There are two ways to get the attention of a senator. The familiar way is to buy a great many impressions in the senator's state and hope that some fraction of them lands somewhere useful. The less familiar way is to identify the people the senator already listens to - staff, donors, advisers, friends, community leaders - and make sure those people keep encountering the same issue. The first method treats influence as acreage. The second treats it as a neighborhood.
Unearth, a public affairs technology company founded in Sacramento in 2011, has built its most distinctive product around the neighborhood. Atlas Influence Targeting constructs digital advertising audiences from the personal, professional and political networks surrounding a decision maker. Unearth says an individual Atlas audience is typically about 2,000 people. The paid message travels through that circle; the hoped-for result is an unpaid conversation.
This is a peculiar business, in the best sense. Unearth is not simply an agency with clever targeting, and it is not simply a software vendor renting seats. Its researchers assemble audiences. Its strategists plan campaigns. Its designers make creative. Its media buyers deliver and optimize ads. The technology holds the method together. Clients can buy Atlas as a fully managed campaign priced by policymaker or license its data quarterly into their own ad accounts.
The useful contradictionNational campaigns built from very small rooms
Unearth serves the institutional cast of American advocacy: government-relations teams, associations, nonprofits, law firms, general consultants, ballot-measure committees and public affairs shops. It says it has run work in 47 states and powered hundreds of advocacy and ballot-measure campaigns. Its customer page names organizations and advisers across agriculture, business, education, energy, construction and telecommunications. The company is national in reach, but the unit of work can be a single person who sits on a committee.
Figures are published by Unearth. The company does not publish the methodology behind its win-rate calculation.
Atlas is the sharp end of a five-product system. Scout listens, adding more than 100 million online conversations a day and asking human analysts to supply context that automated sentiment scores miss. Atlas finds the circle worth reaching. Optic turns forbidding research into interactive, mobile-ready arguments. Portal recruits constituents and controls the release of their emails, social posts and calls. Impact buys and adjusts media across social platforms, streaming television and the web.
Portal contains the suite's most mischievous feature: a throttle. Most advocacy forms send a message the moment a supporter clicks. Portal can collect messages, hold them and accelerate delivery near a hearing or vote. It can also drip them steadily. The software recognizes that one hundred emails are not always one hundred units of pressure. Timing changes their meaning.
The first thing that brokeThe glamorous product had an unglamorous data problem
Atlas began with a problem familiar to anyone who has tried to identify real people across imperfect databases. Unearth gathered employment and professional information from multiple places. Each source arrived in its own shape. Engineers spent time standardizing records and resolving whether two similar-looking entries described the same human being. The audience model might be sophisticated; the plumbing was still plumbing.
Unearth first used People Data Labs' Person Enrichment API to append information to records it already held. Trust in that data led it to use the partner's Person Search API more broadly. The search that once crossed several sources could be centralized. According to the partner's case study, the change filled gaps in employment history, increased the number and quality of matches, and made proofs of concept quicker to stand up and maintain.
The lesson is wonderfully ordinary: the first failure in a data product is often not the model. It is the handoff between sources. Unearth changed its approach when one source proved reliable enough to become a practical source of truth. Other companies can copy the sequence - enrich a known set, test the quality, earn trust, then consolidate - without copying the political use case.
What the precision costsLess media, more judgment
Unearth does not publish dollar prices. It does publish the shape of the bill: full-service Atlas work scales by policymaker; data licenses are sold quarterly. In a one-month model based on July 2022 California media averages, the company claimed Atlas required 89 percent less campaign-setup time and 92 percent less media spending than traditional targeting. The precise budget behind the comparison remains gated, so the percentages are best read as Unearth's sales argument, not a universal benchmark.
The saving makes intuitive sense. A campaign buying access to 2,000 relevant connections should need fewer impressions than one buying an entire legislative district. Yet the cheap part is the media, not necessarily the thinking. Someone must define the target, research the network, choose a credible messenger, shape an issue-specific argument, handle data lawfully and decide when to increase pressure. Unearth has turned those judgments into a mix of software and service rather than pretending the dashboard can make them alone.
The part worth borrowingFind the decider before buying the audience
Unearth calls its preference “small data.” The phrase is a useful provocation. Advertising platforms trained marketers to begin with available audience categories and then search for relevance inside them. Unearth begins with the desired decision and works backward: Who can make it? Whom do they trust? Which of those people can be reached? What message could plausibly move through the group? The audience is the consequence of the strategy, not its starting menu.
Its own advice is just as concrete. Ground the creative in research. Use a messenger the network considers credible. Name the target when a direct approach is appropriate. Increase frequency around a consequential meeting. Watch opposition messaging. Combine online engagement reports with offline follow-up. None of this requires proprietary software to understand. It requires the discipline to resist a million cheap impressions when ten consequential conversations would be better.
Precision is a method, not a miracle
The approach weakens when there is no identifiable decision maker, when the surrounding network cannot be mapped or reached, when the creative lacks a credible messenger, or when a campaign needs broad public awareness rather than concentrated influence. It also depends on responsible, compliant use of identity-linked data. A perfect map cannot rescue an unpersuasive message.
The company itself grew by a similar logic. Co-founders Bryan Merica and Matthew Eagan began with consultancy work in public affairs and candidate communications. Repeated campaign needs became products. Human experience was not discarded; it was encoded into a suite and surrounded by customer success, research and media teams. Unearth's three cultural phrases - Out in Front, More Reps and Fire in the Belly - sound like they were written on the wall of a campaign office because, essentially, they were.
There is a temptation in political technology to describe every new tool as a machine that changes minds. Unearth's more interesting claim is smaller. Machines can find a network, watch a conversation and control delivery. People still decide which facts matter, which messenger belongs and when a message has become pressure. The campaign around the campaign works because politics was a social graph long before anyone gave it that name.