SPF and DKIM are valid and email still lands in spam. What actually decides it: list hygiene, content, frequency and domain warm-up - and the order to check them in.
Key takeaways
- An old list is the number one cause of falling deliverability, even with everything configured correctly.
- Hebrew does not cause spam filtering; sending to dead addresses does.
- A sharp jump in volume is what triggers filtering, not the wording of the email.
- The fix is nearly always to send less, to people who want it - not to change words.
When SPF and DKIM are valid and email still lands in spam, the problem is almost never technical but behavioural: who is on the list, how many of them open, and what happens when you send to a large volume at once. The mail provider is watching recipient behaviour, not the language.
What to check, in order
| Check | How | What it reveals |
|---|---|---|
| Bounce rate | Platform report, last send | Dead addresses that must be removed |
| Open rate over time | Compare the last three sends | A gradual decline means the list is ageing |
| Volume versus history | This month's send volume against previous | A sharp jump triggers filtering |
| Where subscribers came from | The origin of the list | A purchased or very old list |
| Spam complaints | The platform report | A sign frequency or expectation is wrong |
| Links | Which domain they point at | A generic link shortener hurts trust |
The first row explains most cases and is also the easiest to fix: an address that hard-bounces twice must leave the list immediately.
Why does an old list hurt so much?
Because mail providers measure you by what recipients do. A four-year-old list contains addresses that no longer exist, addresses that have become traps, and people who forgot they subscribed - and together those three produce exactly the signals that push you into spam.
The fix is not sending more to "wake them up" but the opposite: reducing to the list that opens. In practice, sending to 2,000 active recipients reaches more inboxes than sending to 10,000 that are mostly dead - which is also why businesses report an immediate improvement after cleaning.
Cleaning a list without deleting customers
- Remove hard-bounced addresses immediately, with no deliberation.
- Flag anyone who has not opened in a year as inactive, but do not delete them.
- Send one re-engagement campaign to that group, with a direct subject line and an option to stay.
- Remove whoever did not respond to that either.
- Keep the removed list separately, so you can recognise them if they sign up again.
Step 3 matters because it turns removal into the recipient's choice rather than your decision, and it recovers the small share who genuinely wanted to stay.
Warming up after a break or a migration
If you have not sent for six months, or you changed platform or domain, sending immediately to the whole list is the reliable route to spam. What works is gradual: start with recipients who opened recently, and grow the volume across several sends.
The logic is simple: the mail provider builds a reputation for the sending domain from history. A small send with good opens builds reputation; a large send with no history looks exactly like bulk mail from an unknown party.
What in the content actually matters?
Less than people think, but three things do:
- A subject line promising something different from the content - it produces fast deletes and complaints.
- An email that is one large image with no text - it looks suspicious, and it is unreadable when images are blocked.
- Too many links, particularly to generic shorteners anyone can use.
What does not matter: Hebrew words, a single exclamation mark, or one emoji. Myths about "spam words" are mostly leftovers from old filtering rules, and they distract from the thing that actually decides it - whether people open.
What happens when a business sends from two systems
This is a common and unrecognised fault: the marketing platform sends from one domain, and the invoicing system or CRM sends from another domain or server. When one of them is misconfigured, the reputation of the whole domain suffers - including the operational email that must arrive.
The way to find it is to count every place email leaves in the business's name: marketing, invoices, order confirmations, reminders, website forms. In most businesses the answer is three to five sources, and some were never checked. The technical settings each of them needs are covered in SPF, DKIM and DMARC operations.
Transactional versus marketing, and why it matters here
Transactional email - an order confirmation, a receipt, an appointment reminder - goes to someone who took an action, so its open rates are high and spam complaints are rare. Marketing email goes to a list, which makes it the one carrying the risk.
The practical consequence is separating them at the sending level: a subdomain, or at least a separate system, for marketing, so a wave of complaints on one campaign does not degrade an order confirmation's ability to arrive. Businesses that do not separate discover this the hard way - on the day a customer pays and receives no confirmation, and the conversation that follows costs more than the campaign earned.
It is also worth making sure transactional email comes from an address that can be replied to. A no-reply address increases complaints, because a recipient who wants to ask something presses whatever is available - and that is often the spam button.
A one-week sequence
- Remove hard-bounced addresses from every list, in every tool.
- Count the sending sources in the business's name and confirm all are configured.
- Segment the list into active and inactive by opens in the last year.
- Send only to actives on the next campaign, and compare the open rate.
- Decide a fixed frequency and stick to it.
The comparison in step 4 is what settles the argument: when the same content to the same audience produces a completely different open rate, the problem is clearly the list rather than the copy.
Sources
Frequently asked questions
How long does it take to repair a damaged reputation?
There is no definite number, because it is built from behaviour over time. What is known is that the direction depends on what you do from now: small sends to active recipients improve it gradually, while continuing to send to the whole list preserves the situation. There is no shortcut around that.
Should marketing go from a separate domain?
It is a common approach: a dedicated subdomain for marketing, so a problem there does not damage operational email. It adds a one-off configuration and separates the risks, and it is particularly worth considering at higher volumes.
What about info@ and office@ addresses?
They tend towards low open rates and high complaint rates, so they are worth managing separately. Do not delete them - they are real customers - but do not include them in campaigns whose purpose is preserving sender reputation.
How do I know whether I reached the inbox or spam?
Platforms report delivery rather than placement, so the practical signal is an open rate that collapses with nothing else changing. It is also worth keeping test mailboxes with several providers and sending every campaign to them, to see with your own eyes where it landed.
Keep reading
Related service
Business Automation
I build custom automations that remove repetitive work end to end.
About the author
Yehonatan Saadia
Freelance automation, web & MVP developer
I'm Yehonatan Saadia, a senior developer who builds business automation, custom websites, and MVPs for small and mid-sized companies across the US, Europe, and Israel. These guides come from real client work, not theory.
Work with meHave a project like this?
Tell me what you're trying to automate or build and I'll tell you the fastest reliable way to ship it.
