· 9 min read · By Marwene Amor
Using these numbers to pick a stack? Our hands-on picks: Instantly for volume at the lowest cost per inbox, lemlist for personalization and LinkedIn steps, Hunter for finding and verifying emails. Full comparison in the best cold email software guide, or get the custom selection report matched to your case.
EmailToolsHub runs its own sending infrastructure, tests the tools it reviews inside real trials, and tracks its own traffic through Cloudflare. When a statistic below comes from our data, we say so and describe the measurement. When it comes from a vendor or a regulator, we name the source. No number on this page is repeated from another roundup. That matters because most cold email statistics circulating in 2026 trace back to vendor content marketing with no methodology attached.
85-90% of server-side affiliate link clicks on our site were non-human. Over a 30-day window in June-July 2026, our server-side tracked partner links recorded 283 clicks. Browser-side (JavaScript) tracking on equivalent links recorded 25 visitors and 1 click respectively for two other partners. After matching click spikes to crawler activity and deploy-triggered recrawls, we attribute the gap overwhelmingly to search crawlers, SEO tools, and AI bots. The practical lesson for anyone reading affiliate dashboards or ad analytics: server-side hit counts without bot filtering overstate human interest by an order of magnitude.
Fix that worked for us: moving every partner link behind an internal redirect path disallowed in robots.txt restored a readable human click signal within a day, with attribution preserved. Details in our methodology.
81 AI crawler and AI assistant requests hit our site in one 24-hour window (July 20-21, 2026), measured by Cloudflare AI Crawl Control. The split is the interesting part: alongside index crawlers (PerplexityBot, ClaudeBot, Applebot, Bingbot), we logged live user-triggered fetches from ChatGPT-User, Claude-User, and MistralAI-User, meaning real people asked an AI assistant a question and the assistant fetched our pages to answer. For a niche review site under three months old, that is evidence that AI answer engines now retrieve small independent sources, not only large publishers.
20 to 100 credits per identified buying signal. That is the disclosed price range we recorded across six AI Signal Agent configurations suggested inside Lemlist's wizard in July 2026 (job posting signals, tech-change signals, LinkedIn competitor-engagement signals, promotion signals). With estimated volumes of 12-45 signals per month per agent, a single agent at 100 credits per signal and 30 signals a month consumes 3,000 credits monthly. For scale: the free trial ships with 200 credits total. Full breakdown in our Lemlist review.
$109 per month is the plan tier required to apply AI-generated campaigns in the same tool; the trial demonstrates the full AI flow but gates the final step. We retested on July 28, 2026 on a partner-extended trial carrying 1,200 credits: the flow runs further than in early July (skippable discovery, a full visual workspace) and the final "create the campaign" click still requires the paid plan. We expect this "watch free, ship paid" pattern to become the dominant AI monetization shape across the category.
5 credits per verified email, 20 per phone number is Lemlist's enrichment pricing, confirmed to us directly by its partner team on July 24, 2026. On a 200-credit trial that means 40 verified emails or 10 phone numbers before the meter stops.
100 credits per inbox placement test is what Lemlist charges for a native seed test, recorded in-product on July 28, 2026, runnable on demand from the deliverability hub or from inside the campaign editor. At Lemlist's own enrichment rates that equals the price of 20 verified emails per test. The panel warns that seed tests send real emails and add to sending volume. Reference point: standalone tools like GlockApps sell this as a separate subscription.
10 tools is what Lemlist's MCP connector exposes in Claude's public connector directory as a one-click Community install, counted live on July 28, 2026 (campaign creation with sequence, lead search, push-to-contacts, inbox conversations, help-center search, among others). The manual wizard route still surfaces the raw 34-tool set with per-tool Allow / Ask / Block permissions. Neither Instantly nor Smartlead ships any MCP as of the same date.
5 credits per lead, 5 credits per reply is what Instantly's new AI agent stack charges (AI Sales Agent and AI Reply Agent respectively), measured in our hands-on test the day after its July 22, 2026 Lead Agent launch. A fresh workspace ships with 1,000 free credits, so the free allowance covers 200 AI-sourced leads. The Lead Finder Agent deploys free and consumes credits per lead found, with no credit price displayed anywhere in its four-step setup wizard. Full walkthrough in our Instantly review.
100 tools, 1 credit per Domain Search page, 0.5 per verification is what we measured on Hunter's official MCP server on July 28, 2026, by reading the account meter after each call on a live free plan. Hunter's remote MCP (version 0.7.0 at test time) exposes 100 tools to Claude and ChatGPT, against 34 for Lemlist and zero for Instantly and Smartlead. Discover company searches, enrichment lookups, and all list management moved the meter by exactly nothing; one Email Finder call registered 1.2 credits including its embedded verification. Our whole test session consumed 3.7 of the free plan's 50 monthly credits. Full walkthrough in our Hunter review.
Since July 14, 2026, open-rate tracking in prospecting emails is unlawful in France following CNIL enforcement guidance, because tracking pixels in cold outreach lack a valid consent basis. Combined with Apple Mail Privacy Protection pre-loading pixels since 2021 (which already made opens unreliable for roughly half of B2B inboxes), the open rate is finished as a decision metric for EU-facing senders: partially fictional before, now also a compliance risk. What to track instead: replies, positive reply rate, and clicks on non-tracking links. Our full analysis: why open rates lie.
Under $100 a month covers a competent solo stack; overbuying is the norm we observe. A solo founder sending under 5,000 emails a month can run finder, verifier, sender, and warmup for roughly $96 a month (dedicated sender plus finder), or as low as about $30 a month on a freemium bundle when volume is lower still. The recurring mistake we see in reader stacks is paying for scale features (10+ rotating inboxes, unlimited warmup slots) at 20 times the volume they send. Full breakdown by profile in how to choose your stack.
14 to 21 days is the minimum domain warmup period before cold sending that we apply and recommend; shortcuts show up as spam placement within the first campaign. 3 DNS records (SPF, DKIM, DMARC) are non-negotiable, and DMARC enforcement matters: moving our own domain from p=none to p=quarantine was flagged as the single riskiest gap by two independent deliverability audits of our infrastructure. 2 to 5 touches is the follow-up range that captures most reply value; beyond 5, incremental replies fall off while spam-complaint risk keeps climbing. Protocols: warmup, DNS setup, cadence.
These statistics are original to EmailToolsHub. You are welcome to cite them with attribution and a link to this page. This page is updated as new test data comes in; the date at the top reflects the last revision. Questions about methodology: contact us or read the methodology page.
In our measured case, 85-90% of raw server-side clicks on partner links were non-human over a 30-day window (283 recorded clicks versus roughly 25-50 estimated human clicks). The share depends on how exposed the links are to crawlers: links repeated across many public pages attract continuous bot hits, and every site redeploy triggers recrawl waves. Any click count measured server-side without bot filtering should be treated as an upper bound, not a human signal.
Yes. In a single 24-hour window in July 2026 we logged 81 AI-related requests to our site, including live user-triggered fetches from ChatGPT-User, Claude-User, and MistralAI-User agents, alongside index crawlers from Perplexity, Anthropic, Apple, and Microsoft. Our site was under three months old at the time. The precondition is technical: the site must not block AI crawlers, must resolve quickly, and benefits from clean structure (llms.txt, structured data, answer-first summaries).
For prospecting emails, yes, since CNIL enforcement guidance effective July 14, 2026: tracking pixels in cold outreach lack a valid legal basis under EU privacy rules. B2B senders targeting France have removed open tracking or excluded French recipients from tracked sends. Reply-based metrics are unaffected. This adds a legal reason on top of the accuracy problem Apple Mail Privacy Protection created in 2021.
Roughly $30 to $100 a month, depending on volume. Below a few thousand emails monthly, a freemium bundle covering finder, verifier, and light sending runs about $30. A dedicated sender plus a specialist finder lands near $96. Costs above that level usually mean paying for scale features a solo sender does not use. The stack composition matters more than the spend: finder, verifier, sender, and warmup each have a distinct job.
Best Cold Email Software 2026 · Lemlist Review · Deliverability Checklist · Why Open Rates Lie · Our Methodology