Email deliverability. The hidden metric that decides whether any of your other metrics matter.

01What deliverability is (and isn't)

Two terms get conflated constantly and they're not the same thing:

Delivery rate measures whether your email was successfully transferred from your sending server to the recipient's mail server. It's a server-to-server protocol-level measure. For any reputable sender, this number is 99%+ all the time. If your delivery rate is below 95%, you have a bounce problem, not a deliverability problem.

Deliverability measures whether your delivered email actually reached the inbox of a human who could see it. The mail server accepted the message, then routed it — to the primary inbox, to spam, to the promotions tab, to a Gmail "All Mail" folder the user never checks, or to a silent filter that drops it without notification. Deliverability is the only number that matters for outbound, and almost no platform reports it accurately because most platforms can only see the delivery confirmation, not what happens after.

A typical modern cold outbound campaign looks like: 5,000 sent → 4,950 delivered (99% delivery) → ~3,500 actually in the inbox (~70% deliverability) → ~3,200 opens (some are MPP pre-fetches, not real opens) → ~120 replies (~3.4% of inboxed). The 30-point gap between delivery and deliverability is invisible in most reports. It's also, by a wide margin, the biggest hidden lever on outbound performance.

02The four pillars

Modern inbox placement is decided by four factors that ISP machine-learning filters weigh together. None alone determines the outcome; weakness in any one degrades the others. The relative weights have shifted substantially since 2022 — content and engagement matter more, technical authentication matters less (as table-stakes-but-not-discriminating), and sender reputation has become the central factor.

The shift in weighting matters operationally. In 2018, authentication + content carried most of the signal; reputation and engagement were secondary. In 2026, authentication is table-stakes, content is mid-weight, and reputation + engagement together make ~80% of the decision. The implication: technical fixes (better SPF, fancier DKIM rotation) deliver almost no marginal gain. Engagement-driving work — better targeting, better personalization, real signal anchoring — is now the only durable deliverability lever.

03SPF · DKIM · DMARC explained

The three authentication records every sender must publish on their domain. Each does a different job. Together they tell receiving ISPs "this is an authentic, authorized email from this domain":

The 2026 reality: all three are required by Gmail and Yahoo for any sender over 5,000 emails/day. Missing any one = your email goes straight to spam or gets rejected outright. There is no "but our content is good so it'll be fine" anymore. Authentication is the lowest-effort, highest-leverage deliverability investment — and the one most outbound teams misconfigure when they switch sending tools, add a sub-domain, or expand to a new mailbox provider.

04The 2022-2026 collapse

Industry-wide inbox placement for cold B2B outbound has fallen sharply over the last four years. The data, drawn from independent inbox-placement testing services that don't have a vendor stake in making the number look good:

Four forces drove the decline, each of which compounds with the others:

1. AI-powered spam classifiers. ISPs replaced rules-based filters (looks for "viagra," counts exclamation marks) with deep-learning models that evaluate hundreds of signals simultaneously. These models adapt faster than sender tricks. Domain rotation worked in 2020; in 2026, the model identifies the rotation pattern itself as a low-trust signal.

2. The MPP fallout. Apple Mail Privacy Protection (2021, iOS 15) artificially inflated open rates for Apple users by 20-40 points, which broke engagement-scoring systems that had been trained on real opens. ISPs adjusted their models to weight replies and human engagement more heavily — which structurally penalized senders relying on opens as their primary engagement signal. The MPP deep-dive covers this in detail.

3. The 2024 Gmail/Yahoo sender requirements. In February 2024, Gmail and Yahoo (representing roughly 60% of B2B inboxes worldwide) jointly mandated stricter sender behavior: mandatory authentication, one-click unsubscribe, and a hard cap on spam complaint rate (≤0.3% or face throttling/blocking). Senders below the bar saw immediate placement drops.

4. The volume arms race. Modern sequencers made it trivial to send millions of cold emails a week from rotating domains and warmed inboxes. ISP filters adapted: pattern-matching on rotation behavior, on warm-up tool fingerprints, and on the email content patterns sequencer-generated outbound tends to share. High-volume senders are now the primary target population for the most aggressive filtering.

The compounding result is the 20-point industry-wide drop. This is structural, not cyclical. The 2022 placement rates will not return. The teams that have adapted treat 60-65% inbox placement as the new normal baseline and design their outbound motion around that constraint — meaning much lower volume, much higher personalization, much more durable engagement-driving work.

05The reputation feedback loop

Sender reputation is the single biggest factor in inbox placement, and it's the one most people misunderstand because it operates as a self-reinforcing loop. Every send updates your reputation; your reputation determines how your next send is treated; how your next send is treated determines the engagement signals that update your reputation. The loop runs in both directions:

Two implications most teams miss. First, reputation is asymmetric. Earning a 5-point reputation lift takes weeks of consistent high-engagement sending; losing 5 points takes one bad campaign with high spam-complaint rate. The loop is positively-reinforcing on the way up and brutally fast on the way down.

Second, reputation is per-domain + per-IP, not per-account. Switching ESPs doesn't reset your reputation if you keep your sending domain. Adding a new sub-domain starts a new reputation from zero — which is why warm-up exists, and why aggressive senders rotate domains to escape their own bad reputation (a strategy ISPs increasingly detect and penalize).

The reputation loop is the deepest answer to "why doesn't this advice from 2020 work anymore." The advice doesn't work because the loop has compounded against most high-volume senders for five years now. Even with perfect authentication and great content, a sender with five years of compounding low engagement starts every campaign at a placement disadvantage that copy tweaks can't recover.

06The 2024 Gmail/Yahoo reset

The clearest single inflection point in cold-outbound deliverability happened in February 2024 when Gmail and Yahoo jointly published bulk-sender requirements that took effect in stages through the year. For senders doing more than 5,000 emails/day to Gmail or Yahoo addresses (which is most B2B outbound), the rules are mandatory:

If you set up your outbound stack before 2024 and haven't audited it since, you are very likely missing at least one of these requirements. The most common gap is DMARC alignment — most sales tools historically used a "via" sender setup that doesn't satisfy 2024 alignment. The fix usually requires custom DNS work and possibly a new sending subdomain.

07How to actually measure it

Most sales tools report "delivery rate" (server-to-server) and call it deliverability. That number is meaningless for outbound. To know your actual inbox placement, you need to measure it directly. Three methods, in order of credibility:

Seed-list testing. Maintain a list of test inboxes at every major provider — Gmail (personal + Workspace), Outlook, Yahoo, iCloud, plus 5-10 Microsoft 365 corporate domains. Send the same campaign to the seed list, then manually check placement at each. This is the gold standard because it shows exactly where messages land at each provider. The downside: it requires manual checking and a representative seed set.

Third-party placement testing services. Tools like GlockApps, Mailtrap, MailGenius, and Inbox Placement (the SparkPost tool) maintain global seed-list networks and report aggregate placement by provider. Cost is $50-300/month and gets you industry-standard testing without manual work. The downside: their seed addresses don't perfectly match your real recipient mix, so the numbers approximate rather than measure exactly.

Postmaster Tools + DMARC reports. Google Postmaster Tools (free) reports your domain's reputation, spam rate, and authentication results from Gmail's perspective — across all of your sending to Gmail recipients. DMARC aggregate reports (set rua= in your DMARC record) give you per-source authentication results. Both are essential and free; neither directly reports inbox vs. spam placement, but together they're the closest thing to a real-time deliverability dashboard for your domain.

08Volume vs. engagement strategy

Modern deliverability is fundamentally a trade-off between volume and engagement. ISP filters reward high-engagement low-volume sending and punish low-engagement high-volume sending. The 2×2:

The dominant strategy in pre-2024 cold outbound — high-volume + low-engagement + domain rotation — is the worst possible quadrant in 2026. The teams still operating this way are paying for a deliverability arms race they can't win.

The teams that have moved to the upper-left quadrant — fewer, better-targeted sends, anchored on real signals, with personalization that drives 10-20% reply rates — are seeing the inverse: their deliverability is improving year over year while everyone else's falls. This is not a coincidence; it's the rational outcome of how ISP filters were redesigned to reward.

09The 7-step recovery playbook

If your deliverability has degraded — declining reply rates, more "did you get my email?" complaints, suspicions that emails are going to spam — here's the 7-step recovery sequence. Run them in order; skipping ahead doesn't work because each step depends on the previous:

1. Audit authentication on every sending source. SPF, DKIM, DMARC must be valid for every tool that sends "from" your domain. The single most common cause of sudden deliverability drops is a tool whose authentication broke when the provider rotated keys. Tools like MXToolbox or Dmarcian make this a 30-minute audit.
2. Set up Google Postmaster Tools and DMARC aggregate reports. Get the visibility before you try to fix anything. You can't recover what you can't see. Postmaster Tools requires domain verification; DMARC reports require adding rua= to your DMARC record. Both are free.
3. Stop sending to disengaged segments immediately. Any recipient who hasn't opened/replied to any of your sends in the last 90 days is hurting your reputation more than helping. Suppress them. Most teams over-send to old lists and underestimate the reputation damage. Aggressive list hygiene is the fastest deliverability win.
4. Drop your weekly send volume by 30-50% for 4 weeks. Counter-intuitive but reliable: sending less with higher per-message engagement quality lifts reputation faster than any technical fix. Use the recovery window to over-invest in personalization and signal-anchoring.
5. Verify your one-click unsubscribe. Send a test campaign to a personal Gmail account, click the unsubscribe link from the Gmail UI (not the email body), confirm it works without a form. If it requires a form, you're violating the 2024 requirement and Gmail is downgrading you.
6. Watch your spam complaint rate weekly. Anything over 0.1% is a warning. Anything over 0.3% is an active emergency. Spam complaints come from clearly unwanted email — usually too-cold prospecting to recipients who don't recognize you. The fix is upstream: better targeting, clearer relevance, real signals.
7. Rebuild on signal-anchored sending, not the old volume motion. The recovery isn't sustainable if you go back to the same volume strategy. Use the recovery as an excuse to redesign the outbound motion around fewer, better-targeted, signal-anchored sends. The new baseline should be 30-50% of the old volume with 3-5× the reply rate.

Most teams running this playbook see meaningful improvement within 4-8 weeks. Teams that try to skip step 3 (suppressing disengaged segments) or step 4 (volume reduction) almost never recover — they keep poisoning their reputation faster than the technical fixes can repair it.

10Common mistakes

11Related terms

Deliverability sits at the intersection of compliance, technical infrastructure, and engagement quality. To go deeper: