Signal-to-Bid Methodology

Five signal layers. Two decisions. Every impression.

Every eligible impression is evaluated across five proprietary signal layers. The composite score determines whether we bid at all — and exactly how much we bid. This is what agentic optimization actually means at the impression level.

01 →

Creative Intelligence

60+ scored dimensions per asset. Format, persuasion, attention, CTA, contextual signals.

02 →

Competitive Landscape

Next-day SOV by publisher × DMA. Whitespace and battleground identification.

03 →

Audience / Identity

Cuebiq affinity + P_Next scoring. Attain purchase propensity. Device graph resolution.

04 →

Conversion Feedback

Daily device-level conversion feeds. Suppression and real-time bid weight adjustment.

05 →

Context Harmony

Creative signal profile × content environment. Harmony Score (0–100) per impression.

What Makes This Different

Traditional CTV Stack

Broad strokes at campaign launch

Audience segments set at kickoff, static through flight
Creative intelligence not connected to media buying
Competitive data quarterly, weeks of lag
Context matching limited to FAST or not present
Conversion feedback post-campaign only
Uniform CPM across qualifying inventory

PurePlay Agentic

Impression-level scoring, recalibrated every minute

Dynamic composite scoring per device, per auction
60+ creative dimensions mapped directly to bid decisions
Next-day competitive SOV at publisher × DMA granularity
Harmony scoring across all CTV inventory types
Daily conversion feeds — suppression and bid adjustment in real time
Differentiated bid price per impression based on composite score

The data behind these signals was technically available before. What wasn't practical until now was cleansing it, organizing it, and optimizing off of it at the impression level, in real time. That is the agentic layer.

Layer 01 · PurePlay Proprietary

Creative Intelligence

60+ scored dimensions per creative asset · Benchmarked against competitive set via index scoring

Every creative asset is analyzed across eight dimension categories before the campaign launches. Each asset receives a signal profile that travels with every impression it generates — and gets matched against the content environment at bid time.

Each dimension is scored 1–10 and indexed against the competitive set: client score ÷ competitor average × 100. A score of 120 = 20% above competitive average on that dimension. This makes gaps visible — urgency, price framing, audience representation, CTA type — so media strategy can reinforce strengths and fill holes.

Dimension Detail

Format types: Narrative, Testimonial, Showcase, Demo, Lifestyle Montage, Presenter/Host, Graphics Only, Promotional Only, Manifesto/Brand Anthem.

Persuasion signals: Ease of Action, Proof, Authority, Social Norms, Guarantees/No Risk, Discounts, Urgency/Scarcity, Missed Opportunity.

Attention signals: Opening Hook, Pattern Interrupt, Direct Address, Pacing Variation, Emotional Peaks, Visual Contrast, Sound Design, Curiosity Gap, Stakes/Conflict, Surprise/Novelty.

How It Feeds the Bid

Each creative gets a signal profile. At bid time, the optimizer matches that profile against the content environment to produce a Harmony Score (Layer 5). Creatives with stronger contextual alignment earn higher bid multipliers in matching environments — and lower ones where the match is weak.

Layer 02 · PurePlay Proprietary

Competitive Landscape

ACR + set-top box data · Next-day delivery · 68+ publishers tracked

For every market in the campaign footprint, we track each competitor's share of voice on every publisher — delivered next-day, not quarterly. Competitive density determines where to bid aggressively and where to conserve spend.

Whitespace: Publishers where all competitors are below a meaningful threshold — these get aggressive bids. Battleground: Publishers where the client is within ±10% SOV of the nearest competitor — these get prioritized or defended. Saturated FAST environments where competitors dominate get conservative bids or pass.

Why This Matters for Publisher-Direct Buys

Run-of-network deals are strong for scale. But even within a single publisher buy, competitive density varies by daypart, content type, and slot position. The competitive layer adds a signal: are the impressions we're winning inside this deal actually the premium ones, or are competitors owning those? Deal curation through OpenX lets us tighten buys at the impression level without disrupting existing publisher relationships.

How It Feeds the Bid

Competitive signals determine where to bid aggressively (whitespace, low-competition premium publishers) and where to conserve (saturated environments where competitors dominate). Updated daily, so a competitor's new campaign launch changes the bid landscape the next morning.

Layer 03 · Data Partner Dependent

Audience / Identity Signals

Cuebiq (live integration) · Attain (active build) · Device graph via Experian/LiveRamp · 60–70% match rate

Device-level propensity scoring from verified third-party data partners — used upstream for pre-bid scoring, not just downstream for post-campaign measurement.

Cuebiq — Live Integration

Two scores per device, refreshed weekly:

Affinity Index (0–1): Brand loyalty based on frequency and recency of physical visitation. Device at 0.95 = strong loyalty, prime re-engagement candidate.

P_Next (0–1): Predicted probability of visiting a brand location in the next 30 days. Device at 0.35 = 35% predicted visit probability.

Composite: Affinity × P_Next weighted by campaign objective. A device at 0.95 Affinity + 0.35 P_Next earns a meaningfully higher bid multiplier than 0.70 + 0.08. Validated across 145K+ unique device IDs, 1,457 brands, 74 verticals — composite scoring successfully drove differentiated bid multipliers in live traffic.

Attain — Active Build · First-to-Market Opportunity

Attain is already validated as a measurement partner for restaurant and QSR campaigns (ShopRite: $10.94 ROAS, #1 programmatic partner). PurePlay is building the same closed-loop architecture with Attain that's been validated with Cuebiq: pre-bid propensity scoring from deterministic purchase data, with conversion feeds flowing back to continuously adjust bid weights. The first brand to activate this gets first-mover advantage on Attain-powered agentic CTV optimization at scale.

How It Feeds the Bid

High-propensity devices get higher bid multipliers. Devices that have already converted get suppressed. The identity layer is what separates reaching the right person from reaching everyone who met a basic targeting criteria.

Layer 04 · Real-Time Closed Loop

Conversion Feedback Loop

Daily device-level conversion feeds · S3 (Cuebiq) or API (Attain) · 83 learning cycles per 90-day campaign

Conversion data doesn't wait until end-of-campaign to influence buying. It flows in daily and gets applied immediately to bid weights and suppression lists — the campaign is smarter on day 30 than it was on day 1.

7x learning speed advantage over traditional batch measurement: 83 rolling daily cycles versus ~11 at a quarterly cadence. Which signal combinations are correlating with actual conversions get weighted more heavily in subsequent scoring — the system self-improves every day the campaign runs.

Suppression Logic

If a device already visited a location or completed a target action, there is no reason to keep paying to reach them. Suppression is applied automatically from daily feeds — not at the end of the measurement window. Budget saved here gets reallocated to higher-propensity devices still in the conversion funnel.

How It Feeds the Bid

Conversion data closes the loop between what the algorithm predicts and what actually happened. Signals that correlate with real conversions gain weight. Signals that don't get down-weighted. Every campaign cycle makes the next one more precise.

Layer 05 · Creative × Content Match

Context Harmony Score

Works across all CTV inventory types · Scene-level on FAST · Publisher + category level across premium

Every creative has a signal profile from Layer 1. Every impression opportunity has a content profile. The Harmony Score (0–100) measures how well this specific creative resonates with this specific content environment at the moment of the bid.

A creative scored high on energy, lifestyle relevance, and situational excitement earns a high Harmony Score in sports, action, or high-energy entertainment content — and a lower score in home renovation or news. This is impression-level context alignment, not segment-level audience targeting. It works on top of audience signals, not instead of them.

Connection to Creative Testing

If a resonance test is in-flight, PurePlay's creative scoring provides the explanatory layer: which signals in the creative are driving resonance, and in which environments. Format, persuasion mechanics, audience representation, CTA type — 60+ dimensions scored against the competitive set. When resonance results come in, the signal analysis shows exactly what to replicate, amplify, or fix.

How It Feeds the Bid

High Harmony Score impressions earn higher bid multipliers — the optimizer bids more aggressively to win slots where creative-context alignment is strongest. Low Harmony Score impressions get passed or bid conservatively, regardless of audience match. It's not enough to reach the right person. The creative has to fit the moment.

The Mechanism

From Five Signals to a Bid Decision

In any given minute, say there are 100,000 available auctions that meet a campaign's targeting criteria. Budget pacing requires winning ~1,000 impressions. At a typical 5–10% win rate, that means bidding on 10,000–20,000 auctions and passing on the other 80,000+.

What Most Platforms Do

Select which auctions to bid on more or less arbitrarily at scale
Apply CPM target uniformly across qualifying inventory
Frequency caps and brand safety rules handle the rest
Audience segments set at campaign launch, static through flight
Optimization happens at end-of-flight or weekly review

What PurePlay Does

Rank all 100,000 impressions by composite signal score
Select the top-ranked impressions to bid on — the highest signal concentration
Set a differentiated bid price per impression based on composite score magnitude
Recalibrate the ranked list every minute based on pacing and incoming signals
Apply daily conversion data to adjust which signals get weighted most heavily

Two Decisions Per Auction

Whether to bid at all

Only impressions that cross the composite score threshold get a bid. Many impressions that technically meet targeting criteria — right audience, right geo, right publisher — still get passed because the combined signal score doesn't justify the spend. This is where budget efficiency is protected.

How much to bid

Within the bidding set, higher-scored devices earn more aggressive bids — increasing win rate precisely where signal strength is highest. Not every impression we bid on gets the same CPM. The bid price is a function of the composite score magnitude, not a flat target.

Agentic vs. Traditional: Capability Comparison

Capability	Traditional CTV Stack	PurePlay Agentic
Audience targeting	Static segments, set at campaign launch	Dynamic composite scoring, recalibrated per minute
Creative intelligence	Not connected to media buying	60+ dimensions scored and mapped to bid decisions
Competitive landscape	Quarterly reports, weeks of lag	Next-day, DMA × publisher granularity
Context matching	FAST-only or none	All CTV inventory, creative-to-content Harmony Score
Conversion feedback	Post-campaign measurement	Daily suppression + real-time bid adjustment
Bid optimization	CPM goals, frequency caps	Impression-level propensity scoring, differentiated bid per device
Learning cycles (90-day campaign)	~11 batch cycles	~83 rolling daily cycles — 7x faster

Engagement Options

Self-Service or Managed — Both Run the Full Signal Stack

The engagement model determines where bid execution happens. Both options run the complete five-layer signal architecture. The difference is whether PurePlay's Augmentor controls the DSP-side bid decision, or whether the client's preferred DSP executes against PurePlay-curated, signal-enriched supply.

Self-Service via OpenX Curation

You buy through your DSP.
We curate what's available.

PMG executes through preferred DSP (The Trade Desk, DV360, or others)
PurePlay curates deals through OpenX — cherry-picks exact publishers, individual deals per pub
PurePlay's enrichment service scores every impression in real time: segment activation, EID injection, deal floor overrides
Composite scoring influences the supply side before any DSP sees the impression
Existing publisher-direct relationships stay intact — PurePlay adds the intelligence layer underneath
PurePlay covers Cuebiq + Attain data costs

One constraint: Impression-level bid price differentiation via Augmentor is a managed-service capability. In self-service, PurePlay optimizes on the supply side; the DSP handles demand-side execution.

Managed Service via Augmentor

PurePlay controls bidding end-to-end.

PurePlay's custom bidder (Augmentor) deploys inside the DSP environment
Both supply-side (OpenX enrichment) and demand-side (Augmentor bid function) are optimized simultaneously
Full composite score → differentiated bid price per device, per impression
Phase-aware behavior: Discovery → Calibration → Optimization → End-of-Flight
Complete pacing control, margin zone management, converter suppression
PurePlay manages inventory sourcing alongside the buy

Capability Comparison

Capability	Self-Service	Managed
All five signal layers	✓	✓
Publisher deal curation (cherry-picked)	✓ Full control	✓ Full control
Creative intelligence scoring	✓	✓
Competitive SOV data (next-day)	✓	✓
Cuebiq / Attain propensity scoring	✓	✓
Conversion suppression (daily)	✓ Via deal floor overrides	✓ Direct bid suppression
Context Harmony scoring	✓	✓
Per-impression bid price differentiation	Supply-side signals only	✓ Full (Augmentor)
DSP used	Client's preferred DSP	PurePlay Augmentor

Your Existing Stack

🔗

Attain

Your measurement partner becomes a pre-bid signal

Attain is already trusted for attribution on this business. PurePlay's architecture uses that same Attain data upstream — to score devices before bidding, not just measure results after. The closed-loop architecture (Attain propensity → pre-bid scoring → conversion feedback → bid weight adjustment) is in active build. The first brand to activate this gets first-mover advantage on Attain-powered agentic CTV at scale.

🎬

Innovid DCO

PurePlay identifies the gaps. Innovid executes the mix.

PurePlay's creative intelligence layer identifies which dimensions are winning or losing against competitors — urgency, price framing, audience representation, CTA type. That analysis informs which creative variations Innovid deploys where. Complementary workflow, not a replacement.

📊

Resonance Testing

PurePlay explains why resonance happens

A resonance test shows which audiences respond to which creative assets. PurePlay's creative scoring answers the follow-on question: which signals in the creative are driving that resonance? When results come in, the signal analysis shows exactly what to replicate and what to fix — 60+ dimensions against the competitive set.

📡

Publisher-Direct Deals

Intelligence on top of your existing supply strategy

Run-of-network buys with major publishers are strong for scale. PurePlay's competitive data answers: within that RON buy, is delivery actually premium or defaulting to FAST inventory? Competitive SOV at the publisher + DMA level shows whether competitors are winning the premium slots inside those deals — and deal curation through OpenX tightens the buys at the impression level without disrupting existing relationships.

Partner	Role	Status	Validated Outcome
Cuebiq	Foot traffic attribution + pre-bid scoring	Live	+17% foot traffic lift · differentiated bid multipliers in live traffic
Attain	Purchase attribution + pre-bid scoring (building)	Active Build	$10.94 ROAS (ShopRite) as measurement partner · pre-bid architecture in development
Experian / LiveRamp	Device graph identity resolution	Live	60–70% MAID-to-CTV match rate on Experian graph