We tested a random sample of 95,000 global, English-language campaigns over the last 12 months (for a total of 2.2bn emails), and have isolated 287 popular ‘trigger words’.

Then, split by sector, we looked at the correlation between the word’s inclusion in the subject line and its variance above or below the average results for key email metrics (Open Rate, CTR, CTOR, and Unsubscribe).

To ensure outliers aren’t confusing things, we’ve also looked at the first, second and third quartiles to give an indication of data spread, not just nominal long-run means.

Bear in mind, these relationships are correlated, but not necessarily causal. There are simply too many variables in an email campaign to pinpoint exact causation. But, where there’s smoke, there’s fire, and at the very least, this will give you many ideas for what to test.

And now some key findings:

People love free stuff

In related news, water is wet. Specific trigger words have a drastic effect on the response rates of offers. For example, ‘Sale’ delivers +23.2% opens, but ‘Save’ only +3.4%.

However, in terms of click throughs, they give +60.7% and -25.2% respectively. When you promote your offers, consider testing those trigger words – it could make or break your campaigns.

And, consider simple quick wins like ‘Free Delivery’ or even ‘% off’ – in the right context these keywords can drive massive response uplifts. But, if you do nothing but hard sell and offer discounts all the time, your customers will become bored and tune out.

Mix up your offer emails in a series of value-adding campaigns.

Content marketing works when the content isn’t crap

The problem with content marketing is the vast majority of content produced is crap. Too many people have outsourced it to agencies that don’t know enough about their clients’ markets, and focus on the wrong metrics. As a result, consumers have become anesthetised to content.

Take, for example, ‘Report’ (-23.7% opens, -54.8% CTR) and Webinar’ (-16.6%, -70.7%.)

Conversely if the content is good, people will consume it. So ‘News’ (+34.8%, +47.7%), ‘Bulletin’ (+15.8%, +12.7%) and ‘Video’ (+18.5%, +64.8%) work well.

Simply put, if your content is crap, it won’t work. If your content is good, you’ll get great results. When doing content marketing, make sure you’re in the latter group, not the former, or else you’ll be what we affectionately call in the industry a “spamming %$*!$#“.

More frequent emails are better than less frequent

Typical for an email guy to say, right? Full disclosure, when people send out more emails, ESPs make more money, this is true. But, if emails drive response and therefore generate revenue, then what’s the problem?

But, don’t take my word for it. We looked at newsletter frequency, and specifically the trigger words that indicate it. ‘Monthly’ brought -26.6% opens and -37.0% clicks. ‘Weekly’ brought +27.1% and +50.6% (not bad) and, amazingly, ‘Daily’ brought +27.8% and +100.3%. Simply put, more email drives more response.

And it’s not just us saying this, check this out if you’re sceptical.

Personalisation works, if your data is clean

Lots of people have played around with subject line personalisation, with varied results. We found the average opens and clicks decrease (-20.7% opens). But, the spread of the data is massive. The first quartile is -73.1%, but the third quartile is +30.8%. Why is this?

The main thing about personalisation is to ensure you don’t deliver users a disjointed user experience. If the subject line is personalised, but the email content isn’t, guess what? You may gain opens, but have done nothing to drive clicks. It leaves users with a negative experience.

Furthermore, if the subsequent online journey isn’t equally personalised, once again you’ve left your users feeling like you’ve given them the old bait-n-switch.

Most importantly, if your database is old and potentially incorrect, you’re in trouble. Eyeball your data first and pick recent, engaged data for testing personalised content.

There’s always money in the banana stand

This report analysed a vast sample of subject lines. More than anything, it showed there is lots of short-run variance in every keyword sample. What matters is that you test things out to your lists, and never stay standing still. What works one week may not work the next, but if you aren’t trying out new things then you’re treading water and ultimately throwing away money.

Email still delivers the strongest revenue of any digital channel by a country mile (source). In all your tweeting, pinning, and facebooking, are you making sure you dedicate enough time to your email subject lines?

Remember – of Obama’s 30 person digital team, 24 of them were focused on email.

They knew that there’s always money in the banana stand.

To see the full list of keywords (287 common words were tested in addition to the ones above), you can download the report. (Registration required – Ed)