Monday morning, stand-up. Someone asks how push did last week. You know your all-time subscriber count and your lifetime click total off the top of your head — but neither one answers the actual question. Was last week better or worse than the week before? Is your click-through rate climbing or sliding? Did the unsubscribe rate creep up while nobody was watching?
A lifetime total can’t tell you if this week was better or worse than last
That’s the gap a lifetime total can’t close, and it’s exactly what week by week push notification performance reporting is for. Instead of one number that only ever goes up, you need buckets — day, week, or month — that you can set side by side and actually compare. A single all-time total tells you where you’ve ended up. A push notification trend report tells you which direction you’re moving right now, and how fast.
If you’ve read PushEngage’s post on lifetime push notification analytics, you already know pushengage_get_analytics_summary answers “how are we doing overall”: all-time subscribers, all-time sends, all-time clicks, all-time goal value. That’s a different tool for a different question, and it’s worth being precise about the boundary — this post never touches that all-time total again. This post is about pushengage_get_analytics_timeseries — the tool that answers “how are we doing right now, and is it moving” — by breaking your account’s performance into buckets you set yourself and letting you ask for it in plain English instead of building a report by hand.
What pushengage_get_analytics_timeseries returns: week-by-week push notification performance, on demand
The tool returns exactly six fields, per bucket, over whatever date range you give it: subscribers, sends, views, clicks, CTR, and unsubscribes. You choose the bucket size — day, week, or month — based on what you’re trying to see. A day-by-day bucket surfaces a single bad send inside a busy week. A week-by-week bucket is the right grain for a Monday stand-up. A month-by-month bucket is what you’d hand to a director who wants the quarter, not the noise inside it.
There’s no separate “weekly report” feature to configure and no template to build first. The date range and the bucket size are just two parameters in whatever you ask, so the same tool produces a push notification trend report at whichever grain the conversation calls for — a week ago, this quarter, or a custom range around a specific campaign. That’s a meaningfully different shape than the dashboards most platforms ship, where “weekly” and “monthly” are fixed views you pick from a dropdown rather than a range you set yourself.
That flexibility matters more at PushEngage’s scale than it would on a smaller platform. The network moves 15.2 billion notifications in a 30-day window across 25,000+ business owners in 150+ countries — volume where a single lifetime figure tells you almost nothing about this week specifically. A trend view, not a total, is the only way to see whether your account’s slice of that volume is moving in the direction you want.
Getting started: connecting the PushEngage MCP server to your AI assistant
None of the examples below work until the PushEngage MCP server is connected to your AI assistant. Add npx -y @pushengage/mcp to Claude Desktop, Claude Code, or Cursor’s MCP config, restart the client, and the “pushengage” server appears in your tool list. The first time you use it, ask the assistant to log you into PushEngage — it opens a browser tab where you click Authorize, and your credentials never touch the assistant itself.
Once you’re logged in, ask it to show your PushEngage sites and tell it which one to use; that selection is remembered across restarts, so every question below runs against the right account by default. For the full walkthrough, including troubleshooting a server that won’t connect, see the full PushEngage MCP setup guide.
Asking for the trend: three ways to bucket the same question
Once you’re connected, the same underlying question — “how is push performing over time” — can be asked three different ways, depending on the grain you need.
Week-by-week, for a monthly review:
“Break down last month’s clicks week by week.”
The assistant returns four or five weekly buckets, each with sends, views, clicks, and CTR, so you can see immediately whether performance was flat, climbing, or front-loaded in one week and dead by the last.
Day-by-day, for spotting a single bad send:
“Show me daily sends and CTR for the last 14 days.”
This is the grain that catches a specific campaign that underperformed inside an otherwise healthy week — a day-level dip a weekly bucket would average away.
Month-by-month, for the quarterly view:
“How did October compare to September, month by month?”
This is the version you’d screenshot for a director: two buckets, side by side, with the delta between them instead of a single number that only ever grows.
All three are the same tool, pushengage_get_analytics_timeseries, with a different bucket size and date range — the bucket is just a parameter in the question you ask, not a different report to learn. That’s the actual work of a push notification analytics AI assistant: it’s the same underlying data every time, reshaped on request instead of pre-baked into whichever view a dashboard happened to ship with.
It’s also the version most retention teams actually need most often: to compare push notification performance month over month, without exporting anything first. Ask for last month and the month before in the same breath, and the assistant hands back both buckets side by side instead of two separate screenshots you’d have to line up yourself.
Reading the comparison: what a moving CTR line is actually telling you
A percentage-point change in CTR sounds small until you turn it into clicks. Say your account sent 400,000 notifications in September at a 4.1% CTR — 16,400 clicks. In October, sends held steady at 400,000, but CTR slipped to 3.6% — 14,400 clicks. That’s 2,000 fewer clicks against the same send volume, and every one of those clicks was a chance at a recovered cart, a win-back, or a repeat purchase that didn’t happen. A manager reporting that upward should say “2,000 fewer recovered-revenue opportunities month over month,” not “CTR dropped half a point” — the second version undersells exactly what moved.
This is the habit worth building into every week by week push notification performance reporting cycle: don’t stop at the percentage. Multiply the delta by your send volume, and you have a number a director can act on instead of a chart they’ll skim past.
The same math works in the other direction, at a smaller grain. If you compare push notification performance month over month and see three weeks flat followed by one week down half a point, that’s not “October was soft” — it’s one specific week, one specific send, or one specific segment worth pulling individually before you write off the whole month. A week over week push notification CTR line is what tells you which week to go pull.
When the trend dips: what to check before you blame the copy
A falling week over week push notification CTR has three common causes, and none of them is “the notification copy got worse.” Before you rewrite a headline, pull the unsubscribe and send-volume buckets alongside the click bucket for the same weeks — the combination usually points straight at one of the three.
Subscriber list decay. A broadcast list that never gets re-segmented ages — the subscribers who opted in eighteen months ago and stopped engaging are still counted in your send volume, dragging the average down. PushEngage’s audience groups let you build behavioral and RFM-style segments (recent, frequent, high-value, or lapsed) directly from the tools available in the MCP server, so a declining CTR is a segmentation problem you can actually fix rather than a mystery. If your dip traces back to this, segmentation’s effect on click-through rate is worth a closer read, and ways to increase your click-through rate covers the tactical fixes once segmentation is in place.
A rising unsubscribe rate. The unsubscribe field in your timeseries data isn’t just a housekeeping number — trended week over week, a rising unsubscribe rate is an early warning that usually shows up before the CTR drop does. Catch it here first, and reduce your unsubscribe rate walks through the fixes.
Deliverability throttling. Chrome’s push rate limits and platform-level policy changes can quietly cap how many of your sends actually reach a device, which looks identical to a content problem in a CTR chart but has nothing to do with your message. If sends and views both dip together while CTR stays flat, that’s the deliverability signature — not a copy problem. You’d catch that pattern only by looking at sends and views as their own bucketed trend, not by staring at CTR alone.
Running all three checks takes the same amount of effort as asking one more question. Instead of guessing, ask directly: “Show me sends, views, clicks, and unsubscribes for the last eight weeks.” One answer, four fields, and the combination usually rules out two of the three causes before you’ve opened a single dashboard.
If you report on more than one account, this is the part that saves the most time
If you’re an agency managing push for several client sites, the alternative to this workflow is building a Looker Studio report for each account, every week, by hand. With the MCP server connected, the same question gets asked once per site: select client A with pushengage_select_site, ask “compare this month to last month, week by week,” get the answer, switch to client B, ask the identical question again. There’s no multi-account rollup built into the tool — you’re still asking site by site — but three client sites and three quick questions is a materially different Monday than three dashboard exports and three chart screenshots.
That speed compounds into a real number when you can attach revenue to it. Measuring push notification ROI covers how to carry a click delta all the way through to attributed revenue per account, once you have the week-by-week or month-by-month numbers in hand.
This is where a push notification analytics AI assistant earns its place in an agency’s Monday routine specifically — not because it rolls multiple clients into one view, but because the marginal cost of the fourth question is close to zero once the first one is answered. A retention manager with one account and an agency with fifteen are running the identical workflow; the agency just runs it more times in the same sitting.
The report only matters if something changes because of it
Week by week push notification performance reporting is only worth building if it changes what happens next. A CTR that slipped 0.5 points should end in a rebuilt segment, not a bullet point in a deck. An unsubscribe rate that ticked up two weeks running should end in a send-frequency review before it becomes a deliverability problem. A month-over-month comparison that shows 2,000 fewer clicks against flat volume should end in a dollar figure in your next report, not a percentage buried in a footnote.
That’s the actual test for this kind of reporting: not whether the trend line exists, but whether it changed a decision. Once the numbers are a question away instead of a dashboard export away, there’s no excuse left for skipping that step. See PushEngage plans to connect the MCP server to your own account and start asking.
