Marketero · Entrevistas
The Muddled Attribution Puzzle Solved: Scott Desgrosseilliers’ 5 Forces
20 ideas de marketing sacadas de «The Muddled Attribution Puzzle Solved: Scott Desgrosseilliers’ 5 Forces», de Perpetual Traffic. Sobre todo de medición y…
- 20 ideas despiezadas
- Fuente Perpetual Traffic
- Dura 53 min
- Publicada en 2025
Ver la entrevista entera en YouTube
De qué habla
Las 20 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 20 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.
Marco de zonas Scale, Chill y Kill para la gestión de presupuesto publicitario
Canales de adquisiciónEn lugar de evaluar las campañas de anuncios contra un número único y rígido de KPI (como el CAC), se deben definir tres rangos operativos previamente acordados: zona de escalado (aumentar presupuesto automáticamente cuando el rendimiento supera el umbral), zona 'chill' o de mantenimiento (mantener el gasto estable sin intervenir mientras esté en el rango esperado) y zona de reducción o 'kill' (recortar presupuesto…
set what's called zone scale, chill, kill zones. Let me go here first scale, increase the budget. If that KPI, if you're in the scale zone, kill is decrease, chill, chills underrated. It means you keep the, you keep spending the same amount of money. If the KPI is in that zone. Also for you as a busy marketer, it means you can chill out and move on to the next thing. Cause you know, this one's going fine. You may want to optimize it and make it go better, but you know, it's good an end CAC is a number when in fact it should be a range really. And if you're shooting for a $50 NCC and you're…
Escúchalo en el minuto 16:46
Uso de zonas de decisión («Scale, Chill, Kill») en lugar de metas numéricas fijas
Medición y experimentosEn lugar de fijar un objetivo numérico único para evaluar una campaña (como un CAC exacto de $50), se deben definir de antemano tres zonas o rangos de decisión: escalar (scale), pausar/apagar (kill) o mantener sin cambios (chill). Esto elimina la parálisis por análisis y las reacciones emocionales ante fluctuaciones menores, permitiendo tomar decisiones de optimización metódicas según el rango en el que caiga el…
Scale. So the idea with scale, chill, kill is that if you just set a single number as a goal, like I want to acquire new customers at $50, then whenever you take a measurement or an outcome, it's always either above or below that number. And then you have to do some thinking. Oh, it's above. Oh, wait, how long has it been above that number? Now is that bad? Oh, do I have to go dig into the campaign now? Or do I have to go look at my who I'm targeting? Do I have to go look at the creatives? Because now it's a couple dollars above $50. Oh, no, wait, I think I can go to 60 until it's a problem…
Escúchalo en el minuto 10:50
Alinear la intención y las expectativas de campaña antes de medir la atribución
Medición y experimentosPara no perderse en la complejidad de las herramientas de atribución, antes de lanzar cualquier campaña deben definirse dos elementos: la intención específica (que dicta qué modelo de atribución, métricas y tendencias observar) y las expectativas temporales (cuánto tiempo debe dejarse correr la campaña y qué criterios exactos determinarán su éxito antes de juzgar ventas prematuramente).
was these five forces. First is the intention. Every campaign should have an intention. What are you trying to do? Then you should have an X that helps us as for measurement and for optimization, narrow down all the different things you can focus on, which is important because we're all busy marketers, but also it helps the measurement know which data points to include, which trends matter, which attribution model to set. And that's important. And so that alone it doesn't look like Facebook's making a lot of money. Well, with the expectations, you set up how long you're going to wait based on…
Escúchalo en el minuto 8:42
Marco de diagnóstico en cinco áreas ante métricas de adquisición fuera de rango
Medición y experimentosCuando el coste de adquisición de nuevos clientes (nCAC) o el ROAS cae en zona crítica, se debe diagnosticar el problema revisando sistemáticamente cinco áreas antes de modificar creativos a ciegas: tráfico (si se atrae público nuevo vs. recurrente), asignación de presupuesto, creativos y oferta, adecuación del producto y calidad del cliente (impacto en AOV y aceptación de upsells).
optimization. There's three areas that we're looking at, excuse me, five areas, traffic, budget allocation, creative and offer product and customer quality, all the different metrics in those areas. We have different metrics and things we look at in these areas to either tell you, Hey, you got a traffic problem or it's the product you're offering because it's new customers and you're offering a product that they don't like, or it could be creative and offer, which could be a lot of things or the customer quality. It might be, well, yeah, you're bringing in volume, but the AOV is way down…
Escúchalo en el minuto 37:42
Concentrar campañas de tráfico frío en el SKU con mejor conversión histórica
Canales de adquisiciónAl optimizar campañas automatizadas de adquisición (como Advantage+), conviene analizar qué SKU ha convertido históricamente mejor al tráfico frío y construir la campaña, los ganchos y los creativos exclusivamente alrededor de ese producto ganador en lugar de dejar la elección al algoritmo.
If you're trying to optimize to go in and look at after segmenting new versus repeat, what SKUs have worked in the past on new on turning people from cold traffic to new is pretty informative because it could be that they're just not getting shown the product that in your history has converted them best. It's been, if you throw it up to the machine. And so then from an advantage plus world focus, you just build the campaign around that one product. Cause then it's easier. You just do the creative for that one product, the hooks around that one and focus on all cold traffic there because…
Escúchalo en el minuto 45:35
Llevar un registro de decisiones con fecha de remedición ligada al ciclo de ventas para evitar cambios impulsivos en anuncios
Medición y experimentosCada vez que se tome una decisión sobre campañas de anuncios (como duplicar o reducir presupuesto), se debe registrar la acción y fijar una fecha de remedición obligatoria equivalente a la duración del ciclo de ventas (por ejemplo, 14 días). Cambiar el presupuesto reinicia la ventana de medición, por lo que evaluar los resultados a los pocos días genera reacciones impulsivas erróneas; el registro protege la…
Now with action, the key piece is what's called the decision log. I'll show one already filled out because the idea is that you want to keep track of what you've decided to do or not to do based on when you make a decision, because also going back to those times, if you've made a decision and cut the budget in half or doubled the budget that resets the measurement time. Now it's not always going to go well in with people, but if you've suddenly doubled your budget because you were in the scale zone the very next day, if you have like a two week sales cycle, the very next day, the numbers…
Escúchalo en el minuto 31:48
Medición y experimentosEl período y la ventana de atribución deben fijarse de antemano según la duración del ciclo comercial y la naturaleza del canal. Aceptar ventanas infladas por defecto (como 28 días para SMS, cuando debería medirse entre 2 horas y 2 días) distorsiona los datos, sobreasigna mérito a canales de impacto inmediato y subestima el aporte del marketing relacional o email.
So then also the intention sets the measurement period. And this is important that when you're launching these campaigns, you communicate the measurement period right up front based on what they're trying to accomplish. we have agreements. If we're not looking at a specific measurement period, that's fair. Like Facebook has that 28 day attribution window. Attentive. Look, your default is 28 days for an SMS. That should be two hours, two days. But if you don't have the type of, I argue two to four hours, I'll give you two days max for SMS attribution, but attentive default is 28. And the only…
Escúchalo en el minuto 27:25
Aislar en campañas independientes los anuncios o palabras clave rentables ignorados por el algoritmo
Canales de adquisiciónCuando los algoritmos de optimización de plataformas publicitarias concentran todo el presupuesto en un solo anuncio o término e ignoran variantes rentables o con cuota de impresiones disponible, se deben aislar esos elementos en campañas separadas para forzar la asignación de presupuesto y permitir su escalado.
may not be AI, maybe, you know, just machine learning. So in that case, that's a budget issue of, Hey, this ad didn't really get any budget. It might actually still work. You just need to try it on a different campaign search campaigns. A lot of times we see that, you know, profitable keywords aren't getting enough budget, even though they have more impression share available. you give stuff up to the machine, the data can point out where the machines made a mistake because they're not foolproof. And then there's different ways to deal with that. You might just have to break out the things…
Escúchalo en el minuto 40:38
Usar atribución de primer clic para evaluar campañas de tráfico frío
Medición y experimentosLas campañas de prospección dirigidas a audiencias frías deben medirse con un modelo de atribución de primer clic (first-click). Evaluar tráfico frío con los modelos predeterminados de plataformas como Google Analytics 4 o Google Ads (que priorizan clics recientes) penaliza injustamente la adquisición al ignorar qué canal inició el contacto con el cliente.
or agreement around the measurement source of truth, what attribution model to use the measurement expectation phase. And we do that by mapping, first of all, the specific intentions to the specific model. So we don't need to go into all of these, but a common one called traffic prospecting, you should use first click because you're trying to find cold traffic. So you need to see what the first click was from the cold traffic. Now, if you're in looking at Google Analytics 4, or if you're in looking at, let's say, Google ads manager reports, they don't use first click. So they're not going to…
Escúchalo en el minuto 25:23
Alinear los umbrales de escalado publicitario con finanzas y capacidad operativa
Medición y experimentosAntes de escalar campañas de adquisición, marketing debe definir zonas claras de decisión (escalar, mantener o pausar según el CAC de nuevos clientes) acordadas con finanzas y validadas con operaciones, para no superar la capacidad de inventario y cumplimiento.
and you're listening to this and you don't have these zones, like these have to come from finance. you know, for our NCAC. So what are all those scenarios? And what do I, you know, where do I tamp the brakes? Where do I press the gas? And understanding that from a profitability standpoint, risk tolerance of the business owner too. I would say yes to finance, but I'd also encourage you to latest. And so in that, like so many companies plan what would happen if they fail, but they really plan what they do if they scale. So if it's like, you need to know what the profitable numbers are and you…
Escúchalo en el minuto 20:39
No mezclar el valor medio de pedido (AOV) de clientes nuevos con el de recurrentes al fijar objetivos de captación
Medición y experimentosEl valor medio de pedido suele ser notablemente superior en clientes recurrentes que en nuevos; calcular objetivos de adquisición basados en un AOV global e histórico infla las expectativas de margen por cliente nuevo y distorsiona el cálculo del coste de adquisición asumible.
And then the customer quality AOV verse what your benchmarks are, particularly again with new versus historical new. And then cause repeat AOVs and obviously repeat LTV is higher because they're repeat customer. But AOV I find across the board is usually higher from a repeat customer. So, but if you're not doing that in your benchmarking and you're thinking, oh, my AOV is normally 200 bucks. So I just need to convert people at a hundred and I'm going to make a hundred bucks. And then lo and behold, you're only making 140 as your AOV. Well, it could be, then you go back. Oh shoot. I was…
Escúchalo en el minuto 43:15
Usar septiembre como prueba de oferta y captación previa a Black Friday
Canales de adquisiciónLas fechas comerciales de septiembre deben tratarse como un ensayo de Black Friday para validar la oferta y construir una base de nuevos clientes/audiencias a menor costo, evitando destinar presupuesto masivo a adquisición fría en octubre y noviembre cuando la pauta publicitaria se encarece.
You know that like September with Labor Day is your mini Black Friday. So you have to make sure you have everything in place to do a soft launch of what your offer is going to be. And that you build all your new customers to retarget. Because if you're spending money on new in October and November, boy, you a sucker.
Escúchalo en el minuto 49:33
Fijar una única métrica North Star por campaña para evitar la parálisis por datos
Medición y experimentosCada campaña publicitaria debe lanzarse con una única métrica rectora (North Star KPI) alineada con su intención principal. Aunque existan indicadores secundarios valiosos (como ROAS, CAC, rebote o AOV), evaluar demasiadas variables simultáneamente distorsiona los criterios de éxito y hace perder el foco de optimización.
a campaign, these are the three things we're first going to fill out, which is what's the intention of the campaign? What's the best KPI for that intention? It doesn't mean we won't look at multiple indicators, but you need a North star because otherwise you get in there and you're like, oh, the ROAS is here. The CAC is here. Oh, the N-CAC is here. Oh, the bounce rate is up on the people clicking to the site. Oh, the AOV is down or up. How many people are taking the upsell? There's so many ways you can just waste time. And so you got to just stick to the North star and then you look at the…
Escúchalo en el minuto 14:01
Canales de adquisiciónUn incremento abrupto en el nCAC en campañas de adquisición fría suele originarse porque algoritmos como Advantage+ desvían el presupuesto hacia clientes existentes (retargeting involuntario), lo que agota la audiencia y satura el anuncio en lugar de deberse a una falla en el mensaje o la pieza creativa.
if you're trying to get new customers, but you're not buying and getting enough new visits. So you're actually getting a lot of volume repeat customers. Yeah. So you see a lot of repeat customers. You're like, well, I'm buying the traffic supposed to be for new customers, but I'm going to tend to repeat something's off. Maybe advantage plus audiences expanded and has turned into a retargeting campaign as often. So I thought I was buying new customers. Yeah. Right. So that would be a case of traffic. You wanted new and you're getting warm or you're getting repeat. It'd be a as a advanced media…
Escúchalo en el minuto 38:25
Medición y experimentosCada campaña publicitaria debe evaluarse y optimizarse respecto a una única métrica primaria alineada con su objetivo (por ejemplo, ROAS de primer clic para prospección en tráfico frío, nuevo CAC para adquisición de clientes o ROAS consolidado para rentabilidad general), evitando la dispersión de optimizar métricas secundarias sin un propósito claro.
everywhere, but there's usually is one key metric. And we call it an MPI, you call it a KPI here, whatever it is, that's your North star. And so this is at a specific campaign, you would select between one of these intentions. You just select one of them. And based on that is the primary KPI, which we're then going to set the zone for, which I'll get to shortly. And then everything flows from there, but it starts with a focus on what am I trying to do with this campaign? But I mean, if you're doing cold traffic prospecting, you know, first click attributed row as is the North star for that…
Escúchalo en el minuto 15:45
Identificar qué productos específicos convierten a clientes primerizos en lugar de promocionar todo el catálogo a ciegas
Oferta y precioPara campañas de captación, se debe analizar qué SKUs o productos convirtieron inicialmente a los mejores clientes actuales en lugar de abrir todo el catálogo, ya que productos con alto volumen global pueden estar sustentados exclusivamente por compras repetidas y fallar al captar compradores nuevos.
Mm-hmm. And then product, product for new, you know, often you might be leading with a, you know, you're just throwing up a shopping campaign and they're taking in your whole feed of 2000 products We'll see some cases it will, but you probably have historical data and you may be leading with one campaign, you, one product you really hope to sell, but it turns out new customers don't like that. It just has a lot of sales on repeat or vice versa. And so segmenting the behavior of what your new, your best current customers did to make them first become customers and using that data. So there's…
Escúchalo en el minuto 44:42
Evolución del media buyer de operador táctico a estratega de datos
Canales de adquisiciónA medida que los algoritmos de IA en plataformas publicitarias automatizan la configuración técnica y la segmentación, el valor del media buyer se desplaza desde presionar botones hacia la interpretación estratégica de datos, la fijación de expectativas y la toma de decisiones de escalado.
I'm saying with the way Zuckerberg and all the ad companies are going, like the networks that allow us to do advertising on, they're trying to eliminate the need for media buying. But the role of a media buyer isn't going to go away. It's just going to evolve. And a lot of it has to be into how do you interpret and leverage the data? Because AI is going to be making a lot of the decisions you don't have to anymore. But you're still going to be needed for decisions like how are you going to scale appropriately? And how do you interpret the data that AI is choosing? And with Facebook, you're…
Escúchalo en el minuto 49:53
Riesgo de subatribución de GA4 frente a ventas reales en Meta Ads
Medición y experimentosLos informes de adquisición en herramientas como GA4 pueden subestimar severamente el impacto de campañas de Meta: en un caso citado, GA4 reportaba solo 16.000 $ en ingresos frente a un gasto de 30.000 $, cuando los ingresos reales generados por el canal rondaban los 130.000 $.
acquisition reports at GA four. And it says we're losing money on Facebook. That happened to a client It was showing they made 16 K in the month and they actually made about 130,000 on Facebook and Google Alex was saying 16 K and he was spending 30. So he's been at a grant a day. He's like, geez, should I, should I cut it? And I was like, no, no, you shouldn't. No. Or Facebook ad manager, you know, not seeing top of the funnel due to its short attribution windows.
Escúchalo en el minuto 24:03
La opacidad y sobre-atribución de las plataformas publicitarias exigen una medición independiente
Medición y experimentosA medida que las plataformas publicitarias operan como «cajas negras» con menor control manual y tienden a adjudicarse el crédito de todas las conversiones para incentivar un mayor gasto, la medición externa e independiente se vuelve crítica para determinar con precisión si realmente es rentable seguir invirtiendo en ellas.
going to mine and get there? And then with the ad platforms getting more and more black box, measurement's going to become more critical. Because you're really not going to have less control in the platforms. You need measurement to know whether you should still spend there. Because they're getting better and better at claiming all the credit in the world. And although they're very valuable to use, they want you to keep spending infinity if you could.
Escúchalo en el minuto 3:39
Fundamentos innegociables ante la automatización por IA: economía y psicología de la oferta
Oferta y precioEn un entorno donde la inteligencia artificial automatiza gran parte de las tareas tácticas del marketing, los componentes estratégicos que no se pueden descuidar son la viabilidad económica del modelo (unit economics) y la psicología que fundamenta la oferta.
Yeah, it's almost like foundational marketing that you're having in a world where AI is taking over so much. These are the pieces you can't neglect. You can't neglect the economics. And you definitely can't neglect the psychology behind your offer. All of those are going to matter tremendously.
Escúchalo en el minuto 50:32
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.