Hacker News Reader

Hacker News

Comment by mbesto | Hacker News Reader

Parent Full thread

mbesto

> I assume that returns are diminishing, and turn negative past a certain point.

Which is why you should test it.

Jan '22 -> once a month

Feb '22 -> twice a month

Mar '22 -> thrice a month

Jan '23 -> once a month

Feb '23 -> twice a month

Mar '23 -> thrice a month

Compare and contrast for both unsubscribe rate and conversion.

View on HN

No, just no. AB testing how much you can squeeze out of the channel has a huge negative backlash. A customer driven "send these more often" or "send these less often" offer in the email will give the customer the ability to set the level of engagement that works best for them and avoid both rage quits and user hostility to being "worked."

A friend of mine was hired by Peet's to try to clean up the mess they caused by doing this to their previously loyal fan base, it has been a lot of work and expense.

> A friend of mine was hired by Peet's to try to clean up the mess they caused by doing this to their previously loyal fan base

Clean up what? You test it once and if it works, it works, if it doesn't you revert back. You're not going to get a huge backlash of brand loyalty for increasing a monthly newsletter to 2x/month. And if you get ANY backlash you simply revert back. Most people wouldn't even notice.

I think you're overthinking my recommendation.

> A customer driven "send these more often" or "send these less often" offer in the email will give the customer the ability to set the level of engagement that works best for them

You should have this option regardless of what techniques you use.

> And if you get ANY backlash you simply revert back. Most people wouldn't even notice.

The people who dropped the newsletter after it became spam definitely won't notice that you reverted back to the expected behavior.

It depends on the size of that group and their sensitivity. I’ve found that consumers generally don’t think as much about it as tech people - it’s much more random than we like to think.

If your test costs 2% of newsletter receipients who were 2 emails away from unsubscribing, it might be worth it.

As always, know your audience.

This works until the people who get 3 emails a month unsubscribe. Don't break it if it works.

Probably better to use a split test. Segment the list into two segments (via random assignment). Send one segment two emails per month, send the other one per month. Compare the segments.

True. I would then test that over a 6+ month interval. Depending on the volume of subscribers the results may not be correlative.

I work for an ecommerce company.

The unsubscribe rate doesn't really change, it's fairly constant at 0.3% of the audience every time.

There are obviously diminishing returns on the revenue with frequencies. Our business is moderately seasonal, so the numbers are not that accurate on the revenue side.

I think a majority of people don't unsubscribe; they just mark your emails as spam and ignore them. That lowers your credibility with email providers, and if that happens enough, you start seeing your emails being considered as spam and you get entirely blocked off the ISP.

Also, it may be a mistake to attribute all of your revenue from the email to email. By its nature, emails are going to the segment of your audience most interested and most likely to purchase. Only a portion of revenue from it is actually incremental; a large part probably would have ordered anyway, and the email gets credit only because they're also clicking through the email.

Of course, it depends on what offers are in the email, etc. If you want to test, the best bet is as was mentioned elsewhere in this thread, segment out a set of customers and change the frequency of emails over the long term, and track their total sales.

Yep, you need to track your open rates too. If you have a 20% open rate, run a daily spam trial, and find you had a 0.3% unsub rate but your open rate is now 10%... You actually had a 50% unsub rate, your customers just aren't letting you know they've unsubscribed

Isn’t open rate obfuscated / blocked by most clients now? How do you measure it?

(I’ve been blocking them for 25+ years; might be out of touch…)

open rates are inaccurate but still somewhat useful because they’re generally consistently inaccurate month to month unless there’s been a change that would affect a lot of subscribers (eg iOS recently).

click rates are probably a better proxy for this.

Saw this comment late, like tl_donson says, yes but it's fairly consistent. So if your open rate drops 20% to 10% in one month, it's probably your fault and not the mail client hiding it.

In my experience most people do not block images in emails, and that's the easiest way to tell if someone opened your mail. Even most people who use an adblocker do not block email images.

You're talking in general, and most of your points are correct, but they don't apply to our business fully.

For example, our newsletters are always offering a time-limited discount for the purchase. They're also not segmented out to people "most likely to purchase", because we don't have that data. It goes out to 100% of our audience every time, unless they made a very recent purchase.

If we knew who the people are that are most likely to buy, it sure as hell would make the business easier to run.

Fries a month seems like the sweet spot. Especially with a drink. And a burger. Where do I subscribe?!

Reply on news.ycombinator.com