Forensic teardown

The tools analysts swear by — and what they're really made of

Every "data-driven" price target leans on a vendor feed. Most of those feeds are modeled guesswork dressed up as measurement. Here's what the big alt-data tools actually source, and how accurate each one really is.

Accurate / direct Modeled / partial Unreliable Not applicable
Tool / Platform Direct data sources Plain-English explanation Website traffic accuracy App downloads accuracy DAU accuracy % free / public data
SimilarwebBrowser/app panels, ISP/telecom, partner sites, own crawlers, WHOIS/registriesTracks browsing samples, partner data, bots, and public records.Often 20–50% off vs. real analyticsNot reliableModeled guesses~90% free/open
SEMrush (Traffic Analytics)Clickstream panel (~200M devices), Google Ads Keyword Planner, Google Trends, partnersFollows a large user sample via apps/extensions; combines with Google's keyword tools.Estimates; relative comparisons onlyWeakWeak~80–90% free
AhrefsOwn crawlers, Google Keyword Planner, Google Trends, Search Console (if owner connects)Bots scan links & rankings; infers traffic via CTR models with Google data.CTR-model guesses (not direct)N/AN/A~95% free
ComscorePanels + census tagsSome sites install Comscore tags; panel extrapolates to audience reach.Strong for large publishersN/ADAU extrapolated (modeled)~40% free, rest proprietary
Sensor TowerDevice panels, App Store/Google Play data, ad datasets, developer connectionsCollects device usage + app-store scraping; exact if developer connects.Weak for webGood directional (~10–20%)DAU modeled~50% free
data.ai (App Annie)Mobile panels, developer accounts, app storesExact if developer connects account; otherwise modeled from panels + stores.Weak for webStrong with direct connectsModeled unless connected~50% free
ApptopiaDeveloper accounts, app stores, metadataExact when developer shares analytics; otherwise estimates from app stores.WeakExact with dev connectDAU modeled~60% free
Cloudflare RadarCloudflare network telemetry, 1.1.1.1 DNS resolver, BGP logsCaptures a big slice of global traffic; good for macro trends/outages.Good macro signals; not per-siteN/AN/A~30% free
StatcounterOn-site JS tracking code (~1.5M sites)Sites install tracking code; aggregated results across participants.Exact for tagged sitesN/AOnly for tagged sites~10% free
ChartbeatOn-site pings every 15sInstalled on publisher sites; measures engaged time.Exact where installedN/AExact DAU~5–10% free
Parse.ly (WP VIP)Tracking pixel + event pipelinesUsed by publishers; tracks every page load, unsampled.Exact where installedN/AExact DAU~5% free
PromptWatchScrapes prompt communities, forums, GitHub, APIsTracks AI prompts shared in public forums & repos.Not traffic-focusedN/APublic-forum DAU only~95% free

The takeaway: when an analyst cites "traffic data" to justify a target, ask which feed — and whether it's measured or modeled. Most of it is the latter. Informational only. Not investment advice.