How are email rates calculated?
One person opening five times is one open, and every rate on the Email tab is out of emails delivered — except bounces, which are out of emails sent. This article gives each rate, its denominator, and the floors a send or a subject-line style has to clear before MarqIQ says anything about it. If you want the wider rule about naming the population behind any figure, read What number is a money figure resting on?
People, not events
Every rate counts people. One person who opens an email five times on the train counts once, and one who clicks three links counts once. The counts MarqIQ reads from your tool are the unique-recipient counts, so a rate here can never climb because one reader was busy.
One figure is deliberately an event count, and it is labelled as one: where a tool reports click events as well as clicking people, MarqIQ prints both so you can see how many raw clicks a lead score would have added.
Each rate and its denominator
| Rate | Top of the fraction | Bottom of the fraction |
|---|---|---|
| Delivery | Emails delivered | Emails sent |
| Open | People who opened | Emails delivered |
| Click | People who clicked | Emails delivered |
| Click-to-open | People who clicked | People who opened |
| Unsubscribe | People who unsubscribed | Emails delivered |
| Bounce | Emails that bounced | Emails sent |
| Spam complaints | Complaints | Emails delivered |
Bounces are out of sent because a bounced email was never delivered; putting it over delivered would divide by a number the bounces are already missing from.
The five tiles at the top of the tab print the denominator under each figure in the same words: “of people who received it”, or “of what was sent”.
Read the open rate as a rough guide
Mail apps now load every message on the reader's behalf, so an open no longer proves a person looked. MarqIQ judges sends on clicks wherever it can, shows opens beside them, and says so on the screen. Opens are still used for one thing: comparing subject lines, because a subject line is all a reader sees before opening, and inflated opens affect both sides of that comparison alike.
What the totals cover, and what is held out
The totals and every rate above are the period's campaigns — emails sent once to a list. Automated and triggered emails are held out of them, listed separately, and kept out of the day, the month and the trend analysis as well.
The reason is that an always-on email has no single send date. Depending on the tool, its numbers are either for this period or for everything since it was switched on, and adding a lifetime to a quarter makes a figure that is neither.
| Tool | What an always-on email's numbers cover |
|---|---|
| HubSpot, Mailchimp, Salesforce Marketing Cloud | Since the email was switched on. The note on the card says so |
| Marketo Engage | This period only, counted from the activity stream. The note on the card says so |
Before a send gets a verdict
A send is called above or below your average only when both of these hold:
- It reached at least 100 people.
- Its result is beyond what chance alone would produce — the cautious end of its own 95% range still clears your average, in whichever direction.
It is called Far above your average or Far below your average when even that cautious end is half again better, or half as good. Anything under 100 delivered reads Too few recipients to judge, and is in no best or worst list.
The average a send is compared with is always your own, across the period on the tab — never an industry figure. A 2% click rate is a triumph for one list and a warning for another.
Note
Sends only get verdicts once the period holds at least 30 clicks in all. Below that MarqIQ falls back to opens and says to treat it as a rough guide; below 30 opens too, no send is judged and the tab says so.
Before a subject-line style is called out
A style — a question, a number, an emoji, a subject over 60 characters — is only called better or worse when every one of these is true:
- At least three sends used it and at least three did not.
- Each side reached at least 500 people in total.
- The typical send on one side beats the typical send on the other by 15% or more. Typical means the median send, so one enormous mailing to a warm list cannot drag its whole side.
- The two pooled rates differ beyond chance, each interval excluding the other's rate.
- Two in three of the sends using the style agree with the verdict.
Even then the wording calls it the best candidate for a split test rather than proof, because different sends went to different audiences.
Styles travel together: if every long subject line is also a question, one effect would otherwise be reported twice under two names. MarqIQ keeps the clearest style, tests every other again among sends that are alike on that one, and withdraws a verdict that does not hold there. The line on the withdrawn style says which heading it is reported under instead.
Take care
One style is never recommended whatever it did to opens: a subject line dressed up as a reply or a forward when it is neither. It misleads the reader, mailbox providers read it as a spam signal, and anti-spam law in most markets forbids a deceptive subject line. MarqIQ counts how many of your sends use it and tells you to stop.
Clicks that were never a person
Security software at many companies opens every link in an email before it reaches anybody. Where your tool flags that activity, MarqIQ counts it apart and leaves it out of every click rate on the tab. Where your tool does not flag it, MarqIQ does not guess: the card does not appear.
- HubSpot filters that activity out of the rates it reports without saying how many. MarqIQ says so on the card.
- A warning is raised when flagged clicks are 10% or more of the addresses that registered a click and there are at least 20 of them, or when flagged click events are a quarter of all click events and there are at least 50.
- A flagged click from an address where a person also clicked changes no rate here, because the person is counted once either way. It still inflates anything that counts raw clicks, such as a lead score adding points per click.
Days, months and the trend
A day of the week is called out only with at least three sends on it reaching at least 500 people, and a gap of 15% or more from your average. Days are in your workspace time zone; where no time zone is set they are in UTC, and the note under the chart says which.
The trend splits the period in half at the midpoint between your first and last campaign, needs at least three campaigns on each side, and calls a change only when the two rates differ beyond chance and the later rate is a quarter higher or a fifth lower. Everything else reads “held steady”, with both figures printed.
The spam threshold is not ours
A warning appears once complaints reach 0.1% of delivered email, because Gmail and Yahoo ask bulk senders to stay under that, and states that they start blocking mail at 0.3%.
What this tab does not measure
- Whether the people who clicked went on to buy. That needs each click tied to a person, and the send totals do not carry it.
- The wording of your emails. MarqIQ keeps facts about each email — its length, where its links go, how many images have no description — and never the copy.
- Whether links carry tracking tags. The sending tool adds its tags as the email goes out, so the body cannot answer it.
If it did not work
Your open rate here is lower than the one in your sending tool
Why: Almost always a denominator difference: a tool showing opens over sent reads higher than opens over delivered, and some tools count total opens rather than opening people.
Fix: Check which the tool is reporting. MarqIQ is unique openers over delivered, every time, and prints both counts.
Your totals here are lower than the tool's
Why: The totals cover campaigns only. Automated and triggered emails are listed separately and are in no total above.
Fix: Add the Always-on emails card to what you are comparing, and check whether that tool reports those over their lifetime or over the period.
No send has a verdict
Why: The period holds fewer than 30 clicks and fewer than 30 opens, or no send reached 100 people.
Fix: Widen the period, or wait for more sends. The tab says which of the two floors was not met.
A subject-line style you believe in is not called out
Why: One of the five floors was not met, or the style was withdrawn because it is the same sends as a clearer style.
Fix: Read the line under that style. It says either that the gap is too small or too inconsistent, or which heading the effect is reported under instead.
The Email tab is greyed out and says your tool does not send marketing email
Why: MarqIQ reads per-send results from four tools only, so on every other tool the tab is greyed out and opens on to “Not something this tool can do.” Its one sentence is the real reason on a CRM. On Pardot (MCAE), Adobe Campaign and Adobe Journey Optimizer it is the tab’s standard wording rather than the real reason, because those three do send marketing email and MarqIQ does not read their send results yet.
Fix: Connect HubSpot, Marketo Engage, Mailchimp or Salesforce Marketing Cloud. Where your sending tool is one of the three not read yet, nothing on your side will make the analysis run — the tab stays greyed out whatever you change in the tool.
Next steps
Related articles
Was this article helpful?