Any advice on budgeting email volume/memory use at scale?

Hi. I’m curious if there’s any consensus or advice on how to plan for operational costs at scale. We’re a few weeks away from opening up a private—paid—community to beta testers and we’d like to avoid having to come back to our members in a year or two with a price increase to cover our costs.

Right now, we’re on the lowest/smallest tier with Digital Ocean and the free MailGun account. While we’ve grown from two to seven users since we started building in August, I don’t feel averaging things down to daily numbers is entirely accurate—and this is proving trickier to research than expected.

Are there any reasonably safe assumptions we could use to ensure we’re prepared to cover our bills at 100, 500, 1,000, 5,000 members? Does anyone have any kind of basic, statistical info on this subject?

For example:

  • What percentage of users would you expect to see active? 5%? 10%? 30%?
  • How many emails would you expect to see sent per user?
  • How much memory is required given the above?

(I know all the above depend on wildly fluctuating variables, but it’s a thought exercise.)

From what I’ve seen, we’d need 150 paying members to cover out bets at US$500/mo, or 300 to cover $1,000/mo. That tells me we should be fine at $50/yr, but thus far, I’m still very green on converting data to actionable insights and I’m pretty sure my existing sample size isn’t up to the job.

Any insights or advice would be appreciate. Thank you.


You’re not going to have any trouble. There are a zillion variables that make answering your questions fairly impossible, but I’d guess that you can support multiple hundreds of monthly users on a 1 or 2 GB droplet (I’d go ahead and bump it to the 2GB-$10/month droplet) and likely not get >10K emails. At the rates you’re charging, the hosting fees will be negligible.


It is difficult to say without knowing what you are trying to do and what operational activities you have to pay for.

USD 50 * 150 = 7,500 p.a. = ~600 p.m. will cover quite a lot if it is only used for technology/hosting costs.

I’d be more concerned getting realistic estimates of user engagement stats. Even 5% active users might be wildly optimistic.

As an aside, I have the question why a private, paid forum would care about active users when it comes to costs? If all users are paying.

There is a lot of work required to get active users, let alone paying users. So why would your value proposition be so attractive with a high conversion rate and a high subscription cost?

Checkout which has a lot of community managers. Feverbee say the average forum has 1 in 50 active users which is 2%.

Here are some stats from some real Discourse sites listed percentage of active users:

Coding education software 191/6.3k = 0.3%
Games software 14.4k/3.9M = 0.4%
Bank 39/3.9k = 1%
Security software shows 151/3.6k = 4.2%
System software 1.9k/43k = 4.3%
Forum software shows 2.1k/40.2k which is 5.2%
Craft making 533/6.6k = 6.7%
Home automation 5.4/81.2 = 6.7%
Community management 208/2.7k = 7.7%
Games software 1.2k/13.6k = 8.8%
Flight simulator software 5.6k/46.9k = 11.2%
Programming language 2k/12k =17%
3d printer product 6.5k/20.8k = 31%

Technically, you are on a free Mailgun plan but you are not on “the free MailGun plan”: The Free plan has a daily limit of 300 emails and no custom domains. Instead, you and I should be on the free Concept plan which is upgraded from the Free plan when you setup your payment details.


Thank you for the deep, thoughtful reply @Remah. Appreciate it greatly!

Your comments on engagement rates are on-point. Absent realistic numbers, I have no way of confidently predicting costs. From my experience with general sales and marketing, 1-3% is the norm, though most of our other, non-forum efforts tend to see 15-40% engagement, and a friend of mine with a 2,000-member Discourse community just told me this week he’s seeing 4,000 unique pageviews per day and enjoying 10% active membership. Granted, his community is publicly visible, which affects memory/bandwidth, but still, I’m doing the needful, as they say.

As to why a private, paid forum would care about active users (when all users are paying), we want to ensure our early adopters have a frictionless experience as long as possible so they can help us all focus on foundational improvements to our process and business model.

I’m an old forum admin from back in the day and I want to make sure the pictures posted today will still be there 10 years from now. Charging people membership means every member has skin in the game, incentive to make that extra stop away from Facebook, but it also means they’re our customers, too.

Chalk it up to personal pet peeve, but I can’t stand it when businesses offer something for free, then pull it behind a paywall. That’s bait-and-switch BS and this dude can’t abide. So I want to make sure that, if I’m charging people $50/yr for membership, I’ve priced things appropriately to cover the costs well into the future, when other revenue streams can pick up any slack, should community membership become something of a loss leader.

It’s all about delivering joy to our customers. That’s all. If I’m charging money—for a forum in the age of Facebook—I need to do everything in my power to make that experience worth every penny.

As to our value prop, we’ve not done a bit of paid promotion yet, but of the few hundred people we’ve been talking to on back channels, several who most reflect our target demographic have literally told us, “Shut up and take my money.” :slight_smile:

Online communities aren’t new. They were some of the first things built on the internet. Social media came along and promised us all global connection and incredible wonder—but ended up segmenting us all into neat, superficial, little boxes for more efficient, data-driven advertising.

People are getting sick of it. We want something more. We want real, actual connect.

Forums are about to make a comeback in a BIG way.

We’re banking on Discourse. :success:

Thanks again for all your help. It feels good being back on a forum. Real good.