vMTA statistics.
Comparing
sending addresses.

The tiles sum up the week's vMTA statistics: the best and worst sending point by bounce rate, the spread between them and the average across all points. It's a quick answer to how evenly your sending addresses perform. In the example 23 vMTAs on 13 servers sent 122,161 messages; the best is inbox.lumaclub.net-vmta with 2.1% bounced, the worst is offers.lumaclub.net-vmta with 4.7%. Points with under 1% of sending are left out of the comparison, so that one random bounce doesn't make them the worst. The color follows the Hub's common thresholds: under 4% — green, from 4% — yellow, from 6% — red. If the spread is large, the next step is the summary below, which shows how each point differs from the average.

Enlarge

• Best — inbox.lumaclub.net-vmta — 2.1%.
• Worst — offers.lumaclub.net-vmta — 4.7%.
• Spread — ×2.2.
• For comparison — 2.9% across all points.
• Included — points from 1% of sending — here all 23.

Color: under 4% — green, from 4% — yellow, from 6% — red.

The summary shows every sending point: the vMTA name, its address, recipients, delivery, bounces and how its bounce rate differs from the average in percentage points. You use it to compare sending IP addresses with each other and find a group that lags behind. In the example the two bank vMTAs send over 60% of all mail with 2.2% bounced, while fifteen points have a rate of 4.4–4.7% — 1.5–1.8 pp worse than average. Why an address is worse — warm-up, PTR, blacklists — you check on the addresses page of “Analytics”. Eight points hold at 2.1–2.2%, so the difference within one list is obvious right away. Clicking a row expands the point's days and worst domains.

Enlarge

• Bank points — id. and secure.corvalis-bank.org — 37 thousand each, 2.2%.
• Fifteen points — 4.4–4.7%, +1.5…+1.8 pp.
• Eight points — 2.1–2.2%.
• Under the table — what the MTA doesn't know: warm-up, PTR, blacklists.

“Who's worse” — here, “why” — in “Analytics”.

Clicking a summary row expands the details of one vMTA: on the left — its days, on the right — up to five domains it delivers to worst. The screen helps you see whether the point lags all the time or on certain days, and with which mailbox providers. In the example news.lumacart-shop.com-vmta is expanded: 223–720 messages a day with 4.3–5.1% bounced, the worst domains are veltrano-group.com (9.4%) and yahoo.com (6.2%). gmail.com has a rate of only 2.7%, but the largest number of bounces — 38. A hint under the bars reminds you: a full bar is 12.5% or more, and a high percentage on five messages means nothing. Oct 15 and Oct 22 are paler — these are the edges of the accounting period.

Enlarge

• Days — 223–720 messages, 4.3–5.1%.
• Edges — Oct 15 and Oct 22 are paler, with a caption.
• In red — veltrano-group.com — 9.4%, yahoo.com — 6.2%.
• In yellow — hotmail.com — 4.6%, outlook.com — 4.2%.
• gmail.com — 2.7%, but the most bounces — 38.

A full bar is 12.5% or more.

The hourly volume map shows how many messages each vMTA sent in each hour of the chosen day. The scale is logarithmic, so a point with twenty messages an hour is visible next to a bank point with hundreds. The map opens on the busiest day by itself — Oct 16 in the example — and shows that all 23 points worked every hour. An empty cell would mean the point was silent that hour; you pick another day in the list on the right. Hours are in your profile's time zone. The edge days are captioned “Oct 15 — from 13:03”, and hours outside the period are hatched: there's no data for them, it's not downtime. Look at the volume map together with the bounce map below.

Enlarge

• Rows and columns — points and hours in your profile's time zone.
• Scale — logarithmic — from 20 to 536 messages an hour.
• Empty cell — would mean the point was silent that hour.
• Edge days — “Oct 15 — from 13:03”, the hours before that are hatched.

The day is picked in the list on the right.

The second map is each vMTA's bounce rate by hour for the same day, compared with the average across all points, 2.9%: green — better, red — worse. It shows in which hours a point's bounces went up. Only hours with at least fifty messages get a rating — in the example 55 cells out of 552, and 22 and 20 hours out of 24 for the bank vMTAs. The other cells are gray: on twenty messages one bounce gives either 0 or 5%, and the color would only confuse you. Compare the red hours with the volume map above: that shows whether a point's bounces grow along with the load. The 50-message threshold is given in the legend, and the message count is in each cell's tooltip.

Enlarge

• Green and red — better and worse than average.
• Bank points — rated for 22 and 20 hours out of 24.
• In total — 55 cells out of 552.
• Gray — fewer than 50 messages an hour — the message count is in the tooltip.

The threshold is given in the legend.

The “point × domain” table shows each vMTA's bounce rate on each of the eight largest domains; the bottom row is the average across all points for that domain. This is the main screen for deciding where the problem is — in the sending address or in the message. In the example fifteen points have every cell worse than average: if the list and messages are shared, the problem is the addresses. veltrano-group.com has a problem of its own — 6.0% versus 1.7% for gmail.com. The Hub doesn't know what goes through the points, so it states its conclusion with a caveat. A cell with fewer than fifty messages for the week would be gray; there are none here. Why an address is worse — warm-up, reverse DNS, blacklists — you check on the addresses page of “Analytics”.

Enlarge

• Bottom row — the average across all points for the domain.
• Bad row — 15 points at about 4.6% — every cell worse than average.
• veltrano-group.com — 6.0% on average versus 1.7% for gmail.com; up to 11.0% for the same points.

The conclusion comes with a caveat: the Hub doesn't know what goes through the points.