solo anuncio 3:2:2; el test se valida únicamente si el nuevo conjunto gana gasto orgánicamente y mejora el rendimiento total de la campaña.

But if you do this properly in a one campaign setup, you're going to have your control ad set where you know how those ads perform, and then an additional one or two ad sets with a single 3-2-2 ad in it. And then when that ad set, if it earns spend, improves the overall campaign performance, you'll be able to measure that impact on your bank account, which means you are actively testing with data to see if you're creating incremental lift or not.

Escúchalo en el minuto 14:24

Menor cantidad de anuncios genera mayor consistencia al concentrar datos

Canales de adquisición

Reducir el número de anuncios activos estabiliza el rendimiento de la cuenta publicitaria, ya que cada anuncio acumula más datos y permite al algoritmo optimizar de forma más predecible.

The fewer ads you have, the more consistent that ad performance will be because each ad will have more data.

Escúchalo en el minuto 41:07

Lanzar anuncios masivos crea fatiga artificial y canibaliza creativos probados

Canales de adquisición

Lanzar un volumen excesivo de anuncios nuevos satura el fondo del embudo porque el algoritmo tiende a mostrarlos a la audiencia existente, lo que roba ventas a las creatividades que ya funcionaban y crea una fatiga publicitaria artificial.

So the more ads you launch, the more money you spend blowing up the bottom of the funnel, But if you're constantly launching new ads, you'll never be able to reach those best customers over and over again, because you're resetting the deck, showing all of your ads, all of your new ads to the people at the bottom of the funnel, and then complaining about growth. Every time you launch those new ads and they get a couple of sales, what you're doing is you're stealing sales from proven ads that could already be successful. So every time you're launching new things, understand that what you're…

Escúchalo en el minuto 25:16

Lanzar anuncios masivos crea fatiga artificial y canibaliza creativos probados

Canales de adquisición

Lanzar un volumen excesivo de anuncios nuevos satura el fondo del embudo porque el algoritmo tiende a mostrarlos a la audiencia existente, lo que roba ventas a las creatividades que ya funcionaban y crea una fatiga publicitaria artificial.

So the more ads you launch, the more money you spend blowing up the bottom of the funnel, But if you're constantly launching new ads, you'll never be able to reach those best customers over and over again, because you're resetting the deck, showing all of your ads, all of your new ads to the people at the bottom of the funnel, and then complaining about growth. Every time you launch those new ads and they get a couple of sales, what you're doing is you're stealing sales from proven ads that could already be successful. So every time you're launching new things, understand that what you're…

Escúchalo en el minuto 25:16

Priorizar la diversidad creativa estratégica sobre el volumen masivo de anuncios en Meta

Canales de adquisición

Saturar una cuenta de Meta con cientos de anuncios o variaciones mínimas de ganchos entorpece el algoritmo de entrega. La verdadera diversidad creativa consiste en utilizar un número reducido de enfoques y formatos distintos (por ejemplo, testimonios UGC frente a imágenes estáticas de beneficios) dirigidos a ángulos concretos de cliente con un mensaje consistente.

And in fact, having more ads isn't necessarily a good thing. I would argue and have repeatedly with data shown that having a lot of ads actually makes the entire machine dumber. You need to nail creative diversity, which means you hit a couple different types of customers in a couple different types of ways. And with that in mind, remember that diversity is not a thousand ads. It's having a few targeted ways of speaking to a few people differently.

Escúchalo en el minuto 3:21

Apagar anuncios sobreindexados en clientes existentes para forzar alcance incremental

Canales de adquisición

Si el desglose de entrega muestra que un anuncio se distribuye desproporcionadamente a clientes existentes en lugar de captar nuevos usuarios, apagar esa creatividad específica eleva de inmediato el alcance incremental hacia tráfico frío.

And if the machine in their best interest has decided to show these ads to people that I don't want them showing this ad to, you could just turn that ad off. And what you will find is your incremental reach will immediately go up. If there are ads that overly deliver to that audience, turn that ad off. And odds are, the worse the ad is, the more often it only gets shown to existing customers, because the machine says, we never want to show this to anybody else.

Escúchalo en el minuto 25:36

Fragmentar en múltiples campañas canibaliza resultados y distorsiona la atribución

Canales de adquisición

Dividir la inversión entre múltiples campañas aisladas (como retargeting, intereses y creatividades por separado) hace que compitan entre sí por el fondo del embudo, canibalizando resultados y aumentando el desgaste de anuncios sin aportar crecimiento incremental real.

The customer journey is happening amongst four different campaigns that don't talk to each other, that are all cannibalizing each other's results with a bunch of different ads that are competing to try to get any data. but we do have a lot of competition for the bottom of the funnel because remember all of the ads are being launched at the bottom of the funnel. So there's more competition to steal the easy wins, which is why the attribution on the one campaign looks worse. And the ad fatigue here is exponential from what you would see in a properly set up one campaign.

Escúchalo en el minuto 43:00

Evitar la canibalización de clientes existentes en la estructura de una sola campaña

Canales de adquisición

Para evitar que una estructura de una sola campaña (one campaign) gaste presupuesto en clientes que ya compraron, se debe configurar el evento de conversión enfocado específicamente en nuevos clientes o desactivar manualmente los anuncios que se muestren desproporcionadamente a compradores actuales.

When you use a one campaign structure, you give Facebook the freedom to spend your budget however they choose. Which is a good thing. So a ton of your budget can be going towards your existing customers, First off, if you don't want your existing customers to see your ads, have new customer as the conversion event. Second, if you notice that some of your ads are being delivered to existing customers in a way that makes you feel uncomfortable, turn off that ad.

Escúchalo en el minuto 24:03

Estructura de conceptos creativos mediante la analogía de los anillos olímpicos y anuncios 3:2:2

Canales de adquisición

Una estructura simplificada y eficaz en Meta combina pocos conceptos creativos usando anuncios flexibles (metodología 3:2:2): por ejemplo, tres conceptos en video orientados a captación inicial (upper funnel) y dos conceptos estáticos diseñados como segundo impacto informativo para la parte baja del embudo.

Perhaps you have a UGC testimonial and a features and benefits static. And if you're using flexible ads in the 3-2-2 method, you can easily have a few permutations If you have three concepts that sit upper funnel, and that could simply be three flexible ads, the 3-2-2 setup, and then two ads that sit lower funnel. Again, flex ads in a 3-2-2 setup, which could easily be three videos, three statics. And understand that the purpose of the statics is to be the second touch to somebody that's

Escúchalo en el minuto 4:19

Evitar saturar los anuncios flexibles con demasiadas variantes simultáneas

Canales de adquisición

Rellenar al máximo las opciones de creativos dinámicos o flexibles (como meter 10 videos en un solo anuncio) introduce una aleatoriedad incontrolable que impide saber qué elemento genera los resultados; es preferible distribuir pocas variantes estructuradas entre anuncios separados para mantener visibilidad y control.

But please don't max out all of the options. Facebook isn't Google. Instead of having six videos in one ad, you could have two ads that each have three and then understand which one does better. This is stacking randomization inside of a black box with a ton of competition. That's not how you build a consistent, complementary, and scalable environment. This produces short wins, that fatigue that makes you dependent on the person running the ads more and more.

Escúchalo en el minuto 9:21

Meta valida los anuncios nuevos mostrándolos primero al fondo del embudo

Canales de adquisición

Al lanzar nuevos anuncios creativos, los algoritmos de plataformas como Meta suelen servirlos primero a personas conocidas o clientes existentes (bottom of funnel) para evaluar su tasa de respuesta antes de decidir si los escalan hacia audiencias frías.

And in fact, one of the reasons that so many people struggle with the one campaign method when they launch so many ads is because every new ad you launch is pushed to the bottom of the funnel. It's showing that ad to somebody that it knows how they're going to behave. Just like with an ad, if the people who are already existing customers don't like it, it's not going to get served to anybody else.

Escúchalo en el minuto 24:41

Segmentar prospección y retargeting en campañas separadas resta eficiencia al algoritmo

Canales de adquisición

Aislar rígidamente la prospección del retargeting en campañas independientes obliga al anunciante a predecir la disponibilidad diaria de inventario y la intención de compra, eliminando la capacidad del algoritmo de Meta para decidir en tiempo real a quién y cuándo mostrar el anuncio más adecuado.

One of the biggest issues, though, when you have them in separate campaigns is that forces you to have to know what's going to happen tomorrow around the available inventory. How many people are in that audience? How eager are they to buy? Where should the machine show the right ad to the right person? The more control you take away from the machine,

Escúchalo en el minuto 40:04

Impactar a clientes existentes entrena al algoritmo para captar usuarios de mayor LTV

Medición y experimentos

Permitir que la pauta publicitaria impacte a clientes existentes aporta datos de compradores recurrentes al algoritmo de aprendizaje automático, lo que le enseña a identificar y atraer a nuevos prospectos con mayor propensión a generar un valor de vida (LTV) alto.

And to be fair, if the machine showing ads to existing customers is the reason your LTV And at the end of the day, all you're really trying to do is have your blended CPA be below your blended AOV. And if you attract the same type of customer over and over again, and you improve the LTV on that customer, it's almost impossible to not scale your business. And more importantly, and I'll end on this, the existing customers that come back and buy again and again, that's data for the machine to know who the best new customers are. If the machine is a machine learning, if we know Andromeda is a…

Escúchalo en el minuto 26:15

Optimizar la publicidad para crear nuevos clientes de alto valor en lugar de inflar la atribución de ROAS

Medición y experimentos

El éxito publicitario no debe medirse por cuántas compras se atribuye la plataforma (ROAS reportado por saturación de anuncios), sino por la capacidad de la pauta para generar intencionalmente nuevos recorridos de compra que atraigan clientes recurrentes y de alto valor acumulado.

I would say what's most important is understanding how to find the customers who are most likely to spend the most money and come back and buy again and again, and then find me as many of those as you can. The problem is that he's using Andromeda's greater level of intelligence to justify spamming the machine instead of saying because of that intelligence, we can take far more educated shots. And understand exactly what our problems are so that we can manufacture the customer journey that most often brings the highest value customer who's likely to buy again and again. where his definition of…

Escúchalo en el minuto 20:14

La caída de ROAS al escalar responde a dinámicas de mercado y prospección, no a manipulación del algoritmo

Canales de adquisición

Meta no entrega primero las conversiones fáciles para luego agotar la audiencia intencionalmente; las caídas de ROAS al incrementar presupuesto suelen explicarse por estacionalidad en la intención del consumidor, variaciones de CPM o porque el algoritmo expande el alcance hacia audiencias frías que generan valor incremental.

Meta isn't front-loading the results. The consumer behavior changes over time. There's no, like, pool of customers and Facebook's giving you all the easy ones right away and then holding off on the other ones. That just doesn't happen. Consumer intent and market forces change. And also, maybe the reason the results look worse is because the machine is now going after new audiences. It's creating more incremental lift for the rest of the business.

Escúchalo en el minuto 36:00

El algoritmo optimiza la secuencia del viaje del usuario, no un único anuncio ganador

Canales de adquisición

Buscar un único 'anuncio ganador' es un error metodológico; los sistemas de entrega modernos de Meta (Andromeda) orquestan secuencias de diferentes formatos (donde las imágenes suelen ser más eficientes a menor inversión y el vídeo opera más arriba en el embudo con mayor escala) para guiar al usuario hacia la conversión a lo largo de su recorrido.

You're not trying to see what is a winning ad. The idea of the best ad is a complete misnomer. What you're trying to do when you test a video and an image is say which one does better. And ultimately, the image is probably going to be more efficient on lower spend. And the video is probably going to be less efficient at greater spend. It's not about the winning ad. What Andromeda is doing is showing the right sequence of ads to produce the most consistent and end-user positive experience customer journey. So it's not show this video to the right people because it's going to get them to…

Escúchalo en el minuto 12:15

Transición algorítmica del gasto en Meta de clientes existentes a nuevos prospectos

Canales de adquisición

Al iniciar una campaña amplia en Meta Ads, el algoritmo suele concentrar parte del presupuesto inicial en clientes existentes para asegurar conversiones rápidas y aprender el perfil del comprador, para luego reducir progresivamente esa asignación y expandir el gasto hacia audiencias netamente nuevas.

But odds are those existing customers came early, and then the machine understood who to show ads to, and then it went to engage and then out to new overwhelmingly. And in a lot of cases, Meta will start to push all of that towards existing customers because they know low-hanging fruit is there. it's going to over-index towards that audience. It's already been cut from 33% to 25 max. It's already throttling that existing customer and moving more towards new.

Escúchalo en el minuto 33:01

Transición algorítmica del gasto en Meta de clientes existentes a nuevos prospectos

Canales de adquisición

Al iniciar una campaña amplia en Meta Ads, el algoritmo suele concentrar parte del presupuesto inicial en clientes existentes para asegurar conversiones rápidas y aprender el perfil del comprador, para luego reducir progresivamente esa asignación y expandir el gasto hacia audiencias netamente nuevas.

But odds are those existing customers came early, and then the machine understood who to show ads to, and then it went to engage and then out to new overwhelmingly. And in a lot of cases, Meta will start to push all of that towards existing customers because they know low-hanging fruit is there. it's going to over-index towards that audience. It's already been cut from 33% to 25 max. It's already throttling that existing customer and moving more towards new.

Escúchalo en el minuto 33:01

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.