VoteHub and Its Founders: From Election Enthusiasts to a Professional Data Company
VoteHub combines capabilities that are often scattered across separate organizations: election-data collection, mapping, statistical modeling, political analysis, interactive development, and institutional partnerships.
My assessment is that it is evolving from a public-facing election publication into a data business serving readers, news organizations, and prediction markets. Its subscriptions and publicly confirmed relationships with Kalshi and CBS News support that interpretation.
The key distinction when studying its founder is this: Garrett Herrin initiated the venture and organizes its business, while its technical, research, and product achievements belong to a specialized team.
27、Garrett Herrin: Confirmed background and current role
Herrin comes from coastal Georgia, graduated from the University of Georgia, and is now based in Los Angeles. He publicly identifies himself as VoteHub’s CEO and co-founder.
His responsibilities include strategy, operations, sales, and partnerships.
His organizational role is therefore to assemble people capable of producing strong election analysis, turn their work into sustainable products, and connect those products with customers and partners.
That is an interpretation of his responsibilities and the company’s trajectory. It does not imply that he personally developed every model or program.
28、How Herrin established a visible industry presence
One clear early milestone was November 4, 2024, when he appeared on Forbes Newsroom as VoteHub’s CEO and publisher to discuss early voting and differences between battleground states.
The interview covered states including Nevada and Pennsylvania, emphasizing differences in participation rather than treating every state as the same story.
This shows that, during VoteHub’s founding year, Herrin was already acting as a public interpreter of election data.
He also appeared among political journalist Chris Cillizza’s election-night interview guests in 2025. That documents participation in a professional media network; it does not establish company revenue or scale.
29、The founding sequence: Collaboration preceded incorporation
VoteHub emerged from the Election Twitter community on X.
According to its official history, Herrin created the account on April 22, 2024, collaborating with Austin Park and Varun Vishwanath on precinct mapping. Luke Wines joined in July, and the company incorporated on October 7.
The sequence matters: public work demonstrated capabilities, collaboration developed, and formal organization followed.
As a business mechanism, the community simultaneously supported recruiting, portfolio-building, and initial distribution. That can reduce the cost of finding collaborators, but continued growth requires turning personal enthusiasm into dependable delivery.
30、Luke Wines: Turning editorial capabilities into products
Wines is a co-founder and Head of Product. He graduated from California State University, Long Beach, with majors in political science and economics and a journalism minor.
His work on the website, polling API, and CivicTracker spans political information, visualization, and software products.
His storytelling skills have external recognition: he placed fifth in a 2022–23 Hearst multimedia narrative storytelling competition, receiving a $1,000 award.
In 2025, he publicly described redesigning VoteHub’s polling page and developing its API, identifying journalists, researchers, and the public as intended users.
The significance is that an article can be read, while a product can be repeatedly queried, integrated, and used in a workflow.
31、Austin Park and Varun Vishwanath: Research and engineering
Park leads research and editorial work. Vishwanath oversees technology and election-night reporting systems and is a Georgia Tech graduate.
Park’s public profile identifies mathematics and political science studies at Stanford, alongside campus media involvement.
Their functions address different reliability problems.
Research determines which evidence to trust, how to interpret it, and how to handle bias. Engineering determines whether data arrives promptly and whether maps and results keep updating.
Analytical correctness and operational reliability are both essential. The latter becomes particularly important when the audience expands from social-media readers to national news organizations.
32、Zachary Donnini: Academic training, election experience, and distribution
Donnini graduated from Yale with majors in Statistics & Data Science and Economics, then pursued a data science master’s degree at Harvard.
He previously worked in data science and race calling at Decision Desk HQ and publishes Data and Divergence.
In his 2025 retrospective, he recalled being fascinated by televised election results during the 2008 presidential election at age five. He described 2025 as a transition from Yale and DDHQ to Harvard and VoteHub.
He reported approximately 60 million impressions and 2 million engagements that year. Those are self-reported personal content metrics, not VoteHub’s unique visitors or paying subscribers.
His contribution combines research, practical election experience, and an existing audience.
33、A flagship project: The 2024 presidential precinct map
VoteHub’s national map presents results below the state and county levels and supports comparisons across election years and demographic estimates.
Its credits identify state agencies, local contributors, and geographic resources including VEST and Redistricting Data Hub.
The underlying difficulty is fragmentation: American election results are distributed across jurisdictions with inconsistent boundaries, formats, names, and release schedules.
Much of the commercial value lies in integration work. Users avoid having to collect, reconcile, and clean numerous datasets themselves. The map is the visible interface; maintaining the underlying information creates continuing costs.
34、Demographic analysis: Turning research into a public tool
Donnini applied his research to national precinct-level demographic vote estimates. The associated methodological paper is titled Multi-Scale Overlap Averaging for National Ecological Inference.
Ecological inference estimates group behavior from aggregate geographic data. It does not reveal how a particular individual voted.
The map acknowledges Yale’s Shiro Kuriwaki, Dave Bradlee and Alec Ramsay of Dave’s Redistricting App, and advice on aggregation bias from Jesse Richardson and Nate Cohn.
These connections demonstrate the role of academic guidance, public data, and professional exchange. Acknowledgment or methodological assistance does not establish investment, ownership, or employment.
35、Polling aggregation: Selection rules create value
VoteHub’s January 17, 2026 methodology considers recency, pollster quality, repeated releases, house effects, and sample populations.
For example, likely-voter samples take priority over registered-voter samples, which take priority over adult samples. Frequent releases from one organization do not automatically receive proportionately greater influence.
An aggregator is therefore not simply transporting numbers. Inclusion decisions, weighting, and treatment of duplicate information shape the trends readers see.
Transparent methods and consistently applied rules are part of the product itself.
36、Pollster Scorecards: Evaluating the producers of information
VoteHub’s scorecards display accuracy, relative error, house effects, and herding-related measures.
The project draws on FiveThirtyEight archives and Split Ticket’s 2024 averages, credits methodological support from Lakshya Jain, and identifies Ethan Singer’s Umpire Scorecards as an inspiration
This moves VoteHub beyond reporting what polls say toward assessing which polls deserve greater weight.
That evaluative role adds responsibility. Assessing historical performance and verifying authenticity are different tasks. A lack of poor historical results does not prove that a new survey actually took place.
37、The 2026 forecast: Combining information into probabilities
VoteHub’s methodology distinguishes two specifications.
The Legacy model uses fundamentals and polling. The Complete model adds expert ratings and prediction-market information.
Validation holds out entire election years, while simulations account for correlated errors across races.
These choices address two practical problems: a future election cycle can differ from previous cycles, and a national polling error can affect many races simultaneously.
Nathaniel Rakich’s October 2026 guide identifies this as VoteHub’s first election forecast while noting Donnini’s prior DDHQ experience. Individual experience matters, but it does not substitute for the new organization’s own long-term forecasting record.
38、Early-vote tracking: Understanding the counting process
On September 23, 2026, VoteHub launched an early-vote tracker with L2, combining ballot requests, returns, and early in-person voting with voter-file modeling.
The information helps estimate electorate composition and understand which ballots remain outstanding on election night.
Three layers must remain distinct: whether someone voted is a recorded event; their party or demographic characteristics may be recorded or modeled; their candidate preference is usually an inference.
Early-vote analysis helps explain why the current leader may not ultimately win. Ignoring voting methods and reporting order can lead readers to misinterpret even detailed dashboards.
39、CivicTracker: Extending election products into daily political work
Created by Wines and labeled a VoteHub project, CivicTracker covers presidential actions, legislation, the Supreme Court, and officials’ social-media activity.
Its changelog records an expansion from POTUS Tracker in July 2025. The public Pro page lists $15 per month, including alerts, historical social data, and filtering.
Commercially, the distinction matters: elections create seasonal peaks, while policy monitoring can be a daily need for journalists, researchers, and public-affairs professionals.
The product shifts the relationship from occasional news consumption toward ongoing monitoring, giving users a workflow-based reason to subscribe.
40、MapLab, campaign tracking, and APIs
MapLab lets users adjust turnout or demographic support assumptions and observe changes in electoral maps—for example, among college graduates, rural voters, or Hispanic voters in South Texas.
VoteHub also provides campaign, endorsement, and redistricting trackers.
Its polling API allows external software to retrieve structured information. Early beta documentation specifies CC BY 4.0 licensing, while another current API document includes API-key authentication. Access conditions should be checked for each interface rather than generalized across all services.
Together, these products serve simulation, monitoring, and integration needs.
41、The business model: Multiple revenue entry points
Subscriptions. VoteHub Pro launched on May 4, 2026, at $9.99 monthly or $74.99 annually, offering detailed forecasts, early-vote dashboards, and deeper analysis. These are launch prices, not a claim that pricing remained unchanged.
Advertising. The privacy policy explicitly mentions Google AdSense.
Tool subscriptions. CivicTracker has separate Pro pricing.
Institutional partnerships and data services. Kalshi and CBS demonstrate institutional adoption, although announcements do not establish contract values or revenue-sharing terms.
Reader contributions. The main website includes a tipping option.
The overall structure combines public information, professional functionality, and institutional use. Free material supports discovery and trust; deeper tools support recurring payments; partnerships extend distribution.
42、Kalshi: From on-site integration to external data supply
On September 17, 2025, VoteHub announced a Kalshi partnership integrating selected political markets into its polling dashboards.
Reuters reported in July 2026 that Kalshi’s midterms hub included VoteHub polling averages alongside fundraising and historical information. An October Kalshi announcement also referenced VoteHub’s Race Intel data.
The relationship therefore expanded beyond displaying a partner’s information on VoteHub’s own pages: VoteHub became a data source inside the partner’s product.
Embedding in a customer’s workflow can be more valuable than one-time exposure. These records establish product cooperation, however, not Kalshi ownership or investment in VoteHub.
43、CBS News: A significant form of institutional recognition
On September 17, 2026, CBS formally announced a VoteHub partnership covering election data, maps, polling aggregation, and analysis across its midterm coverage.
Its October 7 announcement for Election 2026: The Nerd Show explicitly included VoteHub analysis and detailed mapping.
CBS elections and surveys executive Anthony Salvanto highlighted VoteHub’s state- and precinct-level perspectives and mapping capabilities.
This goes beyond a major outlet citing an isolated chart: it establishes participation in a formal reporting workflow.
However, November 3 election night had not occurred by the research cutoff. The partnership was confirmed; its eventual delivery performance could not yet be treated as an accomplished result.
44、What assets has VoteHub accumulated?
Four categories are useful from a business perspective.
Products: websites, interfaces, maps, simulators, monitoring, and alerts that enable repeated use.
Data-processing capabilities: geographic matching, historical-results preparation, polling standardization, and updating. Some underlying information comes from public agencies or partners, so it cannot all be classified as exclusive proprietary data.
Relationships and distribution: institutional partnerships, contributors, media connections, and subscribers.
Trust: methodological transparency, correction practices, and users’ willingness to rely on the service at important moments.
This is an analytical classification, not an audited valuation. Company resources should not be equated with Herrin’s personal property.
45、The resource network: Academia, communities, and commercial partners
The confirmed relationships reveal a complementary structure.
Academic training and research contribute methods. Election enthusiasts and local contributors add information and expertise. L2 supports voter-file analysis. Kalshi supplies a product context. CBS provides a national reporting channel.
These are different relationships and should not all be described as financial backing.
Advice from an academic does not establish university investment; adoption by a media company does not establish equity ownership.
The observed growth is better explained by assembling expertise, processing fragmented information, and gaining institutional adoption.
46、Achievements: Organization and product development matter more than follower counts
In May 2026, Herrin said the team would enter summer with 14 members, including five full-time employees. That was a statement about the anticipated summer team, not a verified October headcount.
A more important achievement is the connected product sequence.
Historical maps explain electoral change. Polling measures current opinion. Forecasts estimate future outcomes. Early-vote tools track participation. Election-night systems interpret counting. CivicTracker extends into subsequent political activity.
My assessment is that VoteHub’s strongest achievement is enabling products across these stages to share data, technology, and user relationships. That offers greater potential for cumulative business value than repeatedly producing disconnected content around each news event.
47、A major controversy: Fabricated Wedgewood polls
On October 20, 2026, Wedgewood Polls admitted fabricating much of its survey data. It claimed its releases had entered aggregators including VoteHub; news organizations reported the statement.
FiftyPlusOne, which investigated the firm, said it had checked identities and methodology but subsequently confirmed that supplied raw data was fraudulent. A plausible methods statement therefore did not establish that genuine fieldwork occurred.
Donnini’s public response was reported as explaining that he had contacted Wedgewood, whose representative described himself as a college student with industry experience, and had extended some trust to a young entrant.
The issue is verification of external data. It does not establish that VoteHub’s team fabricated polling itself.
48、Why the controversy affects commercial value
The following is analytical judgment.
One VoteHub strength is recruiting talent and information from an enthusiast community. Wedgewood exposed the corresponding tension between openness and verification.
Models can discount old information, weak samples, or repeated releases. Weighting cannot resolve the underlying problem when the survey never happened.
Institutional customers ultimately need traceable sources, anomaly handling, withdrawal procedures, and recalculation—not only attractive charts.
Determining the scope of an error quickly is a core data-product capability. Removal counts and recalculation results announced by other aggregators should not be presented as VoteHub’s own remediation record.
49、Other risks worth examining
Feedback between models and markets. VoteHub’s Complete model incorporates market information while its data also appears on a prediction-market platform. The two signals may not always be independent. This is not evidence of manipulation.
Overconfidence from detailed presentation. A precise-looking precinct or demographic estimate does not necessarily rest on equally strong evidence at every level.
Election seasonality. Attention may decline after voting. CivicTracker offers a possible buffer, but does not prove that seasonality has been solved.
Commercial transparency. Partnerships and independent analysis can coexist. Readers need to understand which information comes from partners and which represents modeling judgments.
These are structural risk assessments, not allegations of unlawful or improper behavior.
50、Turning points and present position
The central transition in 2024 was from community collaboration to company operations. In 2025, it was expansion into tools and prediction-market cooperation. In 2026, it was subscriptions and integration into national news workflows, accompanied by greater scrutiny of data credibility.
Nathaniel Rakich’s October 2026 industry guide included VoteHub among election-forecasting platforms while emphasizing the distinction between polling averages and winning probabilities.
I would characterize Herrin as a founder organizing specialist community capabilities into a commercial institution, and VoteHub as an election-data and tools company with institutional adoption that still needs to build a long-term record.
Its next meaningful tests are election-night delivery, forecast calibration, correction practices, and whether customers continue using and paying for its products after peak election attention subsides.
More In-Depth
- Cal.com and Its Founders: Origins, Business Strategy, and the Open-Source Turning PointNov 11, 2026
- Linkfire and Its Founders: The Business Behind Music Links and Its Financial Ups and DownsNov 11, 2026
- DataFast and Marc Lou: Building a Software Business Through Products, Content, and AudienceNov 11, 2026
- TrustMRR and Marc Lou: Turning Verified Revenue into a Marketplace for Software BusinessesNov 11, 2026