Last week, we covered how streaming ads are bought using Demand-Side Platforms (DSPs). DSPs allow political advertisers to serve ads to voters in their target audience while they’re watching streaming content.
Our goal is to efficiently reach voters who will impact the election. Using one DSP alone cannot accomplish this.
To maximize reach, political buyers must use multiple high-quality DSPs and measure deduplicated reach and frequency across them.
How do we target voters?
Voter data is public. Not who you voted for, but when you’ve voted, how often you vote, and party registration (in applicable states). That voter file is the foundation of political targeting and more sophisticated modeling.
But voter files contain offline information: names, home addresses, and sometimes phone numbers. Streaming ads require online identifiers: cookies, IP addresses, mobile IDs, and IFAs.
To bridge that gap, we use identity graphs. Identity graphs match offline voter records to online devices, enabling ads to be delivered when those voters are streaming content.
How do DSPs work?
Very simply:
The advertiser uploads a target audience to the DSP, often via a match partner such as LiveRamp.
The DSP matches that audience to online identifiers using its identity graph.
When a matched device is streaming content, the DSP bids on the available ad slot.
The advertiser with the winning bid serves their ad.
Most DSPs have access to similar streaming inventory. The key difference is not supply. It’s match quality.
Why do Identity Graphs matter?
Every DSP uses its own identity graph to match voters to devices. That’s where performance diverges.
Most matching combines two approaches:
Deterministic: Direct matches using email or subscriber-level data.
Probabilistic: Modeled matches based on IP, device signals, and location patterns.
The weighting matters. The more a DSP leans on loose probabilistic modeling, the more impressions drift outside your intended audience.
How do political advertisers maximize reach?
There are three levers:
Maximize on-target rate. Choose DSPs with strong deterministic matching and disciplined probabilistic methodology.
Diversify identity graphs. Different DSPs reach different pockets of voters. No single graph covers everyone.
Control deduplicated frequency. Measure reach and frequency across platforms to prevent waste and unlock incremental scale.
#3 is the hardest, but most important piece of the puzzle. Some buyers who can't measure across DSPs opt to only use 1, sacrificing reach. Other buyers use multiple DSPs, but if they can't measure frequency across them, they risk over-saturation.
Case Study: Incremental Reach
At CSM, we deploy three trusted DSPs on every campaign. Our measurement solution unifies the separate DSP impression-level data to show a single combined frequency and optimize for incremental reach.
Each additional DSP delivers incremental reach with limited duplication.
Campaign: Statewide Primary
CTV Targeted Reach: 57%
No one DSP reached more than 34%, and each added 8-10% incremental reach.
What does this mean for political advertisers?
Do not rely on one DSP - No matter how strong it is, one identity graph cannot cover the full universe of target voters. You will cap your reach.
Audit on-target reach - Some DSPs expand audiences within their ID graph, serving impressions to voters outside your intended audience. Validate match quality and hold partners accountable.
Monitor Reach and Frequency across DSPs - Use impression-level data to measure true exposure, eliminate waste, and reallocate budget toward incremental reach during the campaign.