We have a triggered send for sending order confirmation emails that we fire over API from our ecommerce platform. The volume isn't huge (2-4k emails per day), but after the triggered send is live for a few days the emails start getting delayed as much as 2-3 days later.
We ended up disabling the triggered send and started sending those emails directly through SendGrid.
There are some unique things about this email as opposed to our other emails, namely that we are sending over 100 subscriber attributes to this particular email (we send chunked HTML content for our order confirmation email listing), as well that we have some semi complicated Ampscript logic that, depending on which Ecommerce server made the request, may render different content in the email.
The scenario has these characteristics: 1. There is no delay if there is no load on the system (i.e. we're sending emails out from our test environment, not a production load). 2. The delay starts to mount up after the first day of being live (1-3 hour delays on the email). 3. By about day 2 or 3 of being live, the emails take 2-3 days to delivery. 4. The issue does not happen with any of our other triggered sends (of which we have a few dozen).
We've been in contact with Marketing cloud support, for a few months now regarding this issue, but they keep coming back with the same responses:
They tell us:
- Its an issue with your sender IP, separate out your marketing from transactional email sender IP addresses
- Its an issue with the ISPs (Gmail, Yahoo, Microsoft, our own internal email servers)
These responses don't make sense to us for the following reasons:
- Shouldn't it happen with all of our emails if its a SPAM issue (Sender IP suggested fix, also, we have been focusing almost entirely on transactional emails lately and haven't run very much marketing content through the system yet).
- It's even an issue with our internal email servers, where we have whitelisted and greylisted the Sender IP
- The delay becomes longer the more emails we throw at the system
Salesforce support swears that it isn't a queuing issue on their end though. They assure us that the a few seconds after our requests are coming in, they are forwarding them outbound.
At this point we'd be open to hiring an outside consultant if they can help us solve the problem. Any advice the community could give us would be helpful? Are we on the right track? Does the premier support response make sense versus what we're experiencing?