How We Think About Programmatic Advertising in B2B (Especially Audience Targeting)

We got a great question recently from a customer building out their programmatic strategy:

‍LinkedIn targeting makes sense — job titles are easy. But how does that same targeting actually work on Meta, Reddit, YouTube, or open-web programmatic? A lot of the vendor collateral sounds like jargon. Are these third-party audiences actually real?”

It’s a fair question.
And honestly… we agree with the skepticism.

A lot of B2B marketers are trying to scale beyond LinkedIn CPMs, and programmatic is the obvious next stop. But the mechanics of targeting across channels are not the same — and misunderstanding that is where most wasted spend comes from.

Here’s how we recommend thinking about it.

First Principle: Not All Targeting Is Created Equal

A lot of vendor collateral talks about:

• “IP-based identity graphs”
  
• “Intent signals”
  
• “Behavioral clusters”
  
• “Scalable audience modeling”
  

Some of that is useful.
Some of it is… marketing copy.

The real question isn’t how the data is described.
It’s how the targeting is validated inside the channel.

This matters a lot in B2B, where the difference between hitting a VP of IT vs. a random consumer is the difference between pipeline and noise.

Walled Gardens vs Programmatic: Same Goal, Very Different Mechanics

The goal of paid social and programmatic is the same:

👉 Reach the right people at scale

But the mechanics — and the risk — are very different.

1) Walled Gardens (LinkedIn, Meta, Reddit, YouTube)

When we use Primer audiences in a walled garden, we’re not really “doing programmatic.”

We’re handing the platform a first-party audience definition.

They do the rest.

Why this works:

• Platforms have logged-in identity graphs
  
• They know who the user actually is
  
• Matching is deterministic (name/email/phone → account)
  
• Inclusion/exclusion is enforced consistently across inventory
  

At Primer, when we push audiences into these platforms, we’re using identifiers like name, email, or phone — not IP guesses. The platform matches those to real users in their graph.

That’s why Primer audiences are:

• Repeatable
  
• Measurable
  
• Consistent
  
• Often outperform native targeting
  

Because the platform controls the last mile.

2) Programmatic / Open Web

Programmatic is a completely different beast.

There is no unified identity layer across the open web.

Every publisher has:

• Their own login base
  
• Their own consent rules
  
• Their own ID framework
  
• Their own data quality
  

Even when hashed emails or UID frameworks are used, they’re still publisher-specific.

And outside logged-in inventory?

Everything becomes probabilistic:

• IP inference
  
• Device signals
  
• Behavioral modeling
  
• Contextual inference
  
• Modeled expansion
  

That’s not a bug — that’s how programmatic works.

But it means:

👉 Audience purity degrades as you push for scale.

3) Enforcement Is Also Different

In a walled garden:

• If Meta says someone is in your audience, they’re in everywhere.
  

In programmatic:

• The same audience can be interpreted differently depending on:
  
  
  • DSP
    
  • SSP
    
  • Publisher
    
  • Region
    
  • Privacy settings
    
  • Inventory type
    

So even if upstream data is good, delivery varies a lot.

This is why people feel like programmatic “doesn’t work.”
It’s not necessarily bad data — it’s fragmented identity and enforcement.

4) DV360 Sits in the Middle

Platforms like Display & Video 360 benefit from Google’s identity graph on Google-owned properties.

But once you move into the broader open exchange, variance goes up.

So evaluation needs to get stricter.

Why We’re More Confident in Walled Gardens

Inside platforms like:

• LinkedIn Ads
  
• Meta Ads Manager
  
• Reddit Ads
  
• Google Ads
  

We can validate targeting much more clearly.

Because:

• Identity is deterministic
  
• Enforcement is consistent
  
• Holdout testing is easier
  
• Measurement is cleaner
  

That’s why Primer audiences are extremely repeatable there.

And why we’re more nuanced with programmatic.

Does Programmatic Work for B2B?

Yes — if you’re disciplined.

No — if you go broad and hope for magic.

Here’s what actually works.

What Works

1) Prioritize Logged-In Publishers

Look for inventory where users are authenticated.

This reduces bot risk and improves matching.

Think:

• Major news publishers
  
• Industry sites
  
• Subscription content platforms
  

2) Use Strong Identity DSPs

Not all DSPs are equal.

You want ones with:

• Real identity frameworks
  
• Strong publisher relationships
  
• Good reporting transparency
  

3) Start Narrow

Loose targeting + open inventory = noise.

Instead:

• Tight account lists
  
• ICP-focused audiences
  
• Controlled expansion
  

This is especially true in B2B where audience sizes are small.

4) Run Real Experiments

At Primer we strongly recommend controlled testing — the same mindset we use when evaluating mobile vs desktop attribution.

Use:

• Holdout groups
  
• Incrementality testing
  
• Channel comparisons
  

Don’t rely on vendor dashboards alone.

5) Measure Pipeline, Not Clicks

Programmatic looks great on CTR.

That’s not the goal.

Look at:

• Qualified leads
  
• Pipeline creation
  
• Account engagement
  
• Conversion lift
  

Especially if you’re integrating performance data across channels into something like our ClickHouse-based reporting stack, you can actually see whether programmatic audiences move pipeline.

Where We Get Skeptical

When we see:

• Wide-open inventory
  
• Loose targeting
  
• Long-tail placements
  
• Heavy reliance on “modeled intent”
  
• No holdout testing
  

You’re basically buying noise.

The identity degrades fast.
Enforcement gets messy.
You get impressions that look fine on paper… but don’t move revenue.

The Right Mental Model

Programmatic isn’t bad.

It’s just less deterministic than walled gardens.

So the right approach is:

• Confident in walled gardens
  
• Nuanced in semi-closed ecosystems like DV360
  
• Test-driven on open web
  

That’s how we approach it at Primer.

A Simple Checklist for B2B Marketers

Before spending real budget on programmatic, ask:

• Can we validate targeting in-channel?
  
• Are we using deterministic identifiers where possible?
  
• Are we prioritizing logged-in publishers?
  
• Do we have holdout tests?
  
• Are we measuring pipeline impact?
  

If not, you’re probably buying expensive noise.

Final Thought

In B2B marketing, the hard part isn’t reaching people.

It’s reaching the right 2,000 people inside a company.

Walled gardens give you deterministic targeting.
Programmatic gives you scale — with more variance.

The trick is knowing which tool you’re using.

And being honest about the tradeoffs.

If you’re trying to figure out where programmatic fits into your B2B mix, happy to share more of what we’ve seen across customers — especially how we structure experiments to actually measure audience lift.