Programmatic vs. direct: when both belong in the mix — defending an 80/20 split

The question keeps recurring in advertiser meetings: programmatic or direct? Over the last two years our answer has settled at an 80/20 split — and that is not a taste preference, it is the result of three months of parallel measurement across 11 campaigns.

How we use each

Programmatic for us: open-auction DSP buying with an invitation-only PMP layer on top of premium inventory. RTB, contextual targeting, frequency control. Average CPM: 380 HUF. Average reach: 4.2M unique users per week.

Direct for us: hand-written IO, flat fee for a specific landing page or newsletter, brand-safe slot guaranteed. Average CPM: 1,100 HUF (roughly 3x programmatic). Average reach: 180,000 unique users per week per slot.

At first glance the numbers say programmatic is 12x cheaper. That is misleading — they measure different things.

The three axes that separate them

Brand fit: direct wins. In the first week of a brand campaign, when the message is still being refined, a direct slot adjacent to human-curated editorial delivered 2.3x higher time-on-page for the same creative versus programmatic. The reader's context matters.

Scale: programmatic wins. An 8-day retargeting campaign on direct IO would have required 47 separate owner negotiations. RTB: two hours. That is it.

Measurability: tie. Programmatic gives us every post-impression event, but brand-lift measurement is noisy. Direct gives cleaner brand-lift surveys (we know exactly who was exposed) but slower behavioral signals.

The logic of the 80/20

Our current default: 80% of advertiser spend programmatic, 20% direct. This is not dogma — the split depends on the campaign objective.

Brand launch (new product, new creative): 50/50. The direct weeks let the message harden; the programmatic scales it once it has matured.

Always-on awareness: 90/10 in favour of programmatic. Here frequency control and cheap reach dominate; direct only on the high-attention slots (homepage takeover, newsletter sponsorship).

Performance / lower funnel: 95/5 in favour of programmatic. Direct is essentially a brand marker — your logo present in case someone is really looking — but conversion comes from programmatic retargeting segments.

B2B / niche: this flips. 30/70 in favour of direct. The B2B audience is so small that on programmatic the frequency cap exhausts the target by lunchtime, while you pay more per CPM than direct would cost.

Two cases that hardened the rule

Two specific campaigns from the last quarter. Case A (FMCG launch, 18M HUF): 50/50 split, 8 weeks. The direct arm produced a 6.1% brand-recall lift; the programmatic arm produced 38% higher assisted conversion. Neither would have been enough alone — direct built the memory, programmatic harvested it.

Case B (B2B SaaS, 4M HUF, 90-day sales cycle): started 80/20 programmatic-direct. After six weeks we flipped to 30/70 because the B2B audience (a 1,200-account named ABM list) was so narrow that programmatic CPMs ran up to 4,200 HUF. On direct, in a targeted industry newsletter, 1,800 HUF. End of campaign: 41 MQL, 11 SQL, 2.4x cost efficiency.

What not to do

Do not run the two in isolation under two different agency teams. We worked that way in 2024 and the programmatic frequency cap had no knowledge of direct impressions. A user saw the same creative 14 times in a week. Today both run on top of a shared frequency-management layer — that is the only non-negotiable.

Takeaway

Programmatic and direct are not competitors; they are two different functions in the same media mix. The 80/20 is our default, not our law. Three questions at the start of every campaign: is it brand-fit sensitive? Is it scale-hungry? How narrow is the audience? The three answers spell out the split.