Not Your Granddad’s MMM: AI, Creativity, and What Comes Next

欢迎收听《现代营销与效果评估》。

Welcome to Modern Marketing and Measurement.

这是 Advertising Week 的播客，带你深入数据驱动营销的快速变革世界。

The Advertising Week podcast that takes you inside the fast changing world of data driven marketing.

每期节目都会为你带来行业领袖的深刻洞察，帮助你从面向未来的视角解读数据分析、归因和效果表现。

Each episode brings you sharp insights from industry leaders as we decode analytics, attribution, and performance through a future focused lens.

从新兴技术到隐私规则的演变，再到下一代效果评估策略，这档节目是你掌握行业前沿的指南。

From emerging technologies and shifting privacy rules to the next wave of measurement strategies, this is your guide to staying ahead of the curve.

在本期节目中，我与 The Brandtech Group 北美区首席执行官兼管理合伙人杰夫·马蒂索夫对话，聊聊人工智能如何重塑从营销效果评估到创意和媒介的方方面面。

On this episode, I sit down with Jeff Matisoff, North American CEO and managing partner at The Brandtech Group, to talk about how AI is reshaping everything from MMM to creative and media.

我们会探讨实时场景化评估究竟是怎样的，对话数据如何改变团队协作方式，以及在工具越来越智能的今天，保持清晰的目标和人性化的视角比以往任何时候都更重要。

We get into what real time scenario based measurement actually looks like, why conversational data is changing how teams work, and how keeping a clear North Star and a strong human point of view matters more than ever as the tools get smarter.

嗨，杰夫。

Hey, Jeff.

今天欢迎来到节目。

Welcome to the show today.

贝基，谢谢你邀请我。

Thank you, Becky.

能来参加这期节目我真的很开心。

I'm really excited to be here.

谢谢。

Thank you.

估计现在没人还不了解Brandtech这家公司了，你们合作的都是大家耳熟能详的行业伙伴。

So Brandtech, if if anybody's hiding under a rock somewhere, clearly, you've got relationships with people we've all heard of.

我知道Gravity Road这家公司的整体情况

I know Gravity Road collectively.

你还参与了Tracker的相关工作

You're involved with Tracker.

不对，应该说你和Griffin有投资层面的合作

If you don't, you're involved with, investment with Griffin.

如果你最近一直关注这些项目的话……我其实也有点记混了

If you've been paying attention to that lately, I I don't know.

你得往前看，拥抱未来。

You need to step into the future here.

你能给我讲讲 Brandtech 是做什么的吗？

But can you give me the rundown of Brandtech and what you do there?

当然可以。

Sure.

Brandtech 其实是几个不同业务的集合体。

Brandtech is a a con agglomeration of couple different things.

首先，我们按不同市场，拥有九到十一家代理公司和企业，这些都属于我们的集团旗下公司。

First, it's a standard model where we own, nine to 11 depending on which market, different agencies and companies, and those are the group companies.

同时，我们还有一支风投团队，专门在广告科技和营销科技的前沿领域进行大规模少数股权投资。

And then we're also have a VC and where we place sizable minority investments in the cutting edge of ad tech and martech across the board.

所以之前提到的那些重要合作关系，确实有不少，有些回报非常可观，有些则帮我们建立了绝佳的资源网络。

So that's where we mentioned they are sizable relationships that we have, and some of those are extremely profitable, some of them just get us to fantastic relationships.

举个例子，我们有个客户提出了三维视觉的创意，而 Zapper 在这方面做得特别出色。

As an example, one of our clients had an idea for three d visuals and that's what Zapper does really brilliantly.

所以我们直接找首席执行官。

So we call the CEO.

不用去找销售总监或者客户经理。

We don't have to get to a head of sales or an account manager.

我们可以给客户带来新东西，最后这些反而成了他们自己的产品。

We're able to bring new things to our clients that then end up being products for both of them.

集团旗下涵盖各种业务，比如你管理的那些，还有三年前收购的Jellyfish——过去几年我大部分时间都在那儿，还有Oliver，这是品牌内部的制作工作室，全球规模最大、获奖最多的。

But the group companies are everything from collectively like you managed and as well as Jellyfish, which was an acquisition about three years ago where I've spent the bulk of my time over the last couple of years, and Oliver, in house production studios for brands, the world's largest and most awarded in housing company.

最后但同样重要的是Pencil，这是我们端到端的AI与智能代理平台，从洞察、媒体投放、创意生成，到分析和优化，全都能覆盖。

And then last but not least, Pencil, which is our AI and agentic platform that goes from end to end, from insights into media, into creative generation, and analysis, and optimization.

我会不断提到Pencil，因为它是我们所有业务的核心纽带。

And we'll talk more and more, and I'll bring up Pencil over and over because it's the spine that links it all together for us.

虽然Brandtech是个母公司，但越来越多客户想要的是一个统一的解决方案，而不是一堆品牌logo，还总要问：‘那个工具到底叫什么来着？’

So while Brandtech is an umbrella company, more and more clients are looking for a single solution out brand logos, without well wait, what was that tool and well wait, what was that?

所以我们在这个领域是个颠覆者，更灵活、更高效，公司总数不到十家，通过AI把所有资源整合起来，为客户提供定制化服务——在北美，这正是我的职责。

So we're a disruptor in this space and we we can be more nimble, we can be more agile, there aren't hundreds of companies, there's less than a dozen and putting it all together through AI for our companies whatever they need and however they need it packaged in North America specifically is my job.

所以我们的目标当然是增长，但这种增长来自于贴近客户和那些在AI时代寻求帮助的资深营销人。

So the remit is of course growth but growth by being close to our clients and prospective senior marketers who are looking for help as they transform in the age of AI.

这工作真有意思。

What a fun job.

你像是在为其他公司量身打造最合适的工具和解决方案，感觉特别有掌控感。

It feels like you're the one in charge of right tooling and right sizing solutions for other companies.

听起来真的很棒。

That sounds really great.

现在干这行特别激动人心，因为终于，AI让我们实现了数字营销、程序化投放和动态创意原本该做到的事，接下来我会聊聊这些在媒体效果评估、洞察分析和数据看板上的落地情况。

It's a super exciting time to be in this job because finally, I feel like AI is allowing us to do the stuff that digital and programmatic and DCO was always supposed to get us to be able to do, and we'll talk about how that lands in MMM and insights and dashboarding.

过去十八个月是我职业生涯中最开心的时光，全是因为AI让我能自由尝试，我在生活中玩的东西，居然在工作中也完全适用——那种用自然语言来编程的感觉，真的太对路了。

But it's the the last eighteen months have been the most fun of my career, and it's all because of what you can dabble with in AI, what I can try in my personal life, and then see that actually that same kind of natural language vibe coding is super applicable at work too.

所以真的特别有趣。

So so it is really fun.

这确实很引人入胜。

Well, that's fascinating.

好的。

Okay.

既然你已经在这儿了，咱们就深入聊聊吧。

Let's go ahead and dig down into it since you're already there.

以前我们用MMM仪表盘主要是复盘，现在有了AI，可以随时随地灵活调整，按需响应。

We used to use MMM dashboards to kind of look back, and now we can use AI and be mobile and pivot on demand as we need.

这具体是怎么帮助你实时做决策的？

How does that actually help shape decisions in real time?

是的。

Yeah.

我们管这叫‘不是你爷爷那代的移动广告’。

So we call it not your granddad's like the old old mobile commercial.

我们其实已经做了好几年了。第一次变革是用自动化和机器学习，哪怕在AI之前，就能做到更实时，做出不再有六个月延迟的内容，也不只是核对那些从MMM里随手拉出来的数据点。

And we've been doing this for a couple The first revolution was just using automation and machine learning even pre AI to be more real time, be able to craft things that aren't six month lag or just verifying the bullets that are, you know, you just need to pull from MMM to use what you want.

现在，Jellyfish的产品已经能用上最近两周的数据，而且精细程度完全由你掌控。

And our product now next soon at Jellyfish is using data that's as recent as two weeks, that is as granular as you want it to go.

所以，如果你需要查看单品、创意概念或目标受众，这些数据和工具都能支持你做到这一点。

So if you can go to SKU or to creative concept or to audience, this data enables that and that the tool enables that.

而在过去六个月里，AI 更是让这一切得到了极大提升。

And that then, even in the last six months, has been supercharged with AI.

我们用的是谷歌的 Meridian，这是个开源工具，我相信你的听众都知道，它非常强大。借助它，我们现在能做情景规划，不再每次都要找数据科学家去跑模型，而是能实时响应——比如预算上调或下调时，系统能直接告诉你各渠道该怎么调整；或者你设定了目标，它能立刻告诉你需要多少资源才能达成。

We've powered it with Google's Meridian, which is open source as I'm sure all of your listeners know and is super super powerful and that's then allowed us to build a scenario planning aspect of it that doesn't have to go back to the data scientists every time to crunch but that is again more real time where we can have scenarios working on if I have budget either up or down that I need to factor in by channel, or if I have goals I need to hit, just tell me what it will take to hit this goal.

我们从两个方向来看，AI 让数据科学家能花更多时间去理解客户的业务，搞清楚哪些问题真正重要，而不是埋头写代码、跑场景、输出一堆 Excel 表格。

We look at it from both ends and AI has made, has allowed our data scientists to spend more time understanding their client's business and understanding the questions that are important as opposed to back office just churning through coding, churning through scenarios and outputting excel sheets.

把结果放进仪表盘，AI 也大大提升了效率。

Then getting that into dashboards is also super powered for AI.

我们不再需要那么多只负责调字体、改配色、搞视觉风格的数据可视化专员了，整个产出流程都发生了根本性变化。

We don't need as many gorgeous data visualizers who are doing nothing but adjusting fonts and colors, vibe coding and helping, and the process of creating the outputs has also been game changing.

所有这些进展，都是在过去六个月里发生的。我们现在还在探索下一阶段 AI 能力该是什么样——要能提供可靠的诊断，模型值得信赖，数据足够支撑客户的需求，还能接入外部数据，同时又能规模化服务像 Bissell、Workday、联合利华、万豪这样的大客户。

And all of that is in the last six months and we're still figuring out what does the next cycle of an AI powered look like that has the right diagnostics behind it, you trust the model, that has enough of the data in it that your client needs and has external data that we can connect to, but they can still be scaled to clients like Bissell and Workday and Unilever and Marriott and any of our group clients.

这正是我们想打造的方向，而 AI 已经彻底改变了我们对可能性的认知和执行速度。

That's what we're aiming to build, and again, AI has just transformed the speed and our view of what is possible with it.

是的

Yeah.

当然

Absolutely.

现在的报告速度大大提升了AI的敏捷性和信息价值

I mean, the the speed of reporting now just adds to the agility of AI and the the information you get.

我觉得这真的非常令人印象深刻

I find it incredibly impressive.

就在过去几个月，系统能直接生成一份精美的报告，把我要的所有数据都整合起来，人人都能一眼看懂，再也不用花几个小时去制作，省下的是真金白银

And, yeah, it's just been the past few months where you're just like, could just spit out a beautiful report that pulled all the data that I want, and everybody can visually understand it without spending hours, which is money, on making something.

对

Mhmm.

我平时是个业余的‘氛围编码’爱好者，生活中是这样，工作中也这样摸索着来

I'm a hobbyist I'm a hobbyist vibe coder both in my personal life and then kind of figuring things out at work.

一个月前，我想找张仪表盘的截图放进提案里，就想说明一下：我们确实做仪表盘

And a month or two ago, I was trying to just get an image of a dashboard to put in a pitch deck to say like, look, we do dashboards.

Gemini说：‘好的，我明白你想做个做仪表盘的应用。’

And Gemini said, okay, I see that you want to build an app to make dashboards.

这是你的应用。

Here's the app.

我说：‘不用不用，不过还是谢谢。’

And I was like, no no no, but thank you.

我会用这个的。

I'll use that.

是的。

Yeah.

不是。

No.

但也许那个梗。

But maybe that that meme.

不是。

No.

但也许吧。

But maybe.

对。

Yes.

没错。

Exactly.

所以是这样。

So yeah.

我们一直在讨论仪表盘的未来，其实两者都要兼顾。

So we we talk all the time about what is the future of dashboarding and it's a both and.

有时候你只是想要一个视觉界面，能快速回答你常问的问题，方便直接复制粘贴到文档或PPT里，或者在团队会议中调出来用；但同时你也希望有个能和数据聊天的界面——让你知道，支撑技术仪表盘的同一个BigQuery实例或数据体系，也在驱动你的洞察学习，实时告诉你当前发生了什么，而你无需再去找数据可视化人员重新做一张图表，就能换种方式提问。

There will be times where you simply want a visual interface that answers the questions you know you have routinely and to be able to cut and paste into into Docs or PowerPoints or bring up in group meetings, but also you want to have an interface where you can chat with your data, where you can know that the same underlying BQ instance or taxonomy that's powering the technical dashboard is also powering your insights learning and what is happening right now, and how do I ask questions a little bit differently without, again, going back to a data viz person to say, well, now I need a whole new chart on this dashboard.

我觉得这正是报告未来正在发生的变化所在。

I I think that's where, and things are just changing in in what the future of reporting even looks like.

明白了。

Okay.

好的。

Okay.

你这是给我搭了个梯子啊。

So you're segueing for me.

谢谢你啊。

I appreciate that.

你这么一说，我可轻松多了。

That makes my job so much easier.

那我先去吃个三明治了。

I'm I'm just gonna leave and go get a sandwich.

不过说真的。

But, no, seriously.

好的。

Okay.

这种转向对话式、问题驱动的分析方式。

So this shift towards more conversational question led analysis.

咱们先就这个话题多聊一会儿。

Let's stick with this for a minute.

我想插一句，那些自己摸索着学习的领导者们。

How and and I one, I wanna just interject right here in a moment that leaders learning this on their own.

你现在身居领导岗位。

You're in a leadership position.

你才是带领这个团队的人，这点很清楚。

You're the one who's, you know, leading the team here.

你现在既要干活，又要学这些，这至关重要。

The fact that you're also doing the work and learning it is imperative right now.

领导岗位上的人，如果既不亲自动手，也不懂这些事是怎么运作的，那是行不通的——因为变化太快了，对一无所知的东西，你根本没法开口谈。

You cannot have anybody in the leadership position that's not doing the work and understanding how these things are moving because they move so fast, and you can't speak to something you don't know anything about anymore.

这种做法根本行不通。

That just doesn't fly.

而你正在这么做，真的非常了不起。

And the fact that you're doing this is super impressive.

谢谢。

Thank you.

我觉得这一点特别重要。

I think that is so important.

我不是做数据分析出身的。

And I am not a analytics, background.

我能跟这个领域对话，也能融入其中，但我不是数据科学家。

I can speak and live in that world, but I'm not a data scientist.

我不是分析师。

I'm not an analyst.

我不是程序员。

I am not a coder.

我没有工程或编程背景，但我发现，要想对这些东西聊得明白、说得在行，就得亲自上手去折腾。

I do not have a background in engineering or in code, but I've learned that in order to be, conversational and smart about this stuff, I have to get my myself dirty with it.

所以我花了很多时间去了解，在Antigravity、Gemini、Claude这些工具里到底能做什么。你只要自己动手解决几个小问题，就能大幅提升对工作场景的预期，明白哪些是局限，甚至会想：如果我自己都能想到一种做法，那专门做这个的人肯定更能搞定。

So I have spent a lot of time to understand what can you do in antigravity, in Gemini, in Claude, and the more that you even just solve little problems yourself, it just ups what you can expect in work settings, what you know the limitations could be, or even if you're like, okay, look, if I can imagine a way to do this, then certainly someone whose job it is can definitely get this.

所以我有个目标，就是今年让团队里的每个人都真正掌握AI应用能力。

So a confidence and education, and so I have a personal goal with my team of getting everybody to be applicable in their AI fluency this year.

别管做什么，先动手试试。

Just do something, do anything.

我现在懂GitHub是干嘛的了。

I know what GitHub is now.

我也知道Versal是什么了。

I know what Versal is now.

我知道该怎么用Twilio。

I know how to use Twilio.

这些事，一年前我根本说不出来。

Like, all of those things I wouldn't have been able to say a year ago.

所以我完全理解你，这确实是我经常提到的平台。

So so I'm with you, and it is obviously a a platform of mine that I talk about all the time.

对。

Yeah.

我们的通才——当然，专家依然很重要。

If our generalists I mean, specialists are still important.

专家我们还得要，但通才也很关键，我自己也算个通才。

We still need our specialists, but our general generalists I'm a generalist as well.

我们必须提升自己，如果做不到，迟早会被淘汰。

We need to level up, and if we can't do it, then we we're not gonna exist anymore.

这会是一场巨大的转变，我觉得变革就发生在这里。

It's gonna be a huge and I feel like that's gonna be where the shift is.

当人们担心AI会抢走工作时，其实不是抢走，而是你得会用它。

When people are worried about AI taking your jobs, well, no, but you better know how to use it.

我很欣赏你一直在推动和引领。

And I love that you're keeping and leading.

结合你的经验和知识，你是怎么在领导工作中，向市场团队解释他们每天该怎么用这些数据的？

And this is kind of like where then with this knowledge of yours and this experience of yours, how do you use this in your leadership on explaining to way marketers need to interact with this data day to day?

在我升任区域CEO之前的一年半里，我花了大约十八个月，带领团队彻底重建了Jellyfish的媒体技术，从头到尾实现智能化和AI驱动。

Over the last year and a half before I, was promoted into the regional CEO role, I spent about the last eighteen months spearheading a team that was focused on rebuilding the jellyfish media technology to be agentic and AI powered from end to end.

那里有五个模块，我现在要讲的是其中一个，叫‘观察者’，这正是我们刚才在讨论的东西。

And there are five modules there, the one I'll talk about now is called Observer, and that is exactly what we're talking about.

我们怎么把一个大语言模型指向BQ实例，或者客户用来驱动仪表盘的任何数据库，然后返回他们提出的相同问题、相同提示，方式要足够直观，可以直接分享给客户，或者直接复制粘贴，作为第一版内容，供主管或总监进一步润色和深化。

How do we take an LLM, point it at the BQ instance or wherever the client's database that's already powering the dashboard and return back the same prompts, same things, the same questions that they're asking in a way that is easily either shared directly with clients or cut and pastable as almost V1 for then a supervisor or a director to nuance and get back further and further.

我们现在已将它重新构建为The Brandtech Group的操作系统——Pencil，让它更互动、更像对话、更有活力，并且能连接更多平台，这也彻底改变了我们所谓的投资分析团队的工作方式。

We've now rebuilt that into pencil, the operating system of The Brandtech Group to be even more engaging, more conversational, energetic and connected to additional platforms as well And that's changed our entire approach to our what we call our investment analytics team.

他们不再只是坐在那里写报告、列前七个要点。

Again, they're no longer just sitting crafting reports and putting out the first seven bullets.

我们现在能获得跨平台的洞察和行动建议，有时候这些建议会让人特别惊讶，毕竟大语言模型并不懂付费搜索团队和自然搜索团队之间的区别。

We can get to cross, cross platform insights, actions, and, sometimes the actions are super astounding because obviously the LLM doesn't understand what is a paid search team versus an organic search team.

但这对我们和客户来说反而是好事，因为它会提出类似这样的问题：‘你们的SEO标签有没有可能出了问题？’——这种问题，付费搜索团队可能一时想不到，但我们团队里正好有这类专家，在Jellyfish里我们都有。

And that's great for us and our clients because it will say things like, could there be something broken in your SEO tagging that a paid search kid may not always think of right away but we have those teams, we have those teams in jellyfish.

我们为客户服务时，这些团队通常都是配套配置的。

They're almost always scoped together on our clients.

所以现在，这个团队成员就可以说：‘我有个想法。’

So now that kid can go and say, I had an idea.

现在我们来深入看看以前很难发现的问题。

Now let's go dive into something that we wouldn't have been able to as easily diagnose before.

这一切都得益于对话式AI的介入。

And it's all the conversational AI interfacing that's enabling that.

这太棒了。

I love that.

明白了。

Okay.

那我们再深入一点，看看AI怎么分析你们的创意内容。

So then let's deep dive into this a little bit more than AI can then look at your creative.

我们把所有数据都整合到仪表盘上，就是为了消除信息碎片化，然后反过来分析归因，看看哪些创意有效，为什么有效。

Is it well, we brought everything up to the dashboard, and we're, like, we're trying to get rid of fragmentality and all of these things, then we wanna turn around and look at attribution and look at the creative and seeing which part of the creatives are working and why it's working.

你们团队是怎么协助做这件事的？

Can you tell me how your team is, helping with that?

是的。

Yeah.

所以，再次说明，在我们叫做 Pencil 的操作系统里，有由 AI 驱动的创意效果评估机制。

So, again, inside our, operating system called Pencil, that we have creative effectiveness measures that are driven by AI.

如果客户有自己特定的格式或缩写来衡量创意效果，我们也会采用。

We adopt either if a client has a specific format or acronym for how they gauge creative, success.

举个搞笑的例子，比如必须是红色、要有棵树，还得出现一个人。

Let's just use something silly that it has to be red, have a tree in it, and just show a human.

不错。

Cool.

很好。

Great.

我们可以把这三点作为他们的基础评估框架。

We can use those three things as their foundational framework.

然后让大模型去判断：红色是不是对的色调？

Use it pointing an LLM at it to say, is it the right shade of red?

树的类型是不是符合要求？

Does it have the right type of tree in it?

而且，那个人是不是我们指定的那一个？

And is there a human that is the right one that we've specified?

然后根据这些给创意效果打分，再进一步分析：这些广告在特定格式下，是否真正推动了我们想要的效果？媒体指标固然重要，但通常不是最终答案；我们要更进一步，通过调研面板和受众数据，判断它是否真正提升了品牌指标或转化效果。

Give it a creative effectiveness score from there and then start to map that to, okay, did those ads uniquely and that specific format drive the performance that we need either in media metrics, which are fine, but generally not the end answer, all the way then past using panels and audience specific data to get to did it move the needle on brand metrics or conversion metrics.

这三个要素缺一不可。

You have to have all three of those components.

首先，得有个标准来判断：这个创意本身到底好不好？

You have to have a format for is this creative even any good?

如果不用AI，我们会不会放行这个创意？

Would we have let this go out in an unai format?

这个创意，会不会让我们的创意总监或者品牌负责人，甚至客户都说：‘我全力支持这个’？

Would this have been creative that our ECD or our head of brand that a client would say, yes, I put my my my weight behind that.

其次，因为我们越来越多的工作都在平台上进行，它必须能对应到正确的媒体指标。

Second, because more and more of everything we're doing is in platform, it has to map to the right media metrics.

如果分类不够清晰，如果把程序化投放和别的东西混在一起，而它们的定义又大相径庭，那你的创意也必须做出截然不同的调整。

If you don't have them isolated the right way, if the taxonomy is not correct, if you hold programmatic to one thing but linked into another and they're wildly different, then obviously your creative has to be wildly different.

第三，如果这些无法连接到面板、受众，或者你们电商或DTC业务的后端系统——而且幸运的是，你们有这种可扩展的指标体系，否则你们在整个流程中就会漏掉关键信息。

And the third, if those don't link to either a panel or an audience or a back end system that's one to one of your e com or DTC and lucky enough to have that scalability of a of metric, then then you're losing something across that pipeline.

我太喜欢这个观点了。

I love that.

这太精彩了。

That's brilliant.

明白了。

Okay.

那我们稍微换个话题。

So let's switch gears for a little bit.

我回头再聊这个。

I wanna come back to this.

我会再回来谈这个，但在这之前，我想也聊聊在注重隐私的环境下其他一些事情。

I'm I'm going to come back to this, but I wanna, like, circle in some other things as well in a privacy first world.

在这样的环境下，营销人员应该怎么整合这些不同的方法呢？

How should marketers be thinking about combining these different approaches?

你需要靠建模、测试和归因，才能获得更完整的视角。

You got modeling, testing, attribution to get a more complete picture.

由于信号丢失和碎片化，现在根本找不到一个明确的真相来源。

It's with the signal loss and fragmentation, there's no clear source of truth anymore.

那现在人们到底该怎么应对这个问题呢？

How do you fit figure people should go about this now?

这个问题绝对值得好好聊聊。

This is one that's gonna be a great conversation.

我没有答案，也没人有答案。

I I don't have the answer to this, and no one has the answer to this.

关键是先搞清楚：你的真相来源到底是什么？

The key is figuring out first, okay, well, what is your source of truth?

你说到点上了。

You said it.

哪怕知道它不是最理想的，只要有一个统一的真相来源，并且明白它还在不断完善中——毕竟99.5%的营销人员都是这样——这一点就至关重要。

Having a single source of truth that is aligned and even if you know that it's not best case, if it's a work in progress which is like ninety nine and a half percent of all marketers, just knowing that this is today's single source of truth is is crucial.

我们每个客户都是从这一点开始的。

And that's where we start with every client we have.

令人惊讶的是，那么多大型跨国营销团队都答不上来这个问题。

And it's amazing how many enterprise, huge global marketers struggle to answer that question.

所以我不确定，尤其是在如今注重隐私的时代，再加上AI不断敲响隔壁的门——像OpenClaw这类技术，我们都既兴奋又害怕。

So I don't I don't know, especially in a privacy first era, especially with AI, like knocking on the walls next door, Open Claw, and all of the things we're all excited but terrified by.

我真的说不准。

I don't know.

最近有没有谁提出过什么方法，是你觉得不错的，或者看到过什么让你印象深刻的？

Has anyone who's been on recently suggested an approach that you liked or seen something you like?

天哪。

Oh gosh.

我的天。

Oh my god.

这还挺有意思的。

It's kind of interesting.

我们还在试用第三方安全数据室这类工具。

We're still playing around with, like, third party clean room data.

你得想办法把事情推起来。

You're trying to get things going on.

你想表达得清晰一点，但说实话，你刚才说得正好。

You're trying to be able to to give a clear voice, but I think you're again, we're as such you'd hit it on the head.

我们现在正处在一个颠覆性的时期。

We're in a disruptive time.

很多事情都在飞速变化。

We've got a lot of things changing super fast.

有位嘉宾最近说，我们正站在矛尖上，而这矛尖正在高速冲向前方。

A guest recently said we are on the tip of the spear, and the spear is hurtling.

你懂的。

You know?

而且这就像是……

And that's like the the there's

确实如此。

the Totally is.

为你而生。

For you.

是的。

Yeah.

真的就是这样。

It totally is.

我觉得你说得对。

And I I think that's right.

我们以前经常聊清洁室，但现在不怎么提了。

You know, we talk about clean rooms a lot in that.

现在几乎没人谈清洁室了，这感觉挺奇怪的。

We don't we aren't talking much about clean rooms anymore, and that feels really weird.

我不确定是不是因为当下变化太快，就像那支飞速前进的矛，我们还没跟上节奏。

And I don't know if that's a sense of, again, the the moment the spear hurdling, and it just hasn't caught back up.

清洁室会以一种更适应AI的方式回归吗？还是说别的什么？

Are are clean rooms gonna come back in an AI friendly way or or what there?

对。

Yeah.

我听过的最好说法就是：不断测试，不断测试，再不断测试。

The best answer I've heard so far is testing, testing, testing.

事实就是，我们目前还处在这样的阶段。

We're still this is the truth of it.

现在我们正处在一个阶段，你必须不停地做测试。

We're in a in a period right now where we are just you gotta keep testing.

你得不断尝试各种方法，看看哪些能行得通。

You gotta keep, like, you know, putting in different things and seeing what sticks.

对我这种老一辈、入行很久的人来说，现在正是该沉下心来的时候。

And and I think for me personally, for us old timers who've been in this for a long time, it's kind of when we needed a hunker back down.

在开始这个播客之前，我们还聊到，如果某个方法不管用，就直接放弃。

And before we started this podcast, were talking about, if something doesn't work, you throw it out.

对吧？

Right?

我觉得我们又回到了营销基础课的第一课。

I think we're back at the old tenants of marketing one hundred one.

营销本质上是推动销售的工具。

Know, like marketing is a tool for sales.

不管什么方法，行不通就直接扔掉。

If it's not working, you gotta throw it out.

了解你的受众，这是最基础的。

Know your audience 101.

得重新回到这些基本功，不然一上来就误用这套新玩法，反而把事情搞砸了。

You gotta get back to some of your basic stuff in order to not apply this brand new changing world incorrectly and completely, like, mess it up on the get go.

你懂我的意思吧？

You know what I mean?

我完全同意，这一点跟我们在MMM里聊的很像。

I couldn't agree more, and that's point one is similar to what we were talking about on MMM.

让我们的数据科学家去和客户沟通、理解客户的需求，变得尤为重要，也特别有价值。

Having our data scientists talk to and understand their clients becomes so so more critical and such a beautiful place.

这就是我们选择加入营销机构的原因。

That's why we joined agencies.

我们之所以做营销，就是因为我们想搞懂这些事。

That's why we're marketers anyways because we want to understand those things.

第二点，我们对营销人员的期待是明确一个北极星目标。

And then the second part is what we want from our marketers is a North Star.

我们做营销，到底是为了什么？

It's what are we trying to do with the marketing?

对。

Yes.

目标是帮助客户成长，但针对这个项目、这场活动、这次AI转型，我们具体想达成的北极星是什么？

It is helping them grow, but specifically for this initiative, for this campaign, for this AI transformation, what is the North Star that we're trying to get?

在营销科技时代，我们太过分散，做了太多孤立的试点和短周期项目。

We devolved too much into isolated pilots and sprints in the MarTech era.

但在AI时代，你没法这么做。

And the AI era, you cannot do that.

如果你没有一个明确的方向，不知道这些工具和AI该为你做些什么，你肯定会失败。

You will fail if you don't have a North Star for where all of these pieces and what you need AI to do for you.

没错。

Exactly.

我们想法完全一致。

We're on the exact same page.

这一点我太认同了。

I love that.

那我们接着说。

So then let's okay.

回头再讲一下，因为你之前提到‘如果不是AI，我们会这么做吗？’，这正好说到我想回过头来讨论的地方。

Circle back then because this hits on where I wanted to circle back when you said, would we have done it if it wasn't AI?

虽然我们的工具越来越智能，但最终的决策依然需要人的视角。

And that brings me to because our tools may be getting smarter, the decisions still need that human point of view.

品牌怎么确保不丢失自己独特的个性？

How do brands make sure they don't lose what makes them distinctive?

比如他们的语气、创意和那种独特的风格。

You know, their voice, their creativity, their edge.

我们这不正好回到刚才说的那个问题了吗？你的北极星到底是什么？

You know, we're we're speaking right here back to, like, what's your North Star here?

他们该怎么保证拥有这一点？

How do they make sure they have that?

完全正确。

Absolutely.

我觉得可以从两个方面来说，一个是创意，另一个是媒介。

And I think there's two I'm gonna pull that for creative and then pull it for media.

在创意方面，人的参与依然和以前一样关键，就像当年创意设计和制作刚转向数字工具时，你得懂怎么用这些工具——这正是我最欣赏的一点。

For creative, the human in the loop is as essential as ever in the same way that when creative design and production moved into digital tools the first time, you needed to know how to use the tools, and it's one of the things I love most.

Pencil的创始人威尔·亨舍尔总说，他们取名‘Pencil’，就是希望它能成为创意人员真正用得上的工具。

The founder of Pencil, Will Henschel, always says he named it Pencil because he wants it to be a tool that creatives use.

这并不是取代。

This is not replacing.

这只不过是放大了他们原本能做的事情。

This is not, doing anything more than amplifying what they can do.

当我们看到创意总监和艺术指导们使用这个工具时，他们能做灯光设计、角色塑造、场景搭建，但用的是自然语言指令，而不是画笔或调整摄影参数。

And seeing our ECDs and creative directors in the tool, they're able to do lighting, character development, environments, but you're doing it with natural language prompts, not with a stylus or with, adjusting photography.

把二维图像转成三维，这本来就是数字工具一直都在做的事。

Turning two d images into three d is what what what, digital tools have always done.

现在只是更快、更便宜、更容易，而且效果还更好。

It's now just faster and cheaper and easier and you get better output from it.

所以在创意和内容方面，人的作用一点都没减弱。

So from creative and content, the human is as important as ever.

我说过，操作的人和操作系统一样重要。

The operator is just as important as the operating system is what I say.

我没法直接在Pencil里生成一个精美的30秒广告，用来在超级碗周日播出，但我们的Oliver创意团队就能，而且他们确实做到了。

I can't go into pencil and get a gorgeous thirty second spot to run on Super Bowl Super Bowl Sunday, but our Oliver creative team sure can and they have.

媒体这块有点不一样，因为这里有规则，有行业惯例，这些都可以被自动化，也能训练进AI系统，而且大致分成三部分。

Media is a little bit different in that, in that there are rules, there are norms that can be automated and that can be trained in an AI and those come in three different parts.

第一部分是所有平台告诉我们的东西，但我们都知道，他们只说了六成真相。

First, there's the stuff that all of the platforms tell us is important and we know that they're telling us 60% of the story.

接下来的三成，是广告公司通过实践摸索出来的，他们得确保媒体投放真正为客户的利益服务，而不仅仅是迎合平台。

The next 30% is what agencies know and learn and adjust to make sure that the media is working for their clients not just for the platforms.

最后那百分之十，是品牌方早就摸透了的经验，他们早就知道该怎么做了，我们真没必要重新再学一遍——所以我们训练所有AI模型和媒体系统时，都会把这三方面都纳入考量，这样就能在每个客户身上既贯彻Brandtech和Jellyfish的最佳实践，又不会搞混方向，而这就是人类介入的关键：首先确保AI不会胡编乱造，说我们没说过的话；其次，就像我们之前提到的数据科学家一样，他们得花时间跟客户深度沟通。

And then the last 10% is what brands have already learned, they already know, they've already figured this out, we don't have to learn it again please and we train all of the AI models and media with those three things so then we are able to give a very distinct yet still apply Brandtech and Jellyfish best practices into every client without muddying the waters and that's where the human in the loop comes in to make sure that first of all, the AI isn't hallucinating that we said that it said what it's supposed to do and that we're able again, like data scientists in the measurement section that we talked about are spending time with clients.

我们的媒体团队也能腾出更多时间，跟客户、媒体方和平台深入交流，了解最新动态，做测试——就像你刚才提到的那一点特别关键：咱们不妨把所有新方法都试一遍，然后用这些实际结果来更新我们的规则和模型训练数据。

Our media teams get to spend more time with clients and also with publishers and platforms to learn what's new, to test, to use your very, very key point on the last point is like, let's try all of these things and use that to update the norms and the training that we're giving to the models.

这个说法太棒了。

I love that.

回答得真好。

That's a great answer.

还有那个‘幻觉’的问题，你刚才说对了。

I also the hallucinations correct me.

我遇到的不是幻觉，而是重复——有时候你得给AI注入点新东西，不然它就会陷入自我循环。

If I don't get hallucinations so much as, the repetitive if sometimes you gotta, like, put something fresh into AI or ends up being the snake eating its own tail.

你就是一直在给我重复同样的内容。

Like, you're just, like, giving me the same thing over and over again.

或者我老得提醒大家：别用行话，千万别用行话。

Or I'm constantly telling us if you don't use jargon, do not use jargon.

别用那些时髦的套话。

Don't use buzz phrases.

天哪。

Oh my gosh.

我从不用那些时髦的套话。

I don't use buzz phrases.

别说了。

Stop it.

我老跟它说，别夸我。

I keep telling it not to compliment me.

我受够了它总是这么客气。

I'm tired of it being very kind.

所以我每次一开始都会说，全程保持客观，别夸任何东西。

So that's always my first prompt is throughout this, stay objective, do not compliment anything.

我的自尊心已经够强了。

I already have enough of an ego.

别说了。

Stop it.

没错。

Exactly.

我知道我在这方面很在行。

I know I'm good at this.

哦，这样啊。

Oh, okay.

考虑到这些，你觉得这个方向该往哪儿走？或者你希望它往哪儿发展？

So with all of this in mind, where do you see this, or where do you want to see this?

我不再问‘你计划用多少年’，现在我会问：六个月、九个月，也就是一两个季度、三个季度之后，你希望看到它发展到什么程度？

I I no longer say tell me years, but now tell me in six months, nine months, we got a couple of quarters, three quarters, where do you want to see this going?

这是个很好的问题。

That is a great question.

无论是 Jellyfish、Oliver 还是整个团队，我们都引以为豪的是，通过 Pencil 这个端到端系统，我们正在打破 P 和 L 之间的隔阂、运营公司之间的壁垒，以及针对不同客户该做什么和真正对他们最有利之间界限。

I think what we are proud of, both in Jellyfish and Oliver and across the group, is that we are that with Pencil as this end to end system, we're breaking down the barriers between p and l's, between opcos, between what scope you have with a certain client versus what's just best for them.

当我们能完整地整合媒体、分析和创意，从brief到内容再到效果实现全链路自动衔接时，我们就能对创意客户说：‘好了，谢谢，客户A，你们团队原本负责分析的部分，其实现在我们已经能直接提供了。’

And so, when we have a fully finished product in media and analytics and creative and we can map brief to content to effectiveness all agentically, then we'll be able to say, cool, thank you creative client A.

我知道你们另有团队做数据分析，但这些功能现在我们已经能直接给你们用了。

I know you have this other team that does analytics, but this is already available to you.

今天，Pencil 里很多工作我们其实已经能做了，但一旦客户帮我们连上广告账户、接入更多数据源，我们的能力就能大幅提升。

And today, we can do a lot of that work already available inside Pencil, but when our clients help us connect ad accounts, connect different data sources in, we're able to do so much more.

所以六个月后，我预计客户不会再因为要对接各种数据源而拖慢进度。

So in six months, I think that clients will stop being blockers and connecting different data sources.

他们会不再把媒体代理和创意团队看作是各自为政、数据互不相通的两个部门。

They'll stop thinking of their media agency as a different owner of data than their creative team.

他们会把营销技术供应商看作是整个组织的资源，而不仅仅是IT部门或高级市场人员的工具。

They'll think of MarTech suppliers across the board, not just for the IT and the senior marketer IT staff.

所有这些都将融合在一起，让智能系统变得更聪明、能获取更多数据，并在不同领域之间无缝协作。

All of this will converge to allow the agents to be smarter, to access more, and be interoperable across discipline.

这太棒了。

I love it.

你们会把事情看得不那么线性，而是更立体、更全面。

You'd visualize things less linear and more three dimensionally.

我们要加上一个Z轴，从全方位来看待所有问题。

We're gonna add a z plane here and look at everything.

我喜欢这个想法。

I love that.

我觉得说得对。

I think that's right.

因为一旦能做到这一点，焦点就会重新回到消费者身上。

Because when we can do that, then it becomes back to the consumer again.

消费者根本不会去区分创意、媒体和追踪这些环节，而智能代理能帮我们重新回到这种整体视角。

The consumer doesn't think about the creative versus the media versus the tracking, and agents can help us get back to that again.

就像你说的，回归营销最基础的原理。

Like you said, getting back to marketing one zero one.

是的。

Yeah.

没错。

Exactly.

杰夫，今天这场对话真的太棒了。

This has been really fantastic, Jeff.

非常感谢你抽出时间。

I appreciate your time.

如果有人想更多了解你，或者想联系你个人以及Brandtech，该去哪里呢？

If people want to learn more about you and connect with you personally or Brandtech, where would they go?

如果想找我，LinkedIn是最合适的地方。

If they want me, Linkedin is the best place to find me.

有人给我发消息，我都会尽量回复，真的很高兴能和大家交流，因为正如你所说，前沿正在快速推进，每次和别人聊这些话题，我都能学到新东西，往往还会冒出一个点子，让人忍不住说：‘对，就是这样！’

I try to respond to anyone who sends a message, and I definitely am thrilled to connect and talk to everyone because, the tip of the spear is hurdling, as you said, and every conversation I have with someone about this stuff, I get smarter, and we find an idea that says, oh, yeah.

咱们可以试试这个主意。

We should try that out.

所以，一定记得去那里找我。

So please, definitely there.

还有，Brandtech，欢迎访问我们的官网。

And Brandtech, come to our website.

去网站上填个表格，想了解BrandtechGroup.com的更多信息，随时都可以行动。

Go to the website, fill out a form, anything you wanna do to find out more about brandtechgroup.com.

很好，Jeff。

Great, Jeff.

非常感谢你今天抽空来参加。

Thanks so much for coming today.

谢谢你，Becky。

Thanks, Becky.

如果你想听更多这样的播客，别忘了关注广告周为广告、营销和科技行业推出的丰富播客阵容。

For more podcasts like this one, be sure to check out Advertising Week's ever growing roster of podcasts for the advertising, marketing, and technology industries.

这些播客都可以在 www.advertisingweek.com/podcasts 上找到。

You can find those at www.advertisingweek.com/podcasts.