Our regular sending will include 100,000 emails/week split between 2 days. For IP warm-up I'm ramping up our sending daily from 5,000/day to 15,000/day over 30 days sending every day and spreading out our sends. After this process is over, will it be harmful to our deliverability if we return to our standard 2x/week sends of around 50,000 emails with Journeys sending 500-1,000 emails on the other 5 days of each week?
general answer first:
Quoting a Salesforce document:
You generally should not send more than double your usual daily send volume from a given IP address, week-over-week. Remember! High volume spikes will cause deliverability problems.
So the answer is yes, in general. Warming means gradually approaching your real sendout behavior. If your normal sendout behavior is 50k twice per week, then you should try and reach that plateau, step by step. Stopping at 15k (40% of the total) while also spreading your sendouts could be too little. You should try and breach the gap between 15k (spread out) and 50k (at once).
It is good to create more send events for reputation buildup, so while you can, do that. But at some point you won't have enough addresses to keep that up.
Also, I would also always try and handle this by domain. You have rather small numbers to reach, but as a general tip, each provider has a different threshold to start with. For example you can find some sources that say google supports 20k from day1, and I have found that to be true. https://help.salesforce.com/articleView?id=sf.mc_es_ip_address_warming.htm&type=5 Look at your top 10 domains and try to stay within their starting numbers in week 1, work from that, doubling your volume.
So: Depending on your address pool, you could maybe setup a plan that is totally fine after two weeks judging your numbers, and your "spikes" might not even reach the beginning threshold in some cases. If you have 20 % gmail addresses, a "spike" of 100k per day will reach 20k addresses on gmail, which they tolerate from a cold IP. It could be that we are "worrying too much" for the load that is actually arriving at ISPs. It's never bad to be TOO careful though, so starting slowly isn't bad, if you don't stop increasing volume too early.
With all that said, for a simplified example. Assuming that you have 50k individual addresses, your normal patterns is two sendouts/ week, your domains are split across several providers as "usual", we have four weeks time and we ignore the split by domain that I just mentioned:
then maybe something like:
5k, 5k, 5k, 5k, 5k, pause on weekend
10k, 10k, 10k, 10k, 10k, pause on weekend
20k,x,x,20k,x,pause on weekend
40k, x, x, 40k, pause on weekend
real sending start of week 5.
Take this with a grain of salt, as I have no idea on your actual address base or surrounding factors. But the patterns and guidelines should help you plan out your warmup.