On this page
Haus
Haus
Haus (haus.io) is a specialist marketing measurement and decision science platform built around Incrementality Testing via geo experiments. Founded in 2021 by Zach Epstein (former Google data scientist and PM), it uses Synthetic Control Method|synthetic control methodology to estimate the causal lift of marketing spend — producing evidence of what would have happened to revenue without a given channel's activity. As of 2026, Haus has expanded from pure geo-experiment tooling into a broader measurement platform covering Causal MMM, causal attribution, and AI-led budget optimisation.
Company and products
Haus is headquartered in Mountain View, California. Its product suite as of late 2026 includes (as-of 2026-11-07):
- Incrementality Experiments (GeoLift) — the core product; geographic treatment/holdout experiments using Synthetic Control Method to build a weighted counterfactual from comparison regions
- Fixed Geo Tests — for measuring offline and always-on channels: billboards, regional activations, OOH (launched June 2025)
- Time Tests — for brand moments and campaigns that cannot be isolated geographically
- GeoFences / Commuting Zones — precise geographic demarcation for urban holdout tests (August 2025 and July 2025 respectively)
- ColdStart — incrementality testing for new products, markets, or channels that lack historical data, using comparable KPIs (August 2026)
- Causal MMM — a Marketing Mix Modeling (MMM) product that uses geo experiments as ground truth rather than adding them post-hoc as calibration (launched October 2025)
- Causal Attribution — daily ad-level incrementality reporting, described as "day-to-day incrementality down to the individual ad" (launched July 2024; expanded July 2026)
- Architect — an AI marketing agent that flags risks and opportunities, runs what-if analyses, and recommends budget actions; described as removing "a large amount of manual data work for marketers" (launched September 2026) (as-of 2026-11-07)
Scale claims on the Haus homepage as of 2026 (vendor-reported, volatile): $30B+ in annual ad spend managed, $1T+ in combined customer revenue. Named customers include Intuit, FanDuel, Wayfair, Dyson, Sonos, Oura, AG1, Grubhub, Hims & Hers, Caraway, Newton Baby, Jones Road Beauty, Coursera, and Pernod Ricard USA.
Haus holds ISO 27001 and SOC 2 Type 2 certifications (as-of 2026-11-07). It was included in Forrester's Marketing Measurement and Optimization Platforms landscape in September 2026. Stanford economist Susan Athey joined as Scientific Advisor in October 2025. Olivia Kory was named CMO in June 2026.
Methodology
Haus's core method is real-world geo experiments: geographic regions are split into treatment and holdout groups, and a Synthetic Control Method|synthetic control — a weighted combination of multiple untreated regions — estimates the counterfactual (what revenue would have been without the tested channel). Haus explicitly contrasts this with matched-market tests in a January 2025 blog post, arguing that synthetic control produces more precise counterfactuals than the matched-market approach.
Pre-experiment setup includes "cAI placebo tests" — running the proposed model on historical data to estimate inherent KPI noise — which informs holdout size and test duration to target at least 80% statistical power (as-of 2026-11-07). Outlier handling outputs include daily, weekly, winsorized, and non-winsorized analyses; winsorizing replaces extreme values with cutoff values rather than removing them.
Haus's platform is built on first-party data only; it requires no cookies or third-party identifiers, making it structurally privacy-durable.
For Causal MMM, Haus argues that experiments should anchor the model from the outset as ground truth, contrasting with "traditional MMM built on historical correlations where models can disagree with each other or with recent experiments." A July 2026 blog post frames this as: experiments and MMM must be used together, with experiments anchoring the MMM — not the other way around.
Haus characterises Meta's Meta GeoLift and Google's Meridian GeoX as "research toolkits, not enterprise platforms" that "can have confidence intervals too wide to act on."
Benchmarks and research output
Haus has published proprietary research drawn from its customer base (as-of 2026-11-07):
- The Meta Report (July 2025) — 640 incrementality experiments across hundreds of brands and billions in ad spend; found Meta Advantage+ automation showed net-positive incremental lift in only 42% of tests, meaning manual campaigns outperformed the automated offering more than half the time (as-of 2025-07; stale-risk applies as Meta's Advantage+ has continued to evolve)
- YouTube Incrementality Report (March 2025 / updated October 2026) — 190 Haus incrementality tests on YouTube ads (as-of 2026-10-14)
- 2026 Marketing Decision Confidence Index — survey of 500 paid marketing and finance professionals on measurement confidence, AI adoption, and planned channel investments (as-of 2026)
- Experiment-Anchored MMM report — Haus's research arm report on the case for causal MMM over traditional unanchored models
Case studies
Customer case studies (self-reported, unverified by independent sources):
- Newton Baby — used Haus to discover that branded search was non-incremental (search ads capturing conversions that would have happened anyway); cut branded search spend with no corresponding revenue decline. Separately, discovered TikTok spend was driving incremental Amazon sales not registered in TikTok's own attribution — a cross-channel halo effect.
- Jones Road Beauty — doubled new customer orders on YouTube by using Haus incrementality tests to validate YouTube as an incremental acquisition channel and then scale investment confidently.
- FanDuel — increased Sportsbook activations by 11% and increased experimentation rate by 800% (six experiments launched in Q4 2022); SVP Marketing Brian Borkowski stated: "The Haus platform provided us clarity in our data and led to better investments that saved us millions of dollars." (as-of Q4 2022)
Funding
Disclosed funding rounds (as-of 2026-11-07):
- August 2023: $17M Series A led by Insight Partners; participation from Baseline Ventures, Haystack Ventures, Upside Partnership, Octave Ventures, Mantis Venture Capital
- July 2024: $20M additional financing led by 01 Advisors (David Fischer, former Facebook CRO; Dick Costolo, former Twitter CEO; Adam Bain, former Twitter COO); participation from Rahul Mehta (DST Global), Gokul Rajaram, Insight Partners, Baseline Ventures, Haystack, Octave, Mantis
Total confirmed public funding: approximately $37M. SegmentStream (a competing vendor) claimed $55.3M total including an $18.3M April 2025 Series B extension, but this could not be independently verified in any press release, law firm news release, or financial aggregator. Treat the $37M as confirmed and the $55.3M as unverified.
Pricing
Haus does not publish list prices; pricing is quote-only through an annual Order Form. Tiered offering as reported by RFP.wiki (as-of 2026, volatile):
- Basics — US-only, self-serve GeoLift experiments
- Core — international testing, dedicated strategist, data warehouse integrations
- Plus — adds Time Tests, Fixed Geo Tests, and Causal MMM
Vendr marketplace data (via RFP.wiki) reports a median annual contract of approximately $150,000 (observed range $132K–$168K), billed annually on Net 30 terms (as-of 2026, volatile). Subscription fees are nonrefundable; subscriptions auto-renew unless cancelled at least 30 days before term end; price increases require 60 days' notice.
Competitive positioning
Haus positions itself as a specialist incrementality platform — experiment-first, not an all-in-one attribution suite. Key competitive comparisons from independent sources and from Haus's own framing (as-of 2026, volatile):
| Vendor | Model | Signal |
|---|---|---|
| Haus | Geo experiments (synthetic control); Causal MMM as add-on | Experiment-first, mid-market to enterprise, $150K/year median |
| Measured | Managed service, more advisory support; enterprise-focused | Higher-priced ($150K–$500K+), less self-serve |
| NorthBeam | MTA-first (always-on); incrementality launched Dec 2025 | Lower price point ($1.5K–$3.5K/month), less rigorous for large brands |
| Triple Whale | All-in-one (MTA + MMM + incrementality) | SMB-friendly, $219–$4.2K/month, scales with GMV |
| Meta GeoLift | Free; geo holdout open-source toolkit | Haus: "research toolkit, not enterprise platform" |
| Meridian GeoX | Free; geo experiments within Google's Meridian MMM | Haus: "confidence intervals too wide to act on" |
What practitioners report
Buyer praise from Alium Research anonymised interviews (as-of 2026): finance teams trust the incrementality data ("the finance team definitely understands the incrementality data"); "very strong correlation with the actual revenue figures"; support team described as "very smart" and helpful in test planning; extensive training provided.
Buyer criticisms from Alium Research anonymised interviews: annual contract inflexibility ("we can't turn off Haus and save on the dollars") with preference for pay-per-test model; onboarding described as "cumbersome, slow … stalls on messy internal data"; mid-size brands question whether Haus's confidence intervals are meaningful at their traffic levels; more self-serve than buyers expected.
RFP.wiki identifies structural weaknesses (as-of 2026, volatile): geo tests require sufficient spend and traffic (small budgets produce intervals too wide to act on); causal reads require multi-week windows (slower than always-on dashboards); statistical concepts require onboarding for teams without data science skills.
Gartner Peer Insights: 4.3/5 from 6 reviews (very small sample; as-of 2026, volatile). Control methodology (4.7/5) and attribution boundary handling (4.6/5) rated highest; governance/permissions (3.2/5) rated lowest.
Key terms
| Term | Meaning |
|---|---|
| GeoLift | Haus's core geo holdout experiment product — treatment vs holdout split, synthetic control counterfactual |
| Synthetic control | A weighted combination of untreated regions used to estimate what treated regions would have experienced without the marketing |
| ColdStart | Haus feature for testing incrementality in new markets/channels without historical data |
| Architect | Haus's AI marketing agent for budget optimisation and risk flagging (launched Sep 2026) |
| Causal MMM | Haus's Marketing Mix Modeling (MMM) product anchored to geo experiments rather than historical correlation |
| Causal Attribution | Daily ad-level incrementality (without geo holdout); launched July 2024 |
| cAI placebo test | Pre-experiment validation: runs the proposed synthetic control model on historical periods to estimate noise and inform power calculations |