What does programmatic buying actually mean in 2026?
Programmatic advertising is the automated buying and selling of digital ad impressions through technology platforms, using real-time auctions or pre-negotiated deal structures rather than direct insertion orders with salespeople. The infrastructure is built on three main standards from IAB Tech Lab's OpenRTB specification, which defines the real-time bidding protocol; VAST (Video Ad Serving Template), which governs how video creatives are served into players including CTV apps; and ads.txt and app-ads.txt, which are anti-fraud mechanisms that let buyers verify authorised inventory sellers.
The core participants are: the demand-side platform (DSP), which the buyer uses to set bids and targeting; the supply-side platform (SSP), which aggregates publisher inventory; and the ad exchange, which runs the auction between them. In practice, the boundary between SSP and exchange has blurred, with most major SSPs running their own auctions.
Two buying mechanisms matter for partner selection. Open exchange (also called open auction or RTB, real-time bidding) means your DSP competes in real time against all eligible buyers for each impression. Private marketplace (PMP) and programmatic guaranteed (PG) are deal-based arrangements where a publisher gives you preferred or exclusive access, often at a negotiated floor CPM (cost per thousand impressions), in exchange for a volume commitment. CTV premium inventory is predominantly PMP and PG rather than open exchange.
Advertisers in Singapore, the US, Canada, Australia, and Malaysia all have access to programmatic infrastructure. The maturity of local supply, publisher relationships, and DSP support teams differs, as covered in the five-markets section below.
How is CTV different from other programmatic inventory?
Connected TV (CTV) delivers video advertising to internet-connected television screens through streaming applications. The DSP bids for impressions served inside apps on smart TVs, streaming sticks, and set-top boxes. The auction uses the same OpenRTB protocol as display and online video, with VAST as the creative delivery standard. What differs from standard digital video is the context: the viewer is on a large screen, often in a household, and typically cannot click an ad.
The measurement infrastructure for CTV is also distinct. Because there are no cookies on television operating systems, attribution relies on IP-address matching, device graph lookups, and household identity signals rather than browser cookies. The IAB Tech Lab's Open Measurement SDK (OMSDK) provides a standard for third-party viewability and verification measurement on CTV, though publisher adoption varies by platform. View-through attribution is the primary conversion metric, and advertisers need a clear methodology to separate CTV-driven conversions from other channels before drawing conclusions.
Inventory quality varies sharply. Premium CTV (named streaming services, broadcast network apps) is mostly available via PMP or PG deals. Open-exchange CTV inventory is available but requires tighter brand safety controls, including domain/bundle whitelists and partner-level filtering. Frequency management across multiple CTV publishers is one of the operational challenges that distinguishes experienced programmatic partners from less experienced ones.
Context note
CTV is not a separate channel to "add on" to a media plan. It is a targeting and buying method for video on television screens. It coexists with linear TV in many plans but requires different infrastructure, different measurement, and a different partner conversation.
The three partner models compared
The table below maps the three models across the dimensions that matter most for a mid-market advertiser choosing a programmatic or CTV partner.
| Dimension | Agency trading desk | In-house DSP seat | AI-native consultancy |
|---|---|---|---|
| Who holds the DSP contract | The agency (you are a sub-seat) | You (direct contract with DSP) | You (direct contract, consultant operates it) |
| Data ownership | Data lives in agency's seat; portability varies by contract | Full ownership; data transfers with you | Full ownership; data transfers with you |
| Fee structure | Bundled CPM or margin on media; may include undisclosed tech fees | DSP platform fee (typically % of media) plus internal headcount | DSP platform fee plus consultancy service fee; no media markup |
| Transparency into bidding | Report-level visibility; bid-stream access uncommon | Full access to bid logic, targeting settings, and logs | Full access; consultant shares rationale for every major change |
| Speed to launch | Fast; agency has existing seat and PMP deals | Slower; DSP onboarding, contract, and team training required | Medium; consultant negotiates seat and deal access, no internal training lag |
| Access to PMP and PG deals | Strong; agencies have pre-existing publisher relationships | Builds over time; DSP platform team can assist | Dependent on consultant relationships; often competitive with agency access |
| CTV-specific capability | Variable; some trading desks have dedicated CTV practices | Available via DSP if team has CTV experience | Available if consultant has CTV-specific experience |
| Exit or internationalisation path | Switching partners means rebuilding audience data from scratch | Low lock-in; seat and data stay with you | Low lock-in; seat and data stay with you, consultant can support handoff |
| Minimum viable team size | No internal team required | Dedicated programmatic manager minimum; director-level to run strategy | One internal stakeholder for briefing and approval |
| Best fit | Large spend, multi-market, deep publisher access priority, comfortable with bundled fees | Long-term commitment to programmatic, adequate internal headcount, full control priority | Mid-market, data ownership priority, no headcount to hire, wants senior strategy without agency overhead |
A few points this table cannot capture. The quality of the people inside each model matters more than the model itself. An agency trading desk staffed by senior practitioners is better than an in-house seat run by a junior analyst on auto-optimise. An AI-native consultancy that treats AI tooling as a substitute for media judgment rather than a multiplier of it is worse than both. Model selection sets the structural conditions; it does not guarantee the outcome.
For advertisers considering media integration as a strategic goal, the in-house or AI-native consultancy model is the only path that preserves the data continuity needed to build cross-channel audience and measurement infrastructure over time.
The Trade Desk's Kokai and its Trading Modes: what buyers should know
The Trade Desk is the largest independent DSP by revenue, meaning it is not owned by a major publisher or ad network. This matters because publisher-neutrality in bidding is a consideration for advertisers who want to avoid structural conflicts of interest in their DSP's supply-side relationships. The Trade Desk's platform, Kokai, is its current buying interface, built with AI-assisted optimisation across supply path, bidding, and audience signals using its Unified ID 2.0 (UID2) identity framework.
Within Kokai, The Trade Desk offers two named configurations that buyers and their partners should understand:
- Control mode: Designed for buyers who want granular manual control over bid parameters, targeting logic, and supply path optimisation. Campaign settings are set by the buyer or their operator; the AI assists but does not override manual decisions. This mode gives the most transparency into how bids are actually placed.
- Performance mode: Applies The Trade Desk's AI optimisation end-to-end, with a bundled CPM structure rather than a separate media-plus-platform-fee breakdown. The trade-off is that the bidding logic is less transparent to the buyer, similar structurally to how Google's Performance Max works on the Google ecosystem. The system is optimising toward your declared KPI (key performance indicator), but the path it takes to get there is not fully visible.
This distinction is relevant to partner selection in a specific way. An agency trading desk or consultant operating in Control mode can show you more of how your budget is being deployed. One operating primarily in Performance mode is asking you to trust the algorithm. Neither is categorically wrong, but you should ask explicitly which mode your partner uses and what the reporting looks like in each.
Neutral reporting note
leapbuzz operates direct The Trade Desk seats for programmatic clients. This section reports the Kokai Trading Mode distinction accurately and without commercial bias. For current minimum spend requirements and feature availability in your market, refer directly to The Trade Desk's sales team or official documentation, as these change by region and partner tier.
Google's equivalent DSP is Display and Video 360 (DV360), with product details and feature documentation available via the official DV360 support centre. DV360 integrates deeply with Google Audience data, YouTube inventory, and Campaign Manager 360 reporting. For advertisers already running significant Google Ads spend, DV360 offers measurement continuity that The Trade Desk does not. Amazon DSP is a third option, differentiated by access to Amazon's first-party shopper data and by its CTV supply through its streaming properties. See also the leapbuzz post on LinkedIn CTV ads and Amazon DSP for a deeper look at the Amazon DSP buying model.
For a broader view of programmatic buying as a channel, the leapbuzz programmatic platform page covers inventory types, identity solutions, and the role of first-party data in an open-web buying strategy.
Market context: programmatic and CTV across Singapore, US, Canada, Australia, and Malaysia
The structural decision between trading desk, in-house seat, and AI-native consultancy plays out differently depending on where your audiences are. Programmatic display and online video are mature across all five markets leapbuzz works in. CTV maturity is the bigger variable.
| Market | Programmatic display maturity | CTV supply depth | DSP local support | Key considerations |
|---|---|---|---|---|
| Singapore (SG) | High; major DSPs fully operational | Growing; concentration in a small number of regional OTT platforms and broadcast apps | The Trade Desk, DV360, Amazon DSP all present | Small addressable market means CTV scale requires regional extension; PDPA first-party data rules apply |
| United States (US) | Highest globally; competitive and sophisticated | Deep; largest AVOD and FAST ecosystem; strong open-exchange and PMP CTV supply | All major DSPs; largest publisher PMP catalogue | Most complex brand safety and fraud environment; state-level privacy laws (CPRA, state equivalents) affect audience segments |
| Canada (CA) | High; closely tracks US ecosystem | Strong; broadcast network apps and streaming services well represented | Full US-equivalent DSP access; some publisher deals US-only | PIPEDA and provincial laws govern data use; French-language market requires separate creative consideration in Quebec |
| Australia (AU) | High; mature with strong local publisher presence | Growing rapidly; broadcast network catch-up apps are primary CTV supply | The Trade Desk, DV360, Amazon DSP present; local SSPs active | Premium CTV inventory concentrated in a small number of broadcasters; Privacy Act 1988 reform ongoing |
| Malaysia (MY) | Mid-maturity; growing quickly | Earlier stage; supply concentrated in regional OTT services; open-exchange CTV limited | Regional DSP coverage via APAC hubs; some local SSP supply | PDPA 2010 applies; multi-language creative (Bahasa Malaysia, English, Chinese) required for broad reach |
Across all five markets, advertisers running programmatic CTV for the first time will find that the agency trading desk model gets them to market fastest, while the in-house or AI-native consultancy model gives them the data foundation to build on. The choice of market also affects minimum viable budgets: CTV PMP floors in Singapore and Malaysia are higher on a relative basis than the US because publisher supply is thinner and deal minimums have not compressed to the same extent. Singapore-based advertisers should note that data practices across programmatic buying must align with the Personal Data Protection Act (PDPA) administered by the PDPC, particularly in how audience data derived from Singapore residents is handled and retained by DSP platforms.
For advertisers managing budgets across multiple of these markets, the retail media and CTV closed-loop post covers how measurement architecture differs when CTV is part of a retail or commerce media strategy rather than a pure brand play.
Decision framework: which model fits your situation?
Seven questions to answer before choosing a programmatic or CTV partner model. Work through them in order; the first question where you hit a hard constraint typically determines the model.
- Do you have an internal programmatic specialist? If no, and you do not plan to hire one in the next six months, in-house DSP is not viable without external help. Move to trading desk or AI-native consultancy.
- Is data portability a strategic requirement? If you are building a first-party data programme, need to retain audience segments long-term, or plan to eventually move in-house, agency trading desk creates lock-in risk. Favour direct seat models.
- What is your annual programmatic media budget? Below a threshold where DSP platform fees and management overhead are proportionate to media spend, programmatic is inefficient compared to self-serve managed platforms. If you are significantly below that threshold, consider whether programmatic is the right channel at this stage at all.
- Is CTV a primary or secondary objective? If CTV is primary, ask specifically about each partner's CTV PMP relationships, OMSDK measurement coverage, and frequency management methodology. A partner without specific CTV experience will treat CTV as display-on-a-big-screen, which is structurally wrong.
- How important is fee transparency? If you need line-by-line cost disclosure for internal governance or board reporting, a trading desk with bundled pricing requires a media transparency contract to be useful. Direct seat models make this simpler.
- What DSP ecosystem do you want to commit to? If you are already deep in Google's stack, DV360 has integration advantages. If you want publisher independence and UID2 identity infrastructure, The Trade Desk. If Amazon first-party data is central to your audience strategy, Amazon DSP. These are not mutually exclusive in the long run, but starting with one and adding others is more manageable than trying to run three simultaneously.
- What is your exit plan? Every partner model should be evaluated for what happens when you want to change. In-house and AI-native consultancy models have the cleanest exit. Trading desk exits require careful contract language upfront to protect audience data and campaign history.
Practitioner observation
The most common mistake in programmatic partner selection is choosing the model that offers the lowest friction to launch and then living with its structural constraints for years. The second most common is hiring for CTV without establishing a measurement framework first, which means the channel produces impressions but not accountability.
Related leapbuzz resources on programmatic and CTV
- Programmatic advertising platform page - inventory types, identity solutions, first-party data strategy in open-web buying.
- Retail media and CTV closed-loop measurement - measurement architecture when CTV is part of a commerce or retail media strategy.
- LinkedIn CTV ads and Amazon DSP - a deeper look at the Amazon DSP buying model and LinkedIn's CTV ad products.
- Media integration service - how leapbuzz structures cross-channel programmes with programmatic and CTV at the core.
