DoWhatWorks
2 known investors
DoWhatWorks tracks A/B tests across millions of public websites, validates them with a research team, and provides a searchable database plus dashboards showing which page variations major brands favor. It serves growth, product, and design teams with competitor insights, testing roadmaps, and wireframe recommendations for website conversion optimization.
Also known as DoWhatWorks
Founders & leadership
Investors · 2
Company profile
researched Aug 2026DoWhatWorks is a conversion rate optimization (CRO) software company whose patented engine crawls millions of public websites daily to detect split tests being run by other brands. The engine identifies what elements are being tested, how many variants exist, and which version the brand ultimately keeps live; a research team then reviews and validates every test, adds metadata, and publishes it to a searchable database. Detection relies exclusively on public data on non-logged-in pages: no cookies, pixels or tracking scripts are placed on either customer or monitored sites, and the approach is independent of whatever experimentation tooling the observed company uses.
Customers access the library through a personalized dashboard where they can subscribe to a sector, track specific competitors, filter by page type or by the specific element tested (for example button shape or color), view desktop and mobile renderings of control and variant with annotations of winning and losing treatments, link to the live page, and save tests to a watchlist. On top of the dashboard the company sells research services built around BetScores, algorithmic predictions of how likely a strategy is to win derived from meta-analysis across the full library of more than 27,000 tests. Service tiers include a Rapid Analysis (BetScore analysis of one to two key areas in one month), Testing Strategies (a Blind Spot Analysis diagnostic followed by monthly test strategies with hypotheses and low-fidelity wireframes), Advanced Analysis (broader diagnostics, BetScores for roughly four to six areas and a prioritized testing roadmap, without wireframes), and an Interactive Page Blueprint for new page development or redesigns, delivering research-based best practices, a lo-fi wireframe and a post-launch testing roadmap in about two months.
Founding story
Co-founders Andres Glusman (CEO) and Will Howard (CTO) worked together at Meetup for about a decade, where Glusman led product and growth roles — including helping launch the company and generating its first $14 of revenue — and Howard rose through engineering to director. Both ran many experiments there and were frustrated by how long results took and how rarely tests won. They left Meetup at the beginning of 2019, after the company's exit, and began prototyping ideas; one prototype was a technology they invented and later patented to detect experiments run by other companies, initially pointed at brands such as Netflix, Warby Parker and GEICO. After showing screenshots to marketers and product managers and observing strong interest at a conference, they validated demand by sending prospective users a Stripe payment link in exchange for a promised monthly PDF report before any user interface existed, then built a v1 dashboard ahead of the delivery date and iterated with those paying early adopters.
Business model
B2B subscription software plus research services. Customers subscribe to dashboard access covering a sector and a chosen set of tracked companies, and can additionally buy tiered research engagements (Rapid Analysis, Testing Strategies, Advanced Analysis, Interactive Page Blueprint) delivered by an in-house research team. The company charged users from the outset rather than offering a free product, and sales run through a book-a-demo/request-access motion.
Recurring subscriptions to the experiment database and custom dashboards, supplemented by fee-based analysis and testing-roadmap engagements of one to two months or delivered monthly.
Traction
At the May 2021 introduction the engine analyzed more than 1,500 companies daily; by September 2023 about 1,600 companies daily and a library of more than 17,000 experiments; company materials in 2024-2025 cite more than 27,000 A/B tests. Reported customer outcomes in an investor interview include one client reporting a $4M lift in annual recurring revenue and others citing double-digit percentage increases in metrics such as sign-ups. Stated client base spans six of the top streaming brands, Fortune 500 SaaS brands, banking, learning and DTC health. The company also reports interviewing 22 website teams over a 12-month period as part of its research content.
Latest developments
The website advertises a preview of a new AI agent and publishes ongoing research, including analyses of competitor comparison pages and their citation in LLMs, findings that AI messaging tends to lose in tests despite widespread adoption, and results indicating customer logos and other social proof lost in most tests. A May 2025 investor post lists DoWhatWorks among Everywhere Ventures' companies.
▸Full profile — market position, technology, go-to-market, history
Market position
Positioned as a CRO data provider whose differentiator is observing competitors' live experiments rather than instrumenting a customer's own site. The company frames the problem via Optimizely research indicating only about 10% of tests beat their controls, and states it is the only party able to assemble this data set because of its patent; an investor profile describes an ambition to become the "AutoCorrect" or "Grammarly" of website UX and copy.
Competitor-observed experiment data collected without pixels, cookies or customer-side instrumentation; patented detection technology; analyst validation and metadata tagging of every published test; and BetScores derived from meta-analysis of a 27,000+ test library, combined with delivery of concrete recommendations, testing roadmaps and lo-fi wireframes.
Technology
A patented crawler-based engine, described by the company as working similarly to Google's web crawl, that ingests public data from millions of websites daily to detect running split tests, the number of variants, and which variant the brand favors and keeps live. Human analysts validate each test and attach metadata so meta-analysis can be run across the corpus; that meta-analysis produces BetScores, algorithmic win-likelihood predictions that factor in the number of variables tested and industry relevance. The system requires no pixel or tracking script and covers only public, non-authenticated pages, excluding tests deep inside registration and subscription flows.
Go-to-market
Inbound demo requests and early-access sign-ups from the website, supported by published research content (industry deep dives, trend analyses, interviews with website teams) and industry-specific test previews. Early growth came from founder-led selling and conference conversations, followed by identifying the customer clusters that succeeded with the product and targeting look-alikes.
Growth, marketing, product and design teams, primarily at enterprise and mid-market companies. Named customer categories include six of the top streaming brands, Fortune 500 SaaS companies, and firms in banking, learning, DTC health, ecommerce, travel and meal subscription.
History
The founders left Meetup in early 2019 and spent roughly two years building the detection engine, which by the May 2021 public introduction was analyzing over 1,500 companies per day and publishing findings to clients via a dynamic dashboard. Early clients cited at launch included three top streaming services, two Fortune 500 SaaS companies and 20 percent of meal subscription companies, and the company was hiring across growth analysis, data science, customer success and community roles while opening limited early access. By September 2023 the engine was analyzing about 1,600 companies daily across a library of more than 17,000 experiments, and by 2024-2025 company materials cite over 27,000 A/B tests underpinning BetScores, alongside a layered set of paid research offerings.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Timeline · 4
launches, deals, and filingsThe company's website promotes a preview of a new AI agent alongside its dashboard and research services.
Company materials describe BetScores, algorithmic win-likelihood predictions from meta-analysis of 27,000+ tests, delivered through four service tiers: Rapid Analysis, Testing Strategies, Advanced Analysis and Interactive Page Blueprint.
The company publicly introduced its experiment-detection engine and client dashboard, reporting analysis of over 1,500 companies daily and opening early access to a limited number of companies.
Andres Glusman and Will Howard, who had worked together at Meetup for about a decade, left the company at the beginning of 2019 to build DoWhatWorks.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
▸Research sources · 8
primary sources listed
- DoWhatWorksdowhatworks.io · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does DoWhatWorks do?
- DoWhatWorks detects and validates A/B tests run by major brands, then sells the results as a conversion-optimization database.
- Who founded DoWhatWorks?
- DoWhatWorks was founded by Will Howard.
- Who are DoWhatWorks's investors?
- DoWhatWorks's investors include Everywhere Ventures, The Fund.