To Build Or Buy An In-House Programmatic Stack? Do Both

tony-ralphBrand Aware” explores the data-driven digital ad ecosystem from the marketer’s point of view.

Today’s column is by Tony Ralph, director of advertising technology at Netflix.

Over the last year, I’ve spoken with at least a dozen representatives from brands both large and small that are looking for advice on first steps as they consider getting more hands-on with their programmatic buying practice.

My initial feedback: This move is not for everyone. My colleague Mike Zeman nicely summarized the criteria to consider before moving forward with bringing programmatic in-house. For those who still feel the timing and circumstances are right, the conversation invariably turns to the balance between choosing platform partners and the internal development efforts required.

In short, what should they “buy” and what is advisable to “build” as they construct their in-house programmatic stacks?

This “build or buy” scenario often applies to ad tech platforms as they evolve toward “full stack” offerings. The choice of whether to pursue an acquisition or initiate internal development to deliver the next component to complete a full suite of serving, bidding and data platforms can be a genuine consideration for those with the dollars and engineering horsepower. As the ecosystem evolves and advertisers are empowered to get closer to the programmatic machinations, they now confront downstream variations of these “build or buy” decisions as they relate to constructing their own internal digital marketing tech stacks.

‘Build And Buy

There are a few big brands whose volume justifies building their own demand-side platforms or data-management platforms. For most advertisers, however, the “build vs. buy” proposition is a false dilemma that can artificially constrain their options.

Rather than an advertiser forcing its thinking into either buying or building for any given component, a prudent path is often to pick the right vendors that allow them to evolve their programmatic practice organically, by building in conjunction with the platform. Rather than “build or buy,” they should “build and buy” by choosing a vendor solution that allows immediate execution while simultaneously enabling iterative development “on top of” the platform over time.

A “build and buy” approach allows advertisers to jumpstart their in-house buying efforts in the short term. As their teams’ buying expertise evolves, they can tailor the tools to meet their advertiser-specific needs. They get both immediate scale and long-term flexibility.

Workflow enhancements, data visualization catered to the business and custom reporting might represent a first wave of enhancement developed internally. A second wave might include using first-party data to influence bidded campaigns, development of an internal data mart or enhanced user-tracking efforts. The prioritization obviously varies with business requirements, development resources and overall state of programmatic evolution.

Vendor Criteria For ‘Build And Buy

The primary step with the “build and buy” approach is picking vendor solutions that liberate advertisers to pursue a sliding scale of end states, because they may not know how far they will take their development efforts. Good development practice dictates that they design extensible solutions that could, in theory, be reused if they later switch platform partners. Inevitably, though, many integrations or features developed will be vendor-specific. This makes it essential for the advertisers to pick partners that can grow with them.

Given the number of players in the ecosystem and the dynamically shifting landscape, how can advertisers ensure that they are picking the best partners for themselves?

The foundational component to look for in a platform that will be built against is a full-featured API. No infrastructure solution will fit perfectly “off the rack.” A comprehensive API gives vendors the autonomy to merge solutions gracefully with an advertiser’s operational nuance. One essential question to ask vendors: Do they builds their apps using their own APIs? If they eat their own dogfood, so to speak, they likely treat their APIs as a first-class feature. If all elements of the vendor UI have corresponding API calls, an advertiser has full liberty to build nearly any solution, up to and including a complete interface outside the vendor solution designed specifically for its workflow.

A second essential feature to allow full build opportunity is a first-class event-level data offering. When pursuing programmatic tactics, teams will inevitably have questions that can only be answered with detailed data for each impression, click and conversion. There should be systematic methods to harvest the data, ideally via API, and something approaching a formal data dictionary describing the schema and associated fields. If these don’t exist, beware.

Aside from these specific features, advertisers should try to get a sense of if they and their prospective vendor are philosophically aligned.

A great litmus test is determining where the vendor sits on the continuum from managed service to self-service. The “build and buy” approach is much more compatible with a self-service mentality, illustrated by vendors that prioritize features that will empower clients to solve their own problems and perform their own optimization. This is different from a platform that is fundamentally a managed service and simply provides UI access upon request. Self-service should be at the vendor’s core. A great question to ask a vendor here: What percentage of its current clients is self-service? And can it provide references?

Because long-term success hinges on evolving past a vendor-client relationship and toward a true partnership, it is useful to understand what the vendor is building in the short, mid- and long term. This may reveal co-development opportunities when priorities are shared and afford the ability to influence when there is misalignment. Sharing at this level also permits “counter-development” – when knowing a vendor’s priorities gives an advertiser the luxury of pursuing soft spots with buyer-developed solutions. This level of sharing takes time and trust. Picking a partner that understands and aligns with the advertiser’s programmatic vision is a useful start.

For advertisers, it also may be worth applying these criteria to the state of their own businesses. If these types of endeavors are not within the current scope of their marketing or engineering organizations, they might question whether an in-house programmatic push is for them. If these do strike a chord, an iterative “build and buy” approach may be a logical path toward their programmatic goals.

Follow Netflix (@Netflix) and AdExchanger (@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. I think a “programmatic stack” view is very short-term, whether it’s buy or build.

    Everything is converging and going “programmatic” so a long view (and ultimately most efficient and effective approach) should be considered for a “marketing stack”. Unlike today’s “managed service” environment where people are used to “connect the stacks”, in the future no component will sit or operate independently from other components.

    You need data to drive smart segmentation, message and content creation. You need ad serving and verification, which operates most efficiently when integrated server side. You might want a complete customer picture by pulling in CRM data for holistic customer management across all channels. You want to ingest analog and digital delivery and sales data to make effective media allocation recommendations. And you need device ID synced across all sources to ensure a complete, persistent and accurate customer view.

    And what about work flow and redundancy between systems.

    The “bolt” solution outlined here allows for flexibility and accommodation to the future looking, tech-savvy organizations. But we know there are many all-in-one solutions in the market which might be the ideal alternative for some. Build or Bolt is not as easy or simple as it sounds.

  2. Tony Ralph

    Keith, the summation of your comment (“not as easy as it sounds”) and the first point made in the piece (“this strategy is not for everyone”) are largely aligned.

    Is building an in-house programmatic solution as described a short-term approach? Absolutely. That is precisely the point. For the advertiser in the right situation, it provides a near-term option allowing internal teams to own activation while leaving open the option of ongoing engineering investment.

    Programmatic technology convergence is an intriguing topic with aspects of inevitability. While prognosticators debate timelines and outcomes, the piece outlines a plan that can work in the ambiguity of today and does not preclude success in a possible world where the “stack” evolves to be monolithic.