Most premium CTV inventory in Southeast Asia is still bought on an insertion order. The agency agrees a plan with the publisher's sales team, signs, sends creative and waits for the post-campaign report. Programmatic CTV grows next to it, through private marketplaces, programmatic guaranteed deals and the open auction. For a planner, the real question is which part of the budget goes where. That depends on 4 things: the inventory the plan needs, the budget per market, what you need to measure, and how much operational work your team can carry.
When the insertion order wins
Some inventory is only sold direct. Live sport, homescreen placements on a TV maker's interface, show sponsorships, the first or last slot in a break: publishers keep these off the exchange because they price them as packages, not as single impressions. If the plan depends on them, the insertion order is the route.
It also wins on certainty. An insertion order names the publisher, the content, the placement, the flight and the rate. You know what ran and where. The terms are written down too: what counts as delivery, what happens on a shortfall, who proposes the make-good. For a brand new to CTV in a market, that clarity is worth paying for.
The cost is time. Every publisher has its own rate card, forms, minimums and report format, and a regional plan across 4 markets can mean a dozen email threads. Publishers feel it as well. A recent AdExchanger column describes CTV sales teams that over-protect guaranteed deals to avoid make-goods and ration premium inventory, because manual work caps how many deals they can handle. Mid-sized budgets are usually the ones that fall through.
When programmatic wins
Programmatic is better at reach across many publishers at once: one platform, one frequency cap, one audience definition applied across apps and devices. If the objective is a defined audience at a controlled frequency, wherever it watches, one platform is simpler than reconciling 6 direct deals.
It lowers the entry bar. Private marketplace deals usually carry lower minimums than a publisher's insertion order, so a brand can test CTV in a market with a budget a sales team would not prioritise. And you can change course mid-flight: move weight between deals, tighten the audience, pause what underdelivers.
Two cautions. First, the open auction carries real fraud risk in CTV. Ad fraud researcher Augustine Fou puts it plainly: buying from real sellers carries almost no fraud, while a CTV bid request is trivial to fabricate with no device involved at all. When viewing is concentrated in a handful of services, a large volume of cheap impressions from the long tail of apps deserves questions. Second, programmatic TV ads tend to be plain 15 or 30 second spots, because auction specifications limit length and structure. Sponsorships and linked formats rarely travel that way.
Compare effective cost, not headline CPM
A programmatic CPM is not what the publisher receives, and often not all you pay. Between the two sit a buying platform fee, a selling platform fee, data costs and sometimes verification and curation fees, each a percentage or a CPM add-on. On an insertion order, the rate is usually close to the whole media cost.
For each option, write down:
- the media CPM the publisher actually receives
- every fee added on the buy side and the sell side
- the share of impressions you expect to exclude as invalid or off-target
- the people cost: trading time on programmatic, negotiation and reconciliation time on insertion orders
Divide the total by the impressions you would accept as valid. That is your effective CPM. Rates and fee stacks vary widely by market and format, so run it on your own numbers. Say an open auction line comes in at a low CPM: once fees are stacked and a share of impressions is excluded after the flight, it can cost more per valid impression than a direct deal at a higher rate card.
The hybrid most experienced buyers run
In practice, most experienced CTV buyers use both. Insertion orders carry the tentpoles: sport, launches, sponsorships, the placements that make a brand look like it belongs on television. Programmatic carries always-on reach against defined audiences, with private marketplaces and programmatic guaranteed preferred over the open auction.
The cost of the hybrid is coordination. Frequency is managed in 2 places, reports arrive in 2 shapes, and nobody sees total household reach without stitching the data. Plan for that work before you commit to it.
A third route: agent-to-agent deals
There is now a way to keep insertion order terms without the manual back and forth. On AdCP, an open protocol for advertising agents, a buying agent sends one brief to the selling agents of every publisher that matches. Each selling agent answers with a plan at the publisher's own prices and within its rules: placements, flight, rates, delivery terms. A trader compares the plans and confirms one. On the publisher side, a person or the rules the publisher has set approve what is offered. The deal reads like an insertion order, and no auction sits in the middle.
AdExchanger reported on Pierre Fabre buying US CTV through agents partly because, with a smaller budget, it was "not really interesting" to many larger buying platforms. The same piece mentions specific buys that auctions handle badly, such as the first or last slot of every break in live sport.
Be clear about the limits. It is early. Only some publishers run selling agents, coverage differs a lot by market, and it does not replace programmatic for reach across thousands of apps. It suits mid-sized direct deals that are too small for a sales team and too specific for an auction. Our guides on direct, programmatic and agentic buying and CTV insertion order terms go further.
A decision framework
Start with insertion orders if you are new to CTV in the market, the plan depends on specific content or placements (sport, sponsorship, homescreen), brand safety has to be named rather than inferred, and the budget per publisher clears its minimum.
Start with programmatic, private marketplace or programmatic guaranteed first, if the objective is audience reach and frequency across many apps, you need to change the plan mid-flight, or the budget per market is below direct minimums. Stay out of the open auction until you can verify by named app where impressions ran.
Run both when each market has the budget and your team has the people to manage 2 workflows and reconcile reach between them.
Try agent-to-agent deals when the publishers you want have selling agents, the deal is mid-sized, and you want insertion order terms (named placements, fixed rates, written delivery terms) without weeks of email.
Where agents fit
Skopa, the buying agent from Oderra, sends the brief to every matching selling agent and returns the plans side by side, and a person confirms before anything runs. Fanera answers for publishers at their own prices. Neither replaces a programmatic setup where one already works.