Marketing Hasn't Really Changed — Until the Day Ad People No Longer Need to Open a Dashboard

A couple of days ago, I was chatting with a friend who works in ad buying.

He said, "Liu Run, guess which piece of software I spend the most time in every day?"

I said, "WeChat?"

He shook his head.

"Excel?"

Still shaking his head.

"The dashboard."

He pulled out his phone to show me. The dashboard of an ad-buying platform. A column of menus down the left side — drop down, drop down again. On the right, chart after chart: conversion rate, click-through rate, spend, ROI. He said he spends at least four or five hours a day, every day, jumping back and forth between a dozen-plus of these dashboards — all to figure out one single thing:

The money I'm spending — is any of it actually working?

When I heard that, my heart sank a little.

Not because it was anything new. Exactly the opposite — because it's so common. So common that all of us have just accepted that this is how it's supposed to be.

But why on earth should marketers have to burn the best few hours of their day clicking through menus and staring at reports?

One

Let's step back for a second.

Over the past decade or so, how have the tools in a marketer's hands actually evolved?

In the beginning, there was the "campaign management system." You'd build a plan, set a schedule, keep an eye on execution.

Later, the tools got an upgrade — they became what people called a "marketing operating system." More features, more things to manage. Sounded like the real thing.

And then?

And then they stopped climbing. The tools kept multiplying, the dashboards kept getting deeper, but underneath it was the same one thing:

You open an interface, look at a pile of numbers, and then you decide what to do.

This model has a name: the "reactive workflow."

What does "reactive" mean?

It means: if you don't click, it doesn't move. If you don't switch to that report, that report won't come tell you where the problem is. If you don't export that dataset, that dataset just lies there asleep.

The tool waits for you to come use it.

And you — you're the one running yourself ragged across a dozen platforms. One for social, one for programmatic buying, one for the data dashboard, another for customer management. Every time you open one, you relearn the menus, you re-stitch the context together.

You think you're using the tools. Really, the tools are using you.

Two

So when I heard that an ad-tech company recently did something kind of interesting, I took a second look.

What did they do?

They decided, basically, to push the whole idea of "the dashboard" into the background.

Not delete it. Demote it.

In its place: an agent that actually understands what you're saying. You don't have to dig through menus, export reports, figure it out yourself anymore. You just tell it what you want. "Push this product to this audience, here's the budget, here's the goal." And then it analyzes, it recommends, it executes.

Sound a little familiar?

Of course it does. The last couple of years, who isn't hanging an "agent" off their software? Slapping a chat box onto your product has basically become table stakes.

But — and here's the but —

Anyone can build a chat box. What actually separates the winners from the rest isn't "whether" you have a bot that can chat. It's whether that bot actually understands advertising.

Three

Think about it.

You take a set of ad-buying data, throw it at a general-purpose large language model, and ask it, "What should I do next?" It'll give you a slick-sounding analysis. Polished wording, clean logic, even nicely bulleted for you.

But the moment you actually try to follow it, something's off. It has no idea how people in your industry actually bid, when you're supposed to scale up budgets, which audiences you should let the system learn on, and which ones you have to lock down manually.

Why?

Because it's a generalist, not a specialist.

A generalist knows ten thousand things, but none of them deeply. You ask it a broad question, it gives you a broad answer. You ask it a specific, only-someone-who's-done-this-job-would-know kind of detail, and it starts talking in circles.

And that's the one thing this company figured out:

The moat for the advertising agent of the future isn't "it can hold a conversation." It's "it genuinely knows the craft."

What should an agent that truly understands advertising look like?

It has to know what kind of context makes a good buying environment. It has to know whether an opportunity is worth chasing. It has to know what "push the budget now" versus "pull back now" actually mean, respectively. And then — it has to actually do the work. Not just talk about it.

Plainly put: it has to be like having a ten-year veteran optimizer sitting next to you. You say one sentence, it gets it instantly, turns around, and the work's done.

Four

Whether this actually pans out — it's still too early to call.

Natural-language interfaces were a fresh idea three or four years ago; today, they're not really a selling point anymore. Everybody's seen them. Nobody's impressed.

The real question was never "can you operate software with your mouth."

The real question is: after you've said it, does anything actually happen?

You say, "I want to bring down customer acquisition cost this month." Does it, after hearing that, spit out a slick analysis report — or does it actually restructure your account, kill the campaigns losing money, and move the budget over to where the money's being made?

The former — that's a slide deck.

The latter — that's doing the work.

If somebody can actually crack this — turn a goal you stated out loud into visible changes inside your account — then at least for this stretch, they'll run a little faster than everyone else.

I say "this stretch" because that kind of lead never lasts long. The rest will catch up.

But at least they moved first.

Five

Coming back to where we started — what I actually want to talk about isn't any one company.

What I want to talk about is a trend.

And the trend is this:

The "interface" between software and humans is shifting, from "you operate" to "it executes."

For the past decade, we've gotten used to a rhythm: open the dashboard, find the feature, read the data, make the call. We've come to treat this whole routine as just the natural order of things.

But it isn't the natural order of things. It was only ever a stopgap — the compromise you make when the technology isn't ready yet.

Once that thing — the thing that understands you AND can do the work for you — actually shows up, you'll realize how much of the time you spent clicking through menus was, honestly, a bit of a waste.

My friend asked me later, "So does that mean I won't have to open the dashboard anymore?"

I said, "Maybe."

He said, "Then what am I supposed to do?"

I said, "Now that's the question you should be asking."

Once the tools are doing the execution for you, your value isn't in execution anymore. It's in knowing what to do, why to do it, and where to go after it's done.

That's where marketers should really be spending their time. Not hunting for data across a dozen dashboards.

It's thinking clearly about what result you actually want.

Maybe in a couple of years, looking back, we'll think the days of "open the dashboard, read the reports" were kind of dumb.

The way we now look back at twenty years ago, when we had to go to the library and flip through a card catalog to find a book.

Back then, that felt like the natural order of things too.