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 |
|---|---|---|---|---|---|---|
| Similarweb | Browser/app panels, ISP/telecom, partner sites, own crawlers, WHOIS/registries | Tracks browsing samples, partner data, bots, and public records. | Often 20–50% off vs. real analytics | Not reliable | Modeled guesses | ~90% free/open |
| SEMrush (Traffic Analytics) | Clickstream panel (~200M devices), Google Ads Keyword Planner, Google Trends, partners | Follows a large user sample via apps/extensions; combines with Google's keyword tools. | Estimates; relative comparisons only | Weak | Weak | ~80–90% free |
| Ahrefs | Own 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/A | N/A | ~95% free |
| Comscore | Panels + census tags | Some sites install Comscore tags; panel extrapolates to audience reach. | Strong for large publishers | N/A | DAU extrapolated (modeled) | ~40% free, rest proprietary |
| Sensor Tower | Device panels, App Store/Google Play data, ad datasets, developer connections | Collects device usage + app-store scraping; exact if developer connects. | Weak for web | Good directional (~10–20%) | DAU modeled | ~50% free |
| data.ai (App Annie) | Mobile panels, developer accounts, app stores | Exact if developer connects account; otherwise modeled from panels + stores. | Weak for web | Strong with direct connects | Modeled unless connected | ~50% free |
| Apptopia | Developer accounts, app stores, metadata | Exact when developer shares analytics; otherwise estimates from app stores. | Weak | Exact with dev connect | DAU modeled | ~60% free |
| Cloudflare Radar | Cloudflare network telemetry, 1.1.1.1 DNS resolver, BGP logs | Captures a big slice of global traffic; good for macro trends/outages. | Good macro signals; not per-site | N/A | N/A | ~30% free |
| Statcounter | On-site JS tracking code (~1.5M sites) | Sites install tracking code; aggregated results across participants. | Exact for tagged sites | N/A | Only for tagged sites | ~10% free |
| Chartbeat | On-site pings every 15s | Installed on publisher sites; measures engaged time. | Exact where installed | N/A | Exact DAU | ~5–10% free |
| Parse.ly (WP VIP) | Tracking pixel + event pipelines | Used by publishers; tracks every page load, unsampled. | Exact where installed | N/A | Exact DAU | ~5% free |
| PromptWatch | Scrapes prompt communities, forums, GitHub, APIs | Tracks AI prompts shared in public forums & repos. | Not traffic-focused | N/A | Public-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.