Not Every Demand-Side Platform (DSP) Is Created Equal: What Is A True DSP?

Data-Driven Thinking“Data Driven Thinking” is a column written by members of the media community and containing fresh ideas on the digital revolution in media.

Today’s column is written by Nat Turner, CEO, at Invite Media.

If you asked a group of people in the display industry to pick the biggest theme for display in 2010, I would be surprised if it wasn’t the concept of real-time bidding and the advent of the so-called “demand-side platform” (DSP).  It seems that there’s a new “DSP” launching every week these days, each with its own twist or differentiation.  Over the last several years, I’ve met or spoken with a number of agencies and marketers on the topic.

The good news is that many of these buyers have started to realize the massive potential impact that a DSP technology can bring to the table.  The bad news, however, is that their decision-making process for figuring out what this all means has become extremely muddled and complex (and is only getting worse).  What I’ve realized is that very few people really know what a “true DSP” is or how to evaluate one (and to no fault of their own), or even how to tell one DSP apart from another.  Any company can claim anything on their website; any company can build a nice PowerPoint deck or tell a great story about real-time bidding or rocket science optimization.  At the end of the day, what really matters is that the agency makes the right decision for their client(s) and chooses a well-evaluated and thoughtful path in order to take advantage of this tectonic shift happening in the industry.

From this, one of the main “ask’s” leaving meetings these days has become, “What are the key characteristics of a true DSP?”  Instead of writing up a different response for each meeting, I instead tried to create a broad list by collecting a bunch of thoughts at our company, ideas from people in the industry, and direct quotes from agencies that have gone through this evaluation process.  In a follow-up article, I will also list out a number of questions that should be asked in a RFP process to evaluate potential DSP partners.

Key characteristics that define a true DSP:

  • The DSP must provide a fully self-service interface.  Clients should be able to have complete control via the interface and build an expertise around its use.
  • If the DSP provides managed services help to the agency, the DSP should be using the same interface that the agency would be using.  The technology should not require any manual work behind the scenes to activate or “set live” a change or a campaign.
  • The DSP must remain neutral and have zero allegiances to any publishers, exchanges, data providers or other vendors.  A true DSP should embody the word “platform” and not just be conduit or pretty interface to a pre-existing business.
  • The DSP must be fully transparent, starting with pricing and fees.  All fees that the DSP earns should be exposed in the interface, and every penny that the DSP makes should be known and visible to the agency.
  • The DSP should not mark-up media cost without the agency knowing.
  • The DSP should not mark-up data cost without the agency knowing.
  • If the DSP works with a publisher directly, it should be in an effort to make that publisher’s inventory “biddable.”  The DSP should not earn any additional margin from revenue sharing with the publisher or arbitraging the inventory.
  • The DSP must allow the agency to use its own exchange seats.  This allows the agency to always have visibility into the exact cost of media to ensure the DSP is not taking any additional margin.
  • Just like media, the DSP must allow the agency to buy or negotiate data cost directly, but flow through a common integration.  Data should be treated like media, it’s another part of the “supply side” that is purchased by the agency and thus should be transparent in cost.
  • The DSP should not, under any circumstances, own or operate an ad network.  This is in direct conflict with the neutrality aspect.
  • The DSP’s goal should be to expose any feature or tool that a supply source provides (either ad exchange or data provider) and not to try and obfuscate/hide or re-brand certain components (ex. “DSP Auto Segment #1”, with no transparency into what that means).  If a supply source provides it, the DSP should expose it for the agency self-service and let the agency decide whether to use it or not.
  • The DSP should not “bulk buy” media in order to re-sell to its clients.  This could either be a function of another way to make margin, a lacking of technology, or a combination of the two.
  • Related to the above, the DSP should not take on media risk.  Every impression should be purchased on behalf of a platform user at that time based on an active campaign that that platform user has created targeting that impression.

I’m sure I’m missing a few things here, and I’m sure people could disagree with a few of the points, but I believe this represents a reasonable list from my conversations in the industry.  One thing I will say is that just because a company or offering does not match all of these points doesn’t necessarily exclude them from being a “demand-side platform.”  This list is purely focused on the concept of what is a “true DSP”, which is just one type of offering available in the ecosystem.  At the end of the day, the real question that the agency or media buyer faces is what kind of solution they’re looking for.  This will guide what questions they should ask, what types of offerings and features they should value, and generally how they should run an evaluation process.

Follow Nat Turner (@natsturner), Invite Media (@invitemedia) and (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. This is a great list, and it’s certainly thought-provoking. That said, I think your “true DSP” is like describing eTrade as the “true asset manager”.

    The qualifications you list serve a specific type of buyer: completely self-directed and so confident in their own expertise that help and simplicity just get in the way. But I think there’s probably as varied a set of needs as there is in the financial universe: lots of people will select hedge funds and ETFs over the DIY approach.

    • Tim – couldn’t agree more. As I allude to at the end, all of this highly depends on what the buyer is ultimately looking for. Ad networks or full-service/black box DSP’s aren’t going away because there will still be agencies who require/need that type of service or don’t require to be in the details. Fair point

  2. Nat – good post. You raise some interesting issues.

    I’m glad to say I hold many of the same values you provide above, especially in terms of transparency around costs and markups on data (which I believe should be zero), serving etc.

    As with any buzzwords, DSP is going to mean different things to different people. This will also evolve over time. My personal opinion is that the “DSP” is just a temporary term that is being used as a shortcut to encompass a segment of companies and services focused primarily on the buyer side and not on the seller side.

    What you describe above is a particular type of technology service provider that adds a serving, access and optimization layer on top of exchanges and aggregated inventory and data sources, and essentially allows quicker and more efficient access to the superset of features they have. Some companies are looking to simplify the media buying process and take away some of the dials and levers that are often inefficiently operated, or are adding display as a component to search campaigns. Other companies who are called (or call themselves) DSPs are likely to themselves evolve into agencies or ad networks over time, or (probably) ad- and data-serving providers.

    If you remember, some early adserving companies split their technology and service/ad network businesses because of the conflicts that could and would arise and these then took divergent paths. You may see this occur again with some of the current players; and no doubt some things that are fancy platform features will get baked into other display technology subcategories (like what we today call “adserving” – and which now includes data-aware serving).

    To say what a “true DSP” should have is a bit silly, because there’s no canonical definition thereof, people are just fitting themselves into a broad category that will surely not much longer have any meaningful value either to describe companies or services (while sometimes synonymous, this is also a useful distinction) and besides which it is very early in the evolution of any kind of demand automation.

  3. Hi Nat,
    This is a very interesting post as we, as an ad network have been struggling to make some sense in all these new technological platforms and their relationships and evolution.
    In continuation to what Rob wrote, I see the “adserving” as just one piece of the puzzle, where DSPs and exchanges are additional pieces which can be separated or integrated in one product. In this case, a “true DSP” (if it can even be defined) is actually a platform that is able to efficiently buy large amounts of media in various platforms and in various buying methods (mainly offline and real-time). The exact billing format and transparency you described and listed are just good business practices and in no means a part of a definition of a “true DSP”. A true DSP should refrain from being involved in the actual bidding process, thus negating all conflicts of interest (even if he owns media) or excess need of transparency. If the price is good for the advertiser (or agency or ad network) he should buy the impression regardless of who makes how much money in the process.

    To me a true DSP is measured by its capability to:
    1. Buy large amounts of media (reach)
    2. Buy this media efficiently (how much media can one media buyer buy)
    3. Facilitate an intelligent selection and pricing of media to buy

  4. Nat, This is really a statement of your company’s beliefs and how you compare them with what you perceive to be the practices of competitors.

    At just 2.5 years old, your industry segment has been pressured to grow up too fast and hurry up and solve all the problems of inventory glut and inefficient buying. Pushers include investors, suitors, pundits and networks and tech providers searching for new business models.

    It’s too soon to stifle innovation by trying to impose standards. With over a dozen competitors calling themselves DSPs, and with buyers adopting formal RFP processes for vendor selection, it’s late for any one provider – even an early market entrant – to dictate what’s “true.” Your marketplace of agencies and advertisers will define the boundaries, if not make the rules, and you are free to ignore or follow them.

  5. Robert Smith

    A lot of interesting comments. I agree with the previous post that pointed out that the DSP space is only 2.5 yrs old. I think it is even younger than that since that it was only a few months ago that the DSP label stuck (I heard RTB, Media Trading, etc.).

    I was surprised that you failed to mention anything about performance. This truly is a statement of your Company’s beliefs (as mentioned in a previous post).

    Most of the new DSP entrants (or those calling themselves DSPs) state to have the exact selling points and beliefs that you do. Boiled down, I view many of the DSPs on the market today as middleware – giving advertisers and agencies the ability to buy across exchanges. You are solving a technical hurdle which others have solved as well. But the exchanges are motivated to have folks hook into them, which makes it a basic technical problem at best.

    The key point that should be the first bullet on your list should be about performance. Sure – anyone can build integration into exchanges, but can the platform help me reach or exceed my campaign goals? THAT is the key. Performance should be the major reason for selecting a DSP (whether it is a brand campaign or performance campaign). If I am focused on performance, then I shouldn’t care if it is delivered as a SaaS platform or if it is delivered as a managed service. The question should be – will I get a higher conversion of my ad dollars spent using one DSP over another, and also compare those results to going direct or using a traditional ad network.

    Sounds like you are running into stiff competition to post an article like this. Competition is good. Natural selection should point to the best performer.

  6. Couldn’t agree more, Robert. At the end of the day, performance and efficiency are what will define the leader in this space.

  7. Bennett – good points all. We are very early in this space and things will take a while to coalesce.

    Robert – it is all about performance and at the end of the day, the model/revshare/whatever doesn’t matter in absolute terms at all, it is just the advertiser’s performance relative to goals that does. Period.

  8. Great sales pitch. Also contradicts some of the other recent posts that a DSP must have RTB (and ADD) who makes all this acronyms up?

  9. If Kafka’s rumors of your acquisition by Google are true, then by your own definition (specifically point #3), Invite Media is no longer a DSP, right? Or does your logic change when the acquisition price is above a certain amount?

    Just askin’…

  10. gfairnak

    Can you tell me if I am missing something here?

    A DSP receive complete URLs from exchanges
    DSPs are matching cookies from URL info they recieve with cookies buyers possess to target aduiences

    Here is an example… If I wanted to target My guess would be that the DSP knows that the page the ad is being served on.
    Is this true? Can DSPs match complete URLs/webpages with buyers white list of URLs webpages?

    thanks for the input