Contoso Ltd is a made-up online shop for garden tools in the EU. All figures are invented for teaching.
Running the case (60 minutes)
60 min- 10 min
Everyone reads the situation and the evidence. Unknown terms are explained with the glossary below.
- 15 min
Pairs do tasks 1 and 2: the calculations and the bought list.
- 15 min
Pairs answer the quiz and put the fixes in order (tasks 3 and 4).
- 10 min
Tasks 5 and 6: the legal question and a short recommendation.
- 10 min
Compare answers as a class. Each pair reads out its recommendation in one minute.
Words you need
Deliverability: how reliably emails reach the inbox instead of the spam folder or nowhere at all.
Hard bounce: the receiving server says the address does not exist. It will never work.
Soft bounce: a temporary problem, such as a full mailbox.
Spam complaint: a recipient clicks "Report spam". Mailbox providers count these.
Mailbox provider: the company that runs people's inboxes and decides what lands where.
Authentication: SPF, DKIM and DMARC records that let a receiver check the sender. See the guide to SPF, DKIM and DMARC.
The situation
For two years Contoso sent a monthly newsletter from its shop software to about 14,000 customers who had ticked a box to receive it. Opens were steady and complaints were rare.
In March the team moved to a new email service. To "make the most of it", they imported every address the company had ever collected: the 14,000 subscribers, 14,000 old addresses from orders and enquiries going back eight years, and a list of 12,000 addresses bought at a trade fair. Then they sent a spring sale to all 40,000 addresses at once.
Within a week, opens fell from 35 to 6 percent and customers said the newsletter was in their spam folder. Even order confirmations, sent from the same domain, started to arrive late or in spam.
The evidence
What the team found | Value |
|---|---|
Addresses in the March send | 40,000 |
Hard bounces in the March send | 9 percent |
Spam complaints | 0.6 percent of delivered messages |
SPF record | Still lists only the old shop software |
DKIM | Not set up with the new service |
DMARC | No record |
Addresses from the bought list | 12,000 |
Bounce messages | Mostly "550 5.1.1 user unknown" |
A word on open rates: they are only a rough signal. Some mail apps load images automatically for privacy reasons, which counts as an "open" that never happened, and others block images, which hides real opens. A sudden fall from 35 to 6 percent is still a clear warning.
Tasks
Calculate. How many messages hard bounced? How many were delivered? About how many people complained? Google asks bulk senders to keep the spam rate below 0.3 percent and recommends staying below 0.1 percent (Google's sender guidelines). How many complaints would each limit have allowed?
The list. What share of the March list came from the bought list? Give two reasons why these addresses cause both bounces and complaints.
Diagnose. Answer the quiz below.
Fix. Put the fixes in order below. Then explain in one sentence why the team should not simply send again next week.
The law. Contoso sells to private customers in the EU. Was it allowed to send its newsletter to the bought list?
Recommend. Write five sentences to the managing director: what went wrong, what to do first and which three figures to watch over the next two months.
Diagnose it
1.Which three problems hurt delivery most? Pick three.
More than one answer is right.
2.Why did order confirmations suffer too?
3.What does a hard bounce with "user unknown" tell the team?
4.Why do the 14,000 old addresses cause trouble even though they once belonged to customers?
Fix it in this order
- 1Send again to recent, engaged subscribers first
- 2Publish a DMARC record with p=none to get reports
- 3Grow the volume step by step over several weeks
- 4Stop sending to the bought list and remove hard bounces
- 5Add the new service to SPF and switch on DKIM signing
Model answers
1. Calculate. Hard bounces: 9 percent of 40,000 = 3,600. Delivered: 40,000 − 3,600 = 36,400. Complaints: 0.6 percent of 36,400 ≈ 218. The 0.3 percent limit would have allowed about 109 complaints, the recommended 0.1 percent only about 36. Contoso was at twice the hard limit.
2. The list. 12,000 of 40,000 is 30 percent. Bought addresses are often old or collected without care, so many no longer exist (bounces). The people behind the rest never asked Contoso for anything, so they report the email as spam (complaints).
3. Diagnose. See the quiz explanations.
4. Fix. Sending again before the list is cleaned and the records are fixed would add more bounces and complaints and push the reputation down further. Starting with the people most likely to open and least likely to complain rebuilds trust.
5. The law. Almost certainly not. In the EU, marketing emails to private people generally need their prior consent. This rule comes from the ePrivacy Directive (Article 13) as put into national law, for example § 7 UWG in Germany; the UK has a similar rule in PECR (ICO guidance). Consent collected by a list seller at a trade fair rarely covers Contoso. The 14,000 old addresses are a problem too. The exception for existing customers only covers addresses that came from a sale, advertising for similar products, and people who were offered an opt-out when the address was collected and in every email since. Addresses from enquiries without a purchase do not qualify. Removing the bought list is a legal step, not only a technical one. This is an overview, not legal advice.
6. Recommend. A model: "Our March send went to 40,000 addresses, a third of them bought and many outdated, from a service that our domain did not authenticate. Mailbox providers saw 9 percent bounces and twice the accepted complaint rate, so they now filter our emails, including order confirmations. We will delete the bought list and all bounced addresses today and fix SPF, DKIM and DMARC this week. We will then write only to the 14,000 subscribers, starting with those who opened recently, and grow slowly. We will watch the hard bounce rate, the complaint rate and the share of emails reaching the inbox each week."
Notes for teachers
Mailbox providers watch bounce and complaint rates closely. Google's sender guidelines ask anyone sending more than 5,000 messages a day to Gmail addresses to keep the spam rate reported in its Postmaster Tools below 0.3 percent, recommend below 0.1 percent, and require SPF, DKIM, a DMARC record and one-click unsubscribe. Yahoo and Microsoft (for Outlook.com) publish similar rules. Bought lists usually breach consent rules in the EU as well, which makes removing them a legal step, not only a technical one.
Typical mistakes. Learners calculate complaints from all 40,000 addresses instead of the delivered ones; accept it if they say so, as the difference is small. Many put "send again" before fixing the records. Ask: what would the mailbox providers see?
Follow-up questions. How could Contoso have tested the new service safely? Who in the company should have stopped the import of the bought list? What should the sign-up form say so that consent is clear?
Differentiation. Easier: give pairs the formula "rate = count ÷ total × 100" and do task 1 together. Harder: learners draft the DMARC record and explain how they would move from p=none to p=quarantine over the following months.
