inside programmatic budget allocation

justauser

I've spent years on the publisher side, and honestly, the buy-side can feel like a black box 🤔. After chatting with a few agency colleagues, here's what I've pieced together about how budgets are actually split.

First off, it varies hugely by advertiser maturity. some brands run direct IO and programmatic as completely separate worlds - different teams, different budgets, different KPIs. Others are much more fluid, optimising holistically across channels. So there's no one-size-fits-all.

When it comes to programmatic channels, I rarely see fixed percentage splits between PMPs and open auction. it's almost always driven by performance goals, brand safety requirements, scale needs, and client preferences. One agency person told me they keep budgets fluid so they can react and push spend toward ad groups that are performing well - clients rarely care about the split unless they're running CTV or audio, which demand heavier justification.

They also mentioned that PMPs are typically set up only for specific contextual alignment - think key tentpole events or premium inventory where you don't want to fiddle with optimisations. meanwhile, CTV leans heavily toward PG/PMP.

so the short version: it's not siloed unless the client is siloed, and PMP vs open auction is a performance decision, not a fixed rule. Hope that helps anyone else trying to decode the buy-side! 😊

ryanseo

From what I've seen over the years, there's no one-size-fits-all answer here - it really comes down to how mature the advertiser's programmatic setup is, and how much internal alignment exists between teams.

Some brands still run direct IO and programmatic as two completely separate worlds. Different teams, separate budgets, conflicting KPIs. You'll see the display team chasing reach on a guaranteed basis while the programmatic team is trying to optimise for performance on the open exchange - and no one talks to each other. On the other end, you've got advertisers that are much more fluid, shifting budget between guaranteed and biddable based on real-time performance signals. That's where you see true cross-channel optimisation.

On the PMP vs Open Auction split, fixed percentages are rare in my experience. It's nearly always driven by a mix of:

Performance goals (CPA, ROAS)
Brand safety requirements (PMPs for premium inventory)
Scale needs (open auction for volume)
Client preferences (some want full control, others let it run)

Where it gets interesting is CTV. That channel leans almost entirely toward PG and PMP deals from what I've observed. Open auction CTV inventory is still thin on quality, so the budgets follow the guaranteed pipes.

budgetburner

Fluid budgets across programmatic channels are the only sane way to run things, honestly. Let's you shift spend to whatever ad group is actually performing without getting bogged down in a fixed split. Clients don't give a toss about the channel breakdown until you want to fire up CTV or Audio - then suddenly every pound needs a bloody justification letter.

As for PMPs? Waste of time unless you need tight contextual alignment for a major event or premium inventory where you don't want to babysit optimisations every five minutes. Otherwise, open auction does the job fine.

socialbutterfly

Oh, this is such a good breakdown - and honestly, it matches a lot of what I've seen negotiating deals on the influencer side. The three-tier advertiser split is spot on. 👏

From my own experience bouncing between in-house, agency, and now publisher-adjacent roles over the last decade, here's what I'd add:

🚀 The "Ignorant Unicorn" - Yep, we see these too. They want the flashy influencer activation, no questions asked, budget is just there to be burnt. IO is basically a blank cheque for them.
📺 The Old Guard - The linear TV lovers who bundle CTV into an IO because it's easier. They don't care about real-time optimisation, they just want the reach to match their TV numbers. And honestly, they're fine with it being ₈₀% of the revenue.
🧠 The Savvy Ones - These are the ones who treat every dollar like it's on fire. IO is a last resort, they want full stop-start control. In influencer marketing, these are the brands that ask for performance clauses before they even look at the proposal.

The publisher-side truth you mentioned about programmatic vs IO really resonates. if you can secure IO revenue, do it - but if the market doesn't support it, you're forced into programmatic. Leaving inventory dark is just leaving money on the floor.

On the pricing side, it's a constant A/B test. I've had to play with floor prices and different deal types in GAM because our finance system is welded to it - no wiggle room for a fancy CTV ad server. The experiment function is a lifesaver for testing hypotheses, even if it's clunky.

There's no one-size-fits-all, but knowing which bucket your client falls into makes the negotiation a whole lot easier.

24fire

The frameworks are broadly similar-historical performance, market sizing, saturation curves, incrementality. But the reality for most agencies is messier. They blur direct and programmatic because the margins demand it. Channel allocation often comes down to what the data has proven: you put 60-80 % into the known winners (Meta and Google), then carve out 10-20 % for something like CTV. That range feels about right. I've been running CTV experiments through one of the newer programmatic partners for a few clients now, and it's become an acceptable slice. The real trick is resisting the urge to over-optimise the proven channels and starve the test budget before it can show anything. Took me a while to learn that.

brandvoice

Yes, that's exactly what I've seen working alongside programmatic teams. Budgets get siloed by team or buying type, especially at larger agencies. There's rarely a fixed PG/PMP vs open auction split - it all depends on the client's KPIs, pricing, trust in the publisher, and how easy the inventory is to activate. Understandably confusing from the outside, but it's genuinely tailored each time.

prpro

Yes and no. Budget allocation in programmatic is fluid and performance-driven, but advertisers often lock in fixed budgets for specific partners or contractual obligations.

That said, if your team can't see the value in being present across all channels programmatically, it's a lost cause. Chances are they still don't fully grasp what programmatic actually does. Once they do, they'll be asking why you aren't already working on the very thing you suggested.

nightowl

The split between IO and programmatic budgets is fascinating because it mirrors the tension between fixed and dynamic allocation that I see every day in Amazon PPC. On-platform spend is fluid by nature-you can't know exactly where the algorithm will distribute impressions until the campaign runs. But clients want predictability, so agencies build these rigid accounting walls that don't reflect how the real-time bidding actually works.

I've spent the better part of a decade watching programmatic share grow steadily, especially among global accounts. The real headache comes when a client brings their own tech stack into the mix-suddenly the DSP isn't just a buying tool, it becomes this opaque middle layer that screws up every forecast model you try to build.

For publishers, it's never going to be perfectly predictable, but the trend line is clear. Treat it like a volatile but essential channel, not a stable revenue stream. You don't cut Amazon ads just because conversion rates dip for a week-likewise, you don't pull programmatic because one month's delivery was lumpy. The macro direction outweighs the micro noise.

metricsmuse

Honestly, the split tends to shift depending on how mature the client is and how much reporting pressure they put on. Bigger clients lock into PG/PMP as soon as they find inventory that keeps conversion quality stable. Everyone talks about wanting scale from open auction until the invalid traffic reports land on their desk.

emailwiz

Had a client who needed absolute brand safety for a premium publisher campaign - PG deal, no question. But for their performance retargeting, open auction gave us the flexibility to scale cheaply. The real kicker is internal silos. IO and programmatic teams end up competing for the same budget pool without even talking to each other. Drives me nuts how much efficiency gets wasted that way.

ryanseo

that lines up with what I've seen on the ground. Keeping budgets fluid rather than locked into channel splits is the only way to let performance dictate spend, but it takes a client who actually trusts the data. Most still want to see "X% on display" even when the numbers scream otherwise.

the PMP point is spot on. From my side of the table, a lot of PMPs are built for operational convenience-cleaner reporting, fewer seats, easier IO management-rather than because that specific inventory drives better outcomes. A colleague once admitted their PMP was just a way to avoid the open exchange mess, not a performance lever. useful to hear that buyers do use them for contextual alignment too, it gives those deals more justification beyond just "control."

CTV feels like the last bastion of curated supply. every buyer I've spoken to still wants to know exactly which apps, devices, and content categories are in the pool before they'll commit budget. The open marketplace just hasn't earned that trust yet

justauser

thanks for breaking that down!!! so in practice, if an advertiser hands you a fixed budget, you'd set up campaigns in a DSP with GEO and audience targeting, plus a domain allowlist/blocklist. But for PMP deals, would you typically spin up a separate campaign linked to multiple PMPs? or can you manage it within one campaign? Just trying to wrap my head around the workflow 🙏

ryanseo

Interesting point about the siloing. In my experience, it's more organisational inertia than genuine strategic reasoning. Most agencies inherit legacy structures where IO and programmatic sit under separate P&Ls, so budgets get ring-fenced for reporting convenience rather than performance efficiency. There's rarely a clear strategic case for keeping them independent - it usually comes down to which director owns the revenue line.

budgetburner

Everyone acts like PMPs are some dark art. They're not. I'd set it up as a separate ad group within the campaign. Brief the PMP partner on targeting and objectives - they push the deal to my partner seat in the DSP. Then I'd run that ad group as part of a prospecting or brand campaign for display, native, or CTV. That's it. Nothing more to it

socialbutterfly

Great questions - I love that you're digging into the mechanics properly. it's rare to see someone actually care about how money flows, not just the vanity metrics.

from my own time dealing with DSPs, here's the nasty truth most people don't want to say out loud:

🔹 There's always an oversell first, clean up later mentality.
When you're the DSP rep, your job is to pull budget through your pipe. Your team's safety depends on it. So budget allocation stops being about "right vs. wrong" and becomes "more is always better." It's not malicious - it's just survival.

🔹 Inventory access is rarely the real blocker.
SSPs have already done the hard work convincing publishers to join their "premium marketplace." Publishers smell money, so they'll let your DSP buy whatever inventory you actually need. If you're after brand-safe premium stuff, any half-decent DSP will have pre-built lists ready.

🔹 For performance campaigns? Go wide as hell.
You want maximum sample size to let the algorithm learn. The only upfront filtering should be obvious fraud-heavy domains. Start broad, pile your targeting on top, then trim the dead weight during optimisation.

🔹 Do grill them on their lists and deals.
Ask what curated lists they already have, push for always-on private marketplace deals, get yourself onto their "frequently used" placements. Fair warning: they might fob you off if you're not worth their time.

and on a side note - Good old SSU. i was one of the first to turn it on in my market years ago. It absolutely murdered my SSP and ad network revenue. Killed it. switched it off and never had the guts to flick it back on. Never looked back.

24fire

Publisher splits aren't typically a KPI on their own - they're more of a steering mechanism. The real north star is usually CPA, ROAS, reach, viewability, attention, or conversions. In a DSP you can attach multiple PMP or PG deals and let the spend gravitate toward whatever's performing best, but you still need guardrails. If a particular publisher is strategically vital - guaranteed inventory, a client mandate, or a premium test - then you might enforce minimums or carve out separate line items. It's the difference between letting the algorithm drive and deliberately steering through a few key intersections.

justauser

That makes a lot of sense, thank you! So just to check my understanding - would you set up a single campaign in your DSP that's linked to multiple deals from different publishers? Meaning the delivery would automatically spread across them?