Deliverability isn't an IT problem. It's a customer experience problem.

When email starts landing in spam, the first reaction is almost always technical. Someone opens DNS, checks SPF, validates DKIM, adjusts DMARC. When all three are fine and the problem persists, the confusion starts.

The confusion happens because deliverability is not a technical problem. Authentication is the price of entry. Who decides whether you land in the inbox or in spam is the subscriber, with every action or inaction they take on your message.

Modern Gmail, Outlook, Apple Mail, and Yahoo see your send through a lens that has little to do with server, IP, or DNS record. It is the lens of the relationship. And that lens is a product and marketing lens, not an IT lens.

Delivery rate is not deliverability

Worth separating two terms often used as synonyms that are not.

Delivery rate measures whether the recipient's server accepted your send. SMTP returned 250 OK. Done, it was delivered. That number easily sits at 98%+ for any minimally configured sender.

Deliverability measures whether the recipient received it in the primary inbox versus another folder (Promotions, Spam, Junk, Archives). The server accepted, but the provider's internal triage may have moved the message somewhere it will never be opened.

Delivery is a server decision. Deliverability is an inbox placement algorithm decision. And the inbox placement algorithm consults nothing in your DNS when it decides.

What the provider is actually looking at

Providers at scale (Gmail processes billions of messages per day) need a scalable signal to decide who lands in the inbox. That signal is built from hundreds of variables, but they group into two families.

Positive signals. Subscriber opens. Subscriber clicks. Subscriber replies. Subscriber marks "not spam". Subscriber drags from Spam to Inbox. Subscriber moves to a custom folder. Subscriber adds the sender to contacts.

Negative signals. Subscriber deletes without opening. Subscriber ignores for weeks. Subscriber drags to Spam. Subscriber clicks "report phishing". Subscriber unsubscribes (smaller effect, but measured). Subscriber archives without opening.

The strongest signal of all is "mark as spam". Internal Gmail research shared at deliverability conferences indicates that this signal can weigh more than all others combined. That is why the recommended limit is below 0.1% and the catastrophic limit is 0.3%.

The second strongest signal is prolonged absence of any action. An address that receives 40 messages from you in three months and never opens is treated by the provider as an address that does not want those messages. The provider starts delivering your messages to that subscriber directly to Promotions or Spam. Worse, it starts applying the same skepticism to other subscribers on the same provider.

Reputation is cumulative and contextual

Important technical detail: the reputation the provider calculates about your sender is not a single score. It is contextual.

Gmail keeps a reputation for your sends to Gmail users. Outlook keeps another for your sends to Outlook users. Reputation calculated by Gmail does not transfer to Outlook. Worse: inside Gmail itself, there are reputation components that are per individual subscriber. A single subscriber may be receiving in the inbox while another subscriber on the same domain is receiving in spam.

This means "my open rate is 22%" does not tell you much on its own. A 22% rate on a well-segmented list is great. A 22% rate on a 200,000-person list with 60% inactives is a disaster in disguise, because real performance among the engaged is lower than it looks and the negative signal from the unengaged is sabotaging you.

Eight practices that change the signal

Moving the signal requires changing how you operate, not how you authenticate.

1. Permission-based marketing as the base, not as a detail. Double opt-in for lists that will receive heavy content. Clear copy on the form about what the person is accepting. Without confirming opt-in, you are building a list that will become a complaint list.

2. Segmentation by engagement, not by persona. Persona helps with creation. But the decision of who receives versus who does not should depend on recent engagement. A subscriber inactive for 90 days enters a different pool. A subscriber inactive for 12 months enters sunset.

3. Relevant dynamic content. Personalization by name is the bare minimum of the bare minimum. What moves the metric is content that changes per subscriber. Recommendations based on behavior. Frequency tuned by preference. Send time tuned by timezone or by historical open pattern.

4. Interactive elements. Polls, votes, inline forms. AMP for Email expands what is possible on some providers. It is not decoration: every interaction is a positive signal logged.

5. Decent preference center. "Unsubscribe from everything" is a design failure. The preference center needs to let the subscriber reduce frequency, pick topics, pause for a few weeks. Each of these paths is an alternative to complaint and to a full unsubscribe.

6. Replyable sender. noreply@domain.com is an old decision. Providers treat addresses that accept replies as a legitimate sender signal. Addresses that reject replies are treated with a bit more skepticism. Running a monitored reply mailbox is worth the work.

7. Recognizable sender name + BIMI. The subscriber opens what they recognize. Consistent name (no changing per campaign), logo visible via BIMI when the domain has enforced DMARC, alignment between sender and brand. Visual recognition reduces "delete without opening" and "spam by mistake".

8. QA before sending. Accessibility (alt text, contrast, keyboard navigation), rendering across different clients (Outlook desktop is still a cruel client), dark mode behavior, message weight (messages above 102KB are clipped by Gmail). QA is the most neglected part and the one that corrects the most per-subscriber outcomes.

Who owns this, inside the company

The most useful political question to ask: inside your company, who is accountable for deliverability?

When the answer is "IT" or "the server people", you are at risk. IT configures SPF, DKIM, DMARC. IT maintains DNS. IT does not decide segmentation, frequency, copy, or preference center. The decisions that move the signal the most are marketing and product decisions.

The model that works at companies with good deliverability is a hybrid team. Marketing drives. IT runs the infra part. Data and analytics measure. Product gets involved when email is part of onboarding, retention, or recovery.

When nobody owns it, everyone points at the other side when spam comes back.

How Email Intelligence helps

Email Intelligence is built on the assumption that engagement is the signal that matters. It connects to your ActiveCampaign account and automatically creates fields that materialize the per-subscriber relationship: current engagement index, engagement trend (rising or falling), last meaningful activity, complaint risk, and indicators that help decide who stays, who gets frequency reduced, and who enters sunset. The IT part stays with IT. The relationship part becomes queryable data.

We are opening access to the free beta.

I want beta access →