Free resource · use it on this page
Push Strategy Planner
Define goals, audiences, channels, and KPIs on one page, so the whole team is aligned before you send anything.
How to use this planner
Most bad sends are not caused by bad copy. They are caused by a team that never agreed on what the message was for, who exactly should get it, or what number would prove it worked. This planner forces those agreements into writing before anything reaches a lock screen.
Work through it in one sitting, ideally in a 30-minute session with everyone who touches the send: whoever writes the message, whoever owns the audience data, and whoever answers for the result. Fill in every field, complete the channel table, then walk the sign-off checklist together. Print the page or keep it open in a shared tab; the finished planner is the brief.
One rule holds the whole thing together: one campaign, one goal, one primary KPI. If you find yourself writing two goals, you are planning two campaigns. Split them and fill this out twice.
Part 1: Goal
Write the goal before you touch copy, audience, or channel. Every later decision gets tested against this section.
Campaign name and single owner
One accountable person, not a team. If two names go here, decide now who breaks ties.
Goal in one sentence
Format: bring [who] to do [what] by [when]. Example: bring customers who have not visited in 30 days back in-store within 14 days of the send. If it does not fit one sentence, it is two campaigns.
The one action you want the recipient to take
Redeem, book, visit, reply, complete a profile. A message with two asks usually converts on neither.
Why now
The trigger that justifies interrupting someone: a launch, an expiring offer, a seasonal moment, a lapse threshold crossed. No reason to send now is a reason to wait.
What this campaign will not do
Scope guardrail. Example: no discount deeper than 15 percent, no send to customers who purchased this week, no follow-up blast if results disappoint.
Send window and hard deadline
A date range, not just a date. Include the last acceptable send day; offers sent after their moment train people to ignore you.
Part 2: Audience
An audience is a written filter, not a vibe. Anyone on the team should be able to re-run your definition and get the same list.
Segment definition, written as a filter
Use reproducible conditions: purchased in the last 90 days AND no visit in 30 days AND has an active pass or subscription. Avoid labels like loyal customers that no query can verify.
Audience count, actually pulled
Run the filter and write the real number. If it is off from what you expected by more than a rough factor of two, the filter is wrong; stop and fix it before going further.
Where consent came from
The specific opt-in moment: app permission prompt, pass install, signup checkbox, SMS keyword. If you cannot name it, do not send to that slice.
Exclusions
Standard set: staff and test accounts, customers who purchased in the last few days, anyone who received a message inside your frequency cap window, anyone mid-complaint with support.
Last time this audience heard from you
Date and message. If it was recent and unrelated, either wait or shrink the audience to people who were not in that send.
Plan for unreachable members
Some of the segment will not have the chosen channel. Decide now: reach them on a fallback channel, or explicitly accept the gap and note the reduced count.
Part 3: Channel selection
Channels differ in permission friction, marginal cost, and where the message physically lands. Mark the last column for each row: Primary, Support, or No.
| Channel | Permission required | Marginal cost per send | Where it lands | Role in this campaign |
|---|---|---|---|---|
| App push notification | Installed app plus an OS-level permission prompt the user must accept | None at typical volumes | Lock screen and notification tray; easily swiped away | Primary / Support / No |
| Web push | Browser permission prompt; on iOS only for sites added to the Home Screen | None at typical volumes | Desktop and Android notifications; limited reach on iPhone | Primary / Support / No |
| Wallet pass update | Customer adds a pass to Apple or Google Wallet once; no app install needed. Wallet-native platforms such as PushNotice deliver here | None per message | Lock screen via the pass itself; the offer persists in the Wallet instead of vanishing | Primary / Support / No |
| SMS | Explicit consent with strict regulatory requirements and mandatory opt-out handling | Paid per message; cost scales with audience size | Native messages app; hard to ignore, and hard to forgive if misused | Primary / Support / No |
| Address capture with consent; no OS prompt involved | Near zero | Inbox, competing with everything else in it; best for detail and reference | Primary / Support / No | |
| In-app message | None beyond having the app; shows only during an active session | None | Inside the product while the user is already engaged; cannot reach the lapsed | Primary / Support / No |
Part 4: KPIs and thresholds
Decide what success looks like before the send, in writing. A threshold chosen after the results arrive is a rationalization, not a target.
Primary KPI, exactly one
Pick the number closest to the business result: redemptions, bookings, visits, completed signups. Opens and taps are diagnostics, not the verdict.
Current baseline
What that number does in a normal week with no campaign. If nobody knows, write baseline unknown and declare this send a baseline-setting exercise, honestly, before it goes out.
Success threshold
The specific value at which you will call this a win, for example 40 redemptions within the window. Write it here and do not move it afterward.
Measurement window
How long after the send you count results before closing the book, commonly 7 to 14 days. Results trickling in later do not retroactively change the verdict.
Stop condition
The signal that pauses or kills the campaign mid-flight: an opt-out or uninstall spike, redemption chaos at the counter, a broken link report. Name the signal and who is watching for it.
Where the number lives
The exact dashboard, report, or query the team will look at, and who reads it. If two people would check two different reports, reconcile them now.
Part 5: Pre-send alignment sign-off
Run this together as the last step. Every unchecked box is a named reason to hold the send. All boxes checked means the team is aligned, not just informed.
The goal sentence is written and everyone involved has read the same sentence
Not a summary from a meeting; the actual sentence in Part 1, seen by every signer.
There is exactly one primary KPI and one success threshold
If someone can still say we will see how it does, Part 4 is not finished.
The audience filter was actually run and the count sanity-checked
A real number from a real query, compared against what you expected before running it.
Exclusions are applied, including staff and test accounts
Internal accounts receiving a customer offer is the most common embarrassing miss.
No recipient falls inside your frequency cap window
Check the last-send date per audience member, not per campaign. Overlapping segments hide double-sends.
The message has a single call to action
One link, one ask. Secondary links siphon taps from the action that defines success.
The message was viewed on a real device lock screen
Truncation, missing images, and awkward line breaks only show up on hardware, not in a compose-screen preview.
The link or landing path was tested from the message itself
Tap through from the delivered test message, on a phone, logged out. Not by pasting the URL into a desktop browser.
The offer is honorable end to end
Redemption works at the counter or in checkout, expiry dates match the copy, and frontline staff know the offer exists before customers do.
Send timing respects the recipient, not the sender
Recipient time zones, local business hours, and no late-night lock-screen buzz unless the content genuinely warrants it.
Someone can pause the campaign, and everyone knows who
Name the person, confirm they know the exact steps in the sending tool, and that they are reachable during the window.
The baseline is recorded before the send goes out
Screenshot or note the pre-send number now. Reconstructing it afterward invites wishful memory.
A post-send review is on the calendar
Book 20 minutes at the end of the measurement window with the same people who signed this page. Compare the result against the threshold in Part 4, in writing.
From the guide: Best Push Notification Tools (2026)
This resource accompanies the full article — worth reading before you commit to a tool.