What is a DSP, and why would you need one if Google and Meta already work well for you? A DSP (demand-side platform) is buying technology that connects your campaigns to participating media owners and technology providers across the open internet. If your current setup works, a DSP should add something useful alongside it, not replace it. This guide therefore compares DSP advertising with walled garden advertising, so you can see where a demand-side platform fits and where it does not.

Key takeaways

• The choice is about buying models and capabilities, not about choosing one company over another.
• A walled garden keeps data, inventory, buying tools and reporting in one system, while a DSP buys across many media owners.
• The extra choice a DSP offers comes with extra operating work.
• Look for a specific gap first, and ask whether changing your existing setup would close it.
• Agree each activity’s job and success measures before launch, not after the first review.

What is a DSP, and how is it different from a walled garden?

A walled garden is a closed platform that combines much of the audience data, inventory, buying tools and reporting within one system. An open-internet DSP, in contrast, connects buying across participating media owners and technology providers. So the comparison is about two buying models, rather than about one company against another.

The line is not always sharp, however. Some companies offer both kinds of product, and some advertising products combine elements of both. For a wider introduction to who buys and who sells, read our guide on how programmatic advertising works. The IAB Tech Lab programmatic standards give background on auction and guaranteed programmatic transactions.

Closed platforms and open-internet buying compared

The differences below are typical, not fixed. Exact capabilities depend on the product, market, contract and campaign setup.

Walled garden advertising

With walled garden advertising, your ads run within the inventory and partner network available through that product. You get strong platform data, plus your own permitted data under the platform’s rules. Optimisation and frequency management work on the activity, signals and recognised audience within that product.

You inspect the reports and data the platform makes available, and independent checks come through approved integrations and measurement methods. The platform also manages much of the inventory system, although you still choose suitable settings. As a result, reporting is relatively integrated within the platform.

DSP advertising on the open internet

With DSP advertising, your ads can run across participating publishers and media owners, subject to access. You can use permitted advertiser data, contextual signals and external segments that the setup supports. In addition, optimisation covers the activity across all inventory you buy through the DSP.

Frequency management works across supported DSP inventory, with limits from identity and environment coverage. You can potentially see more detail on placements and buying paths, although access varies. Verification and research integrations exist too, with varying coverage and methods. On the other hand, you or your partner actively manage publishers, quality settings and buying paths, and several reporting sources may need to be compared and explained.

The extra choice through a demand-side platform therefore comes with extra operating work. Include that cost when you assess whether the choice matters to your campaign.

What do walled gardens do well?

In walled garden advertising, large platforms can combine logged-in audience information, established formats and optimisation within systems that marketing teams know well. As a result, they make it relatively straightforward to run campaigns against defined conversion signals.

Their own advertising environments often provide valuable scale and audience information that you cannot reproduce elsewhere. For many advertisers, these strengths justify keeping walled garden advertising at the centre of the plan.

Where platform reporting stops

Their reporting still has a boundary, though. A platform’s reach figure does not automatically tell you how many of those people also saw ads through another platform. Adding separate reach figures can count the same people more than once.

Independent measurement inside a platform also depends on the integration. Ask what the provider observes directly, what data it receives and which inventory it covers. After all, a verification partner does not make every measure fully independent or comparable.

Do you need a demand-side platform if you use Google and Meta?

Not automatically. First, look for a specific gap that your current setup leaves open. At MINT Square, we often encounter three issues in planning discussions. However, treat them as observations to check against your business, rather than findings that apply to every advertiser.

Media choices have never been evaluated

Investment has followed the channels already in the account. Instead, list the environments your audience plausibly uses and identify which ones you have assessed on evidence. An unevaluated environment is a question to investigate, not an automatic opportunity.

Audience overlap is unclear

Each platform reports its own reach and frequency, but the team has no combined view. A simple sum cannot solve this. So check the scope and definitions of the reports before you commission more measurement.

Customer data has limited use in advertising

The business holds useful information, but the media plan does not use it, or uses it only in one platform. In that case, establish whether broader use would be lawful, practical and valuable.

None of these issues alone proves that DSP advertising is the answer. Ask first whether changing the existing setup would close the gap. Our guide on whether programmatic suits your business gives you the assessment.

How should you compare DSP advertising with Search?

Judge each activity on its job. Imagine Search captures people requesting a quote, while an audio campaign bought through a DSP introduces the brand to potential buyers who are not ready to enquire. Comparing only their last-click CPAs (cost per acquisition) ignores the audio campaign’s purpose.

Search may also receive credit for a conversion after someone has met the brand elsewhere. That possibility does not prove the audio campaign caused the conversion. Instead, it means the review needs evidence suited to both jobs.

So agree each activity’s purpose and success measures before launch. Then ask the stakeholder who approves the budget to accept the comparison you will use. Our article on measuring delivery, response and outcomes explains how to assess each level.

Where two campaigns genuinely have the same job and comparable measurement, a cost or conversion comparison can be useful. In other words, programmatic does not need protection from a fair comparison; it needs a clear one.

In practice: the comparison that arrives in the review

A common pattern looks like this. A first programmatic campaign, bought through a demand-side platform, goes live alongside existing Search and social. Four weeks later, somebody puts the CPAs side by side in a monthly review. Programmatic loses, usually by a wide margin, and the team cuts or stops the activity before it has produced anything readable.

The first question is whether the activities had the same job. You should not judge a campaign that introduces the brand to future buyers solely on the acquisition metric of a campaign that captures existing intent. If both genuinely aim at comparable acquisition outcomes, however, a CPA comparison may be useful, provided audience, attribution and conversion definitions are compatible.

The problem is agreeing that distinction after the numbers have arrived, because it can sound like a defence of a disappointing result. Therefore, record the objective, the accepted evidence and the meaningful comparisons before launch. Share that agreement with the person who chairs the review as well as the team operating the campaign.

The report should still hold each activity accountable. Separate objectives prevent a misleading comparison, but they do not excuse weak delivery, unsuitable exposure or a failure to fulfil the agreed role.

Advice for agencies proposing DSP advertising

Begin with what the client’s current activity does well. Then identify a gap using their own data and objectives, and explain how an additional capability, such as a demand-side platform, might address it and what work it would add.

Be prepared to recommend improving the existing campaigns instead. For example, a useful assessment can conclude that the audience is sufficiently covered, the budget is too small for another test, or the required creative is unavailable. That discussion is easier to evaluate than a general recommendation to “add programmatic”, because it gives the client a problem, an option and a way to judge that option.

Above all, do not open by criticising what is working. A client whose Search and social programme delivers has good reason to defend it. Start from what the current mix does well and reach the boundary from inside it. A useful question is: “Which audience, moment or capability is missing from the current plan, and what evidence would make closing that gap worthwhile?”

FAQ

Where to go from here

Google and Meta can stay at the centre of your plan. The real question is whether a specific gap justifies DSP advertising as another buying arrangement, and what evidence would show it.

Our playbook Getting Started with Programmatic Advertising in 2027 walks you through that assessment, from finding the gap to judging a first campaign fairly.