December 08, 2005
It's an agonizing decision. You slaved over the latest issue of your email newsletter, and yet, only perhaps 30% of recipients opened the darn thing. Should you resend to the 70% who didn't? On one hand these are permissioned names who said at sign-up that they really want your newsletter. On the other hand if you annoy anyone, it could damage your reputation and there goes the brand (let alone delivery past filters). Here's a new MarketingSherpa Case Study about a UK marketer who undertook this slightly dangerous test this fall. Includes samples and lots of useful data.
CHALLENGE Iggy Quazi, Managing Director Mouse2House, had a lovely house email list made up of thousands of customers who'd opted in to get site news.
However, his marketing team hadn't sent anything beyond order acknowledgements to the list for two years.
It's not as crazy as it seems. Mouse2House sells digital imaging supplies such as laser printer cartridges. Just as in the US, British email users' inboxes have been flooded with spam from less scrupulous vendors in that arena.
The Mouse2House team didn't want to send email until they had the time to properly strategize the direction, especially vis-à-vis the competition.
"We wanted email to establish the brand so it's recognized by the public as a trustworthy place to shop, as opposed to all the scamsters," explains Quazi.
CAMPAIGN First the team gathered up examples of every competitor mailing they could get their hands on -- hundreds in all.
"We discovered in this industry nobody sent a newsletter. What they sent was a promotional shot. 'Buy this thing at this price; here's a discount special; it's half price… blah blah blah.'"
This aha-moment was the key to the entire Mouse2House email program. "To make this email desirable to receive, it had to give them something other than sales offers. We decided to differentiate by being trusted experts."
After agonizing over features and a template, the team launched their first monthly newsletter in early August 2005. Like future monthly issues it linked to five new feature articles, including:
- Unbiased Consumer Reports-style shopping advisories such as the article headlined, 'Ink Cartridge Rip-Off'.
- How-to instructional articles on getting the most from devices such as camera phones.
- Q&A from "Digital Dan," a made-up character complete with his own company email address. (Note: It's OK to make up a character, such as Betty Crocker, who can interact with customers for decades even though individual employees come and go. However you can't send an invoice or legal notices from a fictional person.)
In addition, the template featured the site logo at the top, and "ads" for various products in the text, just as you'd expect in a newsletter from a real news site.
The team crossed their fingers and hit the send button. Results were pretty good for a newsletter to such an old list. "We were happy." But, day after day the number of non-opens began to eat at Quazi. By day 14 after the send, he proposed a test.
Why not re-send the newsletter -- the exact same newsletter with the exact same subject line -- to everyone with a non-bouncing address who had not opened and/or clicked? (Note: Anyone viewing the text-only version of the newsletter would not be counted as an open, so you need to watch their clicks.)
This test was far scarier than the first. If any recipients were annoyed, they might report Mouse2House as a spammer and ruin the email party. (See below for a list of MarketingSherpa recommendations if you decide to replicate this test.)
Would it work? Would non-openers seeing the same exact email in their inbox twice in a month open and respond to it?
RESULTS "It's incredible really, why would someone open an email the second time around who didn't open the first? It's quite amazing," says Quazi.
Here's the data for the first two month's tests with send I at the first of the month, and resends two weeks later. (Clicks are percents of sent minus bounces; % of sales is for that month's 100% campaign):
Open rate Click Sales Unsub rate
Aug I 29.2% 8.3% 69% 1.04%
Aug II 11.5% 3.4% 31% .48%
Sep I 32.3% 9.2% 72% 1%
Sep II 12% 4.6% 28% .62%
Best of all, "now we've done this for four months, equaling eight sends, and we've not had a single negative response from a customer." That's not to say there hasn't been feedback. Digital Dan now sees his share of reader mail, and when he made a spelling error recently several readers were quick to let him know.
The team also studied click patterns within issues to see if there were any patterns they could learn from. They discovered readers tended to click on the headlines they thought sounded most interesting -- regardless of where those headlines appeared in order on the page. And perceived-boring topics (such as information about storage devices) wouldn't be clicked on no matter where they were.
Also, invariably more than 10% of total issue clicks, for both send I and send II, are on the site logo at the top of the issue.
MarketingSherpa's kids-don't-try-this-at-home recommendations
As we stated in our introduction, this is a dangerous test although obviously a viable one. For US marketers, we advise the following precautions:
1. Don't resend names at email addresses that are likely to block HTML images and thus not record opens. This includes Gmail and some at-work addresses using Outlook or Lotus Notes.
2. Don't resend text-only version recipients.
3. Don't resend content that's more frequent than monthly. Frequency is the number one cause of spam complaints that permission mailers receive.
4. Put an unsubscribe link prominently at the top of every mailing. Don't make them scroll.
5. Monitor your reply address, opens, clicks, conversions, indications of emailer reputation (blacklists etc) and any kind of customer service feedback like a hawk. This is not a tactic for marketers using less sophisticated analytics systems.
6. Never mail names that are more than 90 days old that you haven't consistently mailed unless you are a well-known, trusted brand in your marketplace and you're darn sure people like and remember you. Even then, do it carefully. Perhaps even send a re-opt-in offer.
7. Take names off your active sending file if they have not opened and/or clicked in the past six months (or even sooner for weeklies and dailies).
Quazi, whose team watched their analytics on a daily basis, has now segmented all never-opened/never-clicked names into a separate test cell. He's strongly considering never mailing those names again if they don't respond to the next newsletter.
Useful links related to this article
Creative samples -- Mouse2House first three newsletters: http://www.marketingsherpa.com/m2h/study.html
Adestra UK -- the email marketing, Fax and SMS broadcast services firm that Mouse2House relies on : http://www.adestra.co.uk