2026年のプログラマティック広告：スケールする購入の実践ガイド

Programmatic advertising now accounts for approximately 90% of global digital display ad spend. The promise was transparency, efficiency, and targeting precision at scale. The reality, in 2026, is more complicated: the average programmatic dollar sees 40 to 45 cents reach a real human user in a brand-safe environment. The rest is consumed by intermediary tech fees, made-for-advertising site inventory, ad fraud, and viewability failures. The brands that win in programmatic are not those with the largest budgets — they are those with the tightest supply chain, the most precise audience definitions, and the most rigorous quality controls. This guide covers the specific practices that move working media above 65% of total programmatic spend.

The supply chain problem: where budget disappears

A programmatic ad impression passes through an average of 4.7 intermediaries between the advertiser's DSP and the publisher's ad server. Each intermediary takes a margin — typically 10 to 20% of the impression value. At 4.7 hops with average 15% margin per hop, approximately 50% of the advertiser's bid is consumed before the publisher receives anything. This is the tech tax. Supply path optimisation (SPO) is the practice of reducing the number of intermediaries by establishing preferred paths to high-quality publishers. In practice: identify your top 20 publisher domains by performance; establish direct deals or preferred SSP paths to those publishers through your DSP; and deprioritise or exclude paths that pass through more than 2 intermediaries. SPO typically recovers 15 to 20% of budget currently lost to unnecessary intermediation.

Made-for-advertising sites: the inventory category that wastes the most budget

Made-for-advertising (MFA) sites are websites built specifically to generate programmatic ad revenue — not to provide genuine content to genuine audiences. They typically feature scraped or thin AI-generated content, extreme ad density (40 to 60 ads per page view), and bot-inflated traffic. The Association of National Advertisers estimated in 2024 that MFA sites account for 11% of all programmatic impressions. The challenge is that MFA sites score well on basic brand safety tools: they are not on blocklists, they do not serve offensive content, and their traffic passes basic bot detection. The solution is inclusion-list-based buying: build a curated list of domains where you are willing to advertise, and exclude everything else. For a mid-market brand, a curated inclusion list of 500 to 1,000 high-quality publisher domains eliminates MFA exposure entirely.

Targeting without third-party cookies: the 2026 reality

Third-party cookie deprecation in Chrome became complete in 2026. The behavioural targeting that programmatic was built on — following users across sites based on their browsing history — is no longer available at scale. The two replacement approaches that are working in our programmatic client portfolio: contextual targeting (placing ads in content contextually related to the product — a sports nutrition ad in fitness content, a travel insurance ad in travel planning content) using semantic contextual engines like Peer39 or Oracle Contextual Intelligence; and first-party audience activation (matching your CRM or loyalty programme email list to publisher and DSP audiences, creating lookalike audiences from your first-party converters). Contextual targeting typically delivers 60 to 80% of the targeting precision of third-party cookie audiences. First-party audience activation typically delivers 85 to 95% of third-party cookie precision — but requires a CRM database of sufficient size (typically 50,000+ records for meaningful match rates).

• Implement supply path optimisation — audit your top publisher paths and eliminate intermediaries with more than 2 hops.
• Build a curated inclusion list of 500 to 1,000 publisher domains — do not buy open exchange without an inclusion list.
• Activate your first-party CRM audience in your DSP — match your customer list to available audiences.
• Test contextual targeting as a third-party cookie replacement — segment by content category relevant to each product line.

Private Marketplace deals: brand-safe scale at predictable CPMs

Private Marketplace (PMP) deals are invite-only programmatic buying arrangements between a brand and a specific publisher or publisher group. Unlike open exchange buying, PMP deals give you: guaranteed brand-safe inventory (you know exactly which sites your ads will appear on), priority access to premium inventory that is not available on the open exchange, transparent CPMs without hidden intermediary fees, and consistent delivery for brand-awareness campaigns. The appropriate use of PMP vs. open exchange varies by campaign objective: use PMP for upper-funnel brand awareness campaigns where environment quality and brand safety matter most; use curated open exchange (with an inclusion list) for lower-funnel retargeting and performance campaigns where scale and cost efficiency matter most.

DSP selection: the decision that shapes everything

The Demand-Side Platform (DSP) is the technology layer through which all programmatic buying flows. DSP choice has a larger impact on working media efficiency than most advertisers realise. The key selection criteria in 2026: inventory quality controls (does the DSP have robust MFA detection and pre-bid brand safety filtering?), supply path efficiency (how many hops does the DSP use to reach premium publishers?), first-party data activation capability (can you activate your CRM audiences without data leaving your environment?), contextual targeting sophistication (what semantic contextual technology does the DSP use?), and reporting transparency (can you see impression-level domain data?). The three DSPs that consistently score highest across these criteria in our independent audits are The Trade Desk, DV360, and Amazon DSP — each with different strengths by campaign type and inventory access.