How we measure things

The Outsolvi tracker tiers every email open Tier 1 through Tier 5. The score is computed at request time from a combination of request-shape signals (IP block, User-Agent, time-of-day, repeat rate, prefetch headers) and recipient-context signals (mailbox provider, device, whether the same recipient has loaded other emails from this sender recently).

Tier 1: high-confidence human read. Request shape matches a real browser session on a consumer device or known business endpoint, no prefetch headers, first or second load in a normal time window.

Tier 2: probable human, weaker signals. Might be a real read on an unusual device or through a privacy-respecting client; might be a slightly disguised proxy.

Tier 3: ambiguous. Often Apple Mail Privacy Protection on a recipient who eventually reads but where the proxy fired first.

Tier 4: low-confidence machine. Corporate email scanners, link-checkers, security sandboxes that pre-open every attachment and image.

Tier 5: bot or scanner with very high confidence. Apple MPP pre-fetches from Apple data centers, Gmail's image proxy refetches, known security-tool User-Agents.

The default dashboard filters Tiers 4 and 5 out of the reported open count. Reps can flip the filter on a per-campaign basis if they want to see the full count for diagnostic reasons.

The reply-rate and open-rate benchmarks we publish (on our B2B benchmarks tools and inside several blog posts) are aggregated from the Outsolvi tracker fleet, with explicit rules about what gets counted.

We include sends where the sender has at least 30 days of consistent activity (so the warmup phase doesn't distort the baseline). We exclude transactional and marketing-blast sends — only one-to-one outbound and warm-thread engagement count toward the rate metrics. We segment by industry and seniority of the recipient so the numbers a SaaS rep sees aren't artificially deflated by recruiting-agency cold outreach in the same dataset.

When we cite external research (InsideSales, Drift, ChiliPiper, Litmus, Postmaster Tools), we link directly to the source. We don't paraphrase or round numbers to make them look better than the original report did.

When we cite a Litmus, Klaviyo, or HubSpot benchmark, we check the publication date. Benchmarks older than 18 months get re-validated against our fleet data before we cite them, because the mailbox-provider behavior has shifted enough since 2023 that pre-2024 benchmarks rarely hold up.

When we test mailbox-provider behavior (Apple MPP prefetch timing, Gmail image proxy refetch, Microsoft 365 SafeLinks scanning), we run controlled sends from a dedicated Outsolvi test domain to seed inboxes on each major provider. The seed inboxes are real accounts on Apple iCloud, Gmail (consumer and Google Workspace), Microsoft 365 (consumer Outlook and enterprise), and Yahoo.

Each test fires a known sequence of pixel loads and link clicks, then we compare the server-side event log against the actual inbox state. The deltas are how we calibrate confidence tiering. We re-run this calibration quarterly and any time a mailbox provider announces a behavior change.

The test methodology and the seed inbox list are not public — we rotate them periodically so the calibration doesn't get gamed. The aggregated calibration results (what percentage of opens fall into each tier on each provider) are published in quarterly notes on the blog.

Every long-form page on outsolvi.com has a "Last reviewed" date at the bottom. That date reflects a real editorial review by Nate, not an auto-bumped timestamp. We review pages on the following cadence:

• Benchmark pages: reviewed quarterly, and any time the underlying source data ships a major update.
• Deliverability pages (SPF, DKIM, DMARC, BIMI generators and explainers): reviewed whenever Gmail or Microsoft publishes a bulk-sender policy change. Otherwise twice a year.
• Apple MPP and proxy-related pages: reviewed whenever Apple ships a Mail behavior change in macOS/iOS, plus quarterly recalibration of the impact estimator.
• Comparison and alternatives pages: reviewed whenever a competitor ships a major feature or pricing change. We track this through competitor changelogs and public announcements.
• Glossary and reference: reviewed annually, sooner if a term's underlying technology shifts.

If you find an error or an out-of-date claim on any page, email us at support@outsolvi.com and we'll fix it. We post-mortem any substantive corrections in a quarterly "What we got wrong" note on the blog.