theseventhsense.com

A 99%+ delivered rate in Marketo only confirms that receiving servers accepted your messages — not that anyone saw them. If your dashboard shows strong delivery but weak opens and clicks, the gap is almost always inbox placement (spam or promotions folders), provider-specific filtering, or list fatigue. The fix is to stop treating "delivered" as a success metric and start diagnosing placement, engagement, and provider-level performance separately.

Delivery rate, open rate, and inbox placement are three different things

Marketo's delivered rate measures one event: the recipient's mail server accepted the message. It says nothing about which folder the message landed in or whether the person ever looked at it.

The practical takeaway: when delivery is high but clicks are flat, you have a placement or relevance problem, not a sending problem. Treat click rate and per-provider breakdowns as your primary health signals.

Break your Marketo metrics down by inbox provider

Aggregate open rates hide the fact that different providers treat your mail very differently. Gmail, Google Workspace, Microsoft 365, Microsoft Consumer, and Yahoo/AOL each apply their own filtering, tabbing, and reputation logic.

Seventh Sense's platform surfaces exactly this view — an "Audience reach by inbox provider" panel that shows the share of your audience on each provider alongside a health rating (Excellent, Good, Needs work, Poor), and an "Email metrics by inbox provider" breakdown covering delivery rate, open rate, click rate, bounce rate, and unsubscribe rate per provider.

To diagnose in Marketo itself, segment your smart lists or reports by email domain:

1. Build smart lists filtered on Email Address containing gmail.com, googlemail.com, outlook.com, hotmail.com, yahoo.com, aol.com, and your own corporate domains.
2. Run an Email Performance Report for each segment over the same date range.
3. Compare open and click rates across segments. A provider that is 20+ points below your average is a placement red flag.
4. Cross-check bounce and unsubscribe rates per segment — a provider with rising bounces is throttling or rejecting you.

If Gmail looks healthy but Microsoft Consumer and Yahoo/AOL look poor, you have a provider-specific reputation problem, not a global one. That distinction changes what you fix.

Read bounce rate, unsubscribe rate, and engagement decay as early warnings

Deliverability rarely collapses overnight. It decays through a predictable sequence:

• Rising soft bounces on a specific provider — the server is temporarily deferring you, often a reputation signal.
• Rising hard bounces — stale list data; you're sending to addresses that no longer exist.
• Rising unsubscribe rate — your content or frequency is mismatched to the segment.
• Falling open and click rates on previously engaged contacts — the classic list-fatigue pattern.

The Seventh Sense framing is useful here: "Every email list has two kinds of people: the ones engaging, and the ones being worn out." Sending the same cadence to both groups is what drives the decay above. When you keep mailing worn-out contacts, their non-engagement drags down your sender reputation, which then hurts delivery to your engaged contacts too.

Reduce list fatigue with send-time optimization and audience recycling

Two operational levers address the root cause rather than the symptom:

Send-time optimization sends each contact's email at the time they are most likely to engage, based on their individual historical behavior rather than a single batch send. Seventh Sense describes this as per-person send-time optimization. The mechanism: instead of one send event for the whole list, the platform schedules each recipient's message at their predicted high-engagement window. Higher engagement per send signals to inbox providers that your mail is wanted, which supports placement over time.

Audience recycling identifies contacts who have stopped engaging and routes them out of your active send stream — either to a re-engagement track or to a suppression segment — so you stop burning reputation on people who aren't reading. Seventh Sense lists audience recycling as one of its core functions alongside deliverability monitoring.

Neither is a one-time fix. Both are ongoing operational practices that keep your engaged audience healthy and your reputation from being dragged down by the disengaged portion.

Concrete troubleshooting steps

Work through these in order, since each step changes what the next one tells you:

1. Authenticate your sending domain. Confirm SPF, DKIM, and DMARC are correctly configured for every domain you send from in Marketo. Misalignment is one of the most common causes of silent spam placement. Verify with your DNS provider and a DMARC reporting tool.
2. Separate placement from delivery. Pull per-provider open and click rates. Any provider far below your average is a placement suspect.
3. Check your bounce and unsubscribe trends per provider, not just in aggregate.
4. Identify and suppress chronically inactive contacts. Define "inactive" by your own data (for example, no opens or clicks in 90–180 days) and stop mailing them on your main cadence.
5. Move to per-person send timing rather than a single batch send, so engagement signals improve.
6. Recycle disengaged contacts into a separate re-engagement or suppression track.
7. Re-measure per provider after 30–60 days and compare against your baseline.

What to expect

Seventh Sense reports a range of +7–100% email performance improvement across its customer base, with 500M+ personalized emails orchestrated and scheduled monthly and 12 years serving HubSpot and Marketo customers. Those are vendor-reported figures, so treat them as directional rather than guaranteed — your results depend on list health, content relevance, and how much of your problem is placement versus fatigue.

The core reframe is the one worth keeping: your dashboard's delivered rate is a server-level fact, not a performance metric. Diagnose with per-provider click and bounce data, fix authentication first, then reduce fatigue through send-time optimization and audience recycling.

An AI email assistant is a tool that works in the background of your existing inbox: it sorts incoming messages and writes reply drafts that are waiting for you to review and send. EmailTriager, for example, plugs into Gmail, trains on your past emails, and produces drafts in your tone rather than a generic one. It fits best if you already live in Gmail, receive a steady volume of repeatable questions, and want to keep your current workflow instead of adopting a new email client.

What the assistant actually does

Two jobs run automatically, without you opening a separate app:

• Organizes incoming email. New messages are triaged as they arrive, so the inbox is grouped by what needs a response rather than sitting as one undifferentiated list.
• Drafts ready-to-send replies. For messages that need an answer, a draft is written and placed where you'd normally reply. You review, edit if needed, and hit send.

The point is to convert email from "write everything from scratch" into "review and approve." EmailTriager frames this as turning email into "a simple task of reviewing and sending drafts."

How the drafts end up sounding like you

The mechanism is training on your own history, not on a prompt you write each time.

EmailTriager's "True Voice" feature learns from your past emails, so drafts pick up your phrasing and your sign-off. In the product's own example, a support question about crediting an image gets a draft that opens with the customer's name, answers the question directly, supplies the exact credit format, and closes with the sender's usual sign-off — all before the human touches it.

The practical consequence: there's no model-training step for you to perform. You don't write instructions about how you like to sound. The tool infers it from mail you've already sent.

What setup looks like

1. Connect your inbox. EmailTriager plugs directly into Gmail, so drafts appear inside Gmail rather than in a separate dashboard.
2. Let it train on past emails. This happens automatically; the product describes it as "plug and play" with no learning curve.
3. Review incoming drafts. When a message needs a reply, a pre-written draft is waiting.
4. Edit and send. You stay the final checkpoint — nothing goes out until you send it.

Because the drafts surface in Gmail, you aren't asked to change your email workflow or learn a new interface.

Privacy and data handling

Two claims matter if you're deciding whether to connect a real inbox:

• Your emails are not used to train general AI models. EmailTriager states this explicitly.
• Third-party security verification. The company says it has been verified by a Google-designated third-party security auditor and holds CASA Tier 2 accreditation.

If your organization has its own data-handling requirements, treat these as the vendor's stated position and check them against your own policy before connecting a work account.

Who this suits — and who it doesn't

A good fit if:

• You use Gmail as your primary inbox.
• A meaningful share of your email is repetitive — support questions, routine requests, standard follow-ups.
• You want to stay in control of what's sent, with review rather than full automation.

A weaker fit if:

• You don't use Gmail, since the product's described integration is Gmail-specific.
• Your email is mostly one-off, high-stakes correspondence where a draft saves little time.
• You need fully autonomous sending with no human review — the described flow is review-then-send.

FAQ

Do I have to train the AI myself? No. It trains on your past emails automatically; the product describes the setup as plug and play.

Will the drafts sound generic? The stated goal is the opposite — drafts are generated in your voice from your own sent mail, including your sign-off.

Does it send email without me? No. Drafts are prepared for you to review and send.

Where do the drafts appear? Directly in Gmail, so your existing workflow stays intact.

Are my emails used to train other AI models? EmailTriager says your emails are never used to train general AI models, and that it holds CASA Tier 2 accreditation from a Google-designated third-party auditor.

For current plan details and any usage limits, check the vendor's pricing page directly, since terms can change.

Send time optimization (STO) predicts the moment each individual recipient is most likely to open or click an email, then schedules that person's send for that moment instead of releasing the whole list at one fixed time. It fits teams sending recurring or lifecycle email to a list with enough engagement history to learn from — typically HubSpot and Marketo users who want higher open and click rates without rewriting content. It works less well on brand-new contacts, very small lists, or shared inboxes where one address represents many people.

How it differs from a fixed send time

A fixed send time treats the list as one audience: everyone gets the email at 9 a.m. Tuesday. STO treats the list as many individuals and spreads delivery across the day, so each recipient's message lands near their own active window.

The practical effect is that a single campaign no longer arrives as one batch. Opens and clicks can rise because timing improves, not because the subject line or offer changed. Seventh Sense describes this as per-person send-time optimization and reports email performance improvement in a range of +7–100% across customers, with 500M+ personalized emails orchestrated and scheduled monthly. Treat that range as a vendor-reported outcome that varies by list, not a guaranteed lift.

What signals feed the prediction

The model needs evidence of when a person actually engages. The signals that matter most:

• Engagement history — when this contact has opened or clicked past emails. This is the strongest input; without it the prediction is close to a guess.
• Timezone — a 9 a.m. send is a different local hour for each recipient, so timezone corrects the raw timestamp into the person's own day.
• Inbox provider signals — where the recipient's mail is hosted. Seventh Sense surfaces audience reach by inbox provider (Gmail, Google Workspace, Microsoft Office 365, Microsoft Consumer, AOL/Yahoo, Orange, and others) and breaks delivery, open, click, bounce, and unsubscribe rates down by provider. Provider context helps because filtering and engagement behavior differ across Gmail, Office 365, and consumer Microsoft, so the same send time does not behave identically everywhere.

A useful way to picture it: a contact who reliably opens between 7 and 8 p.m. local time, on a Gmail address, gets scheduled into that window — while a colleague on the same list who opens at 7 a.m. on Office 365 gets a different slot from the same campaign.

What changes in practice

• Sends spread across the day rather than firing in one batch.
• Each recipient's message is timed to their own predicted window.
• Open and click rates can improve with no change to content, list, or offer.
• Reporting has to be read per person and per provider, not just as one campaign-level number.

Limits and failure modes

STO is a prediction, and predictions degrade when the inputs are thin:

• New contacts with no engagement history — the model has nothing personal to learn from, so early sends fall back to weaker signals.
• Small lists — too few events per person to establish a reliable pattern.
• Shared or role inboxes (info@, sales@) — one address stands in for several people with different habits, so a single "best time" is a compromise.
• Lists with declining engagement — if people have stopped opening, timing alone will not revive them. Seventh Sense frames this as two kinds of people on every list: the ones engaging and the ones being worn out, and pairs STO with audience recycling and deliverability monitoring for that reason.

Expect results to vary by list. A large, long-lived, actively engaged list is where STO shows the clearest effect; a cold or brand-new list is where it shows the least.

How to measure it before rolling out

Do not switch the whole audience at once. Run a controlled comparison:

1. Split the audience into a test group (STO on) and a control group (fixed send time), matched on engagement level and provider mix.
2. Hold content constant — same campaign, same subject line, same offer — so timing is the only variable.
3. Run for several sends, not one, since a single campaign is noisy.
4. Compare open rate and click rate between the two groups, and check bounce and unsubscribe rates to confirm the spread-out delivery did not hurt deliverability.
5. Check results by inbox provider as well as overall, because a lift on Gmail may not appear on Office 365.
6. Expand to the full audience only if the test group wins on the metrics you care about.

If the test group does not beat the control, the list likely lacks the engagement history STO needs — fix data quality or engagement first.

HubSpot's "delivered" metric counts messages the receiving server accepted — not messages that landed in the inbox. A 99.8% delivered rate is fully compatible with a large share of your list sitting in spam, promotions, or the "other" tab. To fix it, stop treating delivery as the finish line and start diagnosing placement and engagement by inbox provider. This applies to anyone sending through HubSpot who sees healthy delivery numbers alongside weak opens and clicks.

What "Delivered" Actually Measures in HubSpot

When HubSpot reports an email as delivered, it means the recipient's mail server returned an acceptance response (a 250-level SMTP reply). That handshake happens before the provider decides where the message goes. Gmail, Microsoft, Yahoo, and others then run their own filtering — spam classification, tab placement, throttling — after acceptance.

So the gap looks like this:

The dashboard line "Your dashboard says 99.8% delivered. Your audience says 'I never saw it.'" describes exactly this. Delivery is a transport metric; placement is a filtering outcome. They are not the same thing.

Break Down Performance by Inbox Provider

The single most useful diagnostic is segmenting your metrics by where recipients actually read mail. Seventh Sense's inbox provider analytics break an audience down like this (illustrative figures from its dashboard):

• Gmail — 50.17%
• Microsoft Consumer — 21.64%
• Verizon / AOL / Yahoo — 18.36%
• Google Workspace — 1.27%
• Apple Mail — 0.93%
• Orange — 0.25%
• Other / Unknown — 6.77%

The same view also rates each provider's health — for example "Excellent" for Gmail, "Good" for Google Workspace and Office 365, "Needs work" for Microsoft Consumer and AOL/Yahoo, "Poor" for Orange and unknown providers. When one provider's open rate collapses while others hold steady, you have a provider-specific filtering problem, not a content problem.

How to pull this yourself

1. Export your HubSpot email performance by contact domain (or use a deliverability tool that maps contacts to inbox providers).
2. Group recipients into Gmail, Google Workspace, Microsoft 365, Microsoft Consumer, Yahoo/AOL, and "other."
3. Compare open rate, click rate, bounce rate, and unsubscribe rate per group rather than as one blended number.
4. Flag any provider whose open rate is materially below your account average.

A blended 22% open rate can hide a 35% Gmail open rate and a 6% Yahoo open rate. The second number is the one that needs action.

Common Causes of HubSpot Deliverability Decline

1. Worn-out, non-engaging contacts

Every list contains two groups: people who engage and people being worn out by repeated sends. Sending to the second group drags down your engagement signals, and providers read low engagement as a spam signal. The more you mail people who never open, the worse your placement gets for everyone on the list.

2. Authentication gaps

SPF, DKIM, and DMARC records tell providers your mail is legitimately from you. Missing or misconfigured records push messages toward spam or rejection, especially at Gmail and Microsoft. Verify all three are published and aligned for your sending domain before blaming content.

3. Sending to unengaged segments

Continuing to mail contacts who haven't opened in months — without suppression or re-engagement — trains providers to distrust your domain. Volume sent to cold contacts is one of the fastest ways to degrade placement.

Step-by-Step Diagnostic Workflow

1. Check bounce and unsubscribe rates by provider. A rising bounce rate at one provider signals list-quality or reputation issues there specifically.
2. Isolate low-engagement contacts. Build a HubSpot list of contacts with no opens or clicks in the last 90–180 days.
3. Run a reduced test send to a clean segment. Send your next campaign only to recently engaged contacts and compare open rate by provider against your normal full-list send. If the clean-segment open rate is much higher, the problem is list hygiene, not content.
4. Confirm authentication. Re-check SPF/DKIM/DMARC alignment for the sending domain.
5. Compare timing. If opens are low across the board but placement looks fine, the issue may be when you send rather than whether it arrives.

Remediation Actions

• Suppress or recycle chronically unengaged contacts. Stop mailing people who never engage, or move them into a separate re-engagement track. Recycling — routing worn-out contacts away from your main sends — protects the engagement signals of your active audience.
• Warm domains and IPs gradually. If you've added a new sending domain or IP, ramp volume slowly rather than blasting your full list.
• Adjust send timing per recipient. Per-person send-time optimization delivers each message when that individual is most likely to open, which lifts engagement signals and, over time, placement.
• Fix authentication records if any of SPF, DKIM, or DMARC are missing or misaligned.

Metrics to Monitor After Fixes

Confirm the fix worked by tracking these over the following weeks:

• Open rate by inbox provider — the primary signal that placement improved.
• Risk index — a composite health score; watch for it moving from High toward Medium or Low.
• Bounce rate — should fall as list quality improves.
• Click rate and click-through — engagement depth, not just opens.
• Unsubscribe rate — a spike can mean you're still over-mailing.

Seventh Sense reports a risk index that shifts categories (for example, from High to Medium) and an influence score benchmarked against the prior period, which gives you a before/after read on whether remediation actually moved placement.

The Short Version

HubSpot's delivered metric stops at server acceptance. Real inbox placement depends on engagement, authentication, and list hygiene — and it varies by provider. Diagnose by breaking performance down per inbox provider, isolate your non-engaging contacts, verify authentication, then suppress or recycle the worn-out segment and optimize send timing. Track open rate by provider and your risk index to confirm the fix held.

Email engagement is the set of active behaviors recipients take after an email reaches their inbox — opening, clicking, replying, forwarding, or converting — as opposed to passive delivery. A 99.8% delivered rate tells you the message left your system and reached a mail server; it says nothing about whether a human saw it, let alone acted on it. To measure engagement properly, you track click-through, reply, and unsubscribe behavior over time, and you break those metrics down by inbox provider so you can see which segments are actually engaging and which are being worn out.

Why "Delivered" Is Not Engagement

Delivery is a server-to-server handshake. Engagement is a human decision. The gap between them is where most email programs quietly fail.

A dashboard can report near-perfect delivery while the recipient's inbox shows thousands of unread messages. Seventh Sense frames this exact contradiction: "Your dashboard says 99.8% delivered. Your audience says 'I never saw it.'" The message arrived; the person did not engage.

This matters because inbox providers (Gmail, Microsoft, Yahoo, and others) increasingly weight engagement signals when deciding whether to place your future mail in the inbox or the spam folder. Low engagement doesn't just mean fewer opens today — it erodes your ability to reach that person tomorrow.

The Engagement Signals That Actually Count

Open rates are the weakest signal, because privacy features in Apple Mail and others now pre-fetch pixels, inflating opens without a human reading anything. Weight your measurement toward behaviors a person must actively perform:

Track these per contact, not just per campaign. A single campaign's 40% open rate can hide the fact that the same 200 people open everything while 8,000 never do.

Break Engagement Down by Inbox Provider

Aggregate metrics average away the problem. The same email can perform well with one provider's users and poorly with another's, and you won't see it until you split the data.

Seventh Sense's inbox provider analytics show this directly — a dashboard breaking down delivery rate, open rate, click rate, bounce rate, and unsubscribe by provider:

• Gmail — 42% of audience, rated "Excellent"
• Google Workspace — 28%, "Good"
• Microsoft Office 365 — 22%, "Good"
• Microsoft Consumer — 14%, "Needs work"
• AOL / Yahoo — 8%, "Needs work"
• Orange — 4%, "Poor"
• Other / Unknown — 3%, "Poor"

The insight isn't the numbers themselves — it's that a provider rated "Needs work" or "Poor" is telling you something specific about that segment. Maybe those recipients are stale, maybe they've been over-mailed, or maybe deliverability to that provider is degrading. You can only act on it if you can see it.

Separate Engaged From Worn-Out Audiences

Every list contains two populations: people who engage and people being worn out by mail they ignore. Treating them identically is the core operational mistake.

Seventh Sense describes its AI as knowing "which is which — and what to do about it." The practical decision framework:

• Engaged contacts — keep sending at their preferred cadence; these are your deliverability assets.
• Worn-out contacts — reduce frequency, pause, or recycle them out of active sends before their non-engagement drags down your sender reputation.
• Unknown contacts — test with a re-engagement send; if they don't respond, move them to the worn-out group.

Audience recycling — removing or suppressing chronically non-engaging contacts — protects the engagement rate of everyone else. A smaller, more engaged list typically outperforms a large, decaying one.

A Practical Measurement Routine

1. Pick your core metrics. Click rate, reply rate, and unsubscribe rate per contact — not just opens.
2. Segment by inbox provider. Use your platform's provider breakdown (Seventh Sense provides this natively for HubSpot and Marketo) to spot segments rated below "Good."
3. Track engagement over time per contact. Flag anyone whose opens and clicks have declined across recent sends.
4. Classify into engaged / worn-out / unknown. Set thresholds that fit your cadence — for example, no click in 90 days marks a contact as worn out.
5. Act on the classification. Reduce frequency or recycle worn-out contacts; protect engaged ones.
6. Re-measure. Watch whether overall engagement rate and provider health improve after recycling.

Common Traps

• Trusting opens alone. Pixel pre-fetching makes opens unreliable as a human signal.
• Averaging across providers. A strong Gmail segment can mask a failing Microsoft Consumer segment.
• Never recycling. Keeping non-engagers on the list feels safer but actively harms deliverability for everyone.
• Measuring campaigns, not people. Campaign-level metrics hide per-contact engagement decay.

Where This Fits

Seventh Sense is built specifically for HubSpot and Marketo, combining per-person send-time optimization, deliverability monitoring, and audience recycling in one place. Its stated results include 500M+ personalized emails orchestrated monthly and a +7–100% email performance improvement range across customers. If you run email on HubSpot or Marketo and your delivered rate looks healthy while engagement doesn't, provider-level engagement analytics and audience recycling are the two levers most likely to close that gap.