Promotion Analytics: How to Read Delivery, Survival Rate, and CPM Before You Scale

Most operators track views and stop there. That is the wrong stopping point. A view count tells you what arrived; it says nothing about what survived, what it cost per thousand impressions, or whether the delivery curve matches what you actually ordered. Promotion analytics only becomes useful when you are reading at least three layers simultaneously: delivered volume, survival rate, and CPM.

These three metrics form a diagnostic triangle. High delivered volume with a collapsing survival rate points to a source quality problem. A healthy survival rate with an inflated CPM points to a pricing or packaging problem. Low delivered volume against a healthy CPM usually means a pacing problem. Knowing which corner of the triangle is broken tells you exactly what to fix before you move budget to the next order.

Delivery rate is the percentage of your ordered volume that has actually reached the target asset within the agreed window. If you ordered a 50,000-view TikTok package over 72 hours and the dashboard shows 31,000 delivered at the 48-hour mark, your delivery rate is 62 percent at the two-thirds time mark — roughly on pace. If it shows 12,000, you have a fulfillment gap worth escalating before the window closes.

The shape of delivery matters as much as the raw percentage. A front-loaded curve — say, 80 percent of volume arriving in the first 24 hours of a 72-hour order — can trigger platform suppression signals on some content types. A flat, even curve is usually safer for organic-looking growth. The delivery chart in your promotion dashboard should let you see hourly or daily cadence, not just a final tally. If the only number visible is a total, you are flying blind on pacing.

When pacing goes wrong, document the deviation with timestamps. That record becomes the basis for a make-good conversation or a delivery adjustment through the scaler before the order concludes.

Survival rate measures what percentage of delivered views, followers, or plays remain on the asset after a defined retention window — typically 30 days post-delivery. A campaign that delivers 100,000 views but retains only 40,000 after 30 days has a 40 percent survival rate. That number belongs in every client report, not just the initial delivery screenshot.

A survival rate below 70 percent on a views-based campaign usually indicates source quality issues: traffic that the platform's own systems downrank or remove over time. Rates above 85 percent on a well-paced order suggest the traffic is behaving in ways the platform treats as organic-adjacent. Neither threshold is a guarantee of anything, but the range gives you a defensible benchmark for comparing vendors and packages over time.

The practical reason to track survival rate is cost accuracy. If you paid for 100,000 views at a $4 CPM and 50,000 dropped, your effective CPM just doubled. Report the post-retention CPM to clients, not just the delivery CPM. The difference is where your credibility lives.

Standard CPM (cost per thousand) in paid promotion is calculated against delivered impressions. That is fine for display advertising where an impression is an impression. In organic-growth promotion, where the asset persists and continues accumulating value, the more honest metric is retained CPM: total spend divided by retained volume at the 30-day mark, multiplied by 1,000.

A 50,000-view YouTube package priced at $150 shows a delivered CPM of $3.00. If 42,000 views survive at day 30, the retained CPM is $3.57. That is a small difference — and that is what a high-quality package looks like. On a lower-quality package where only 25,000 views survive, the retained CPM jumps to $6.00 on the same $150 spend. Reporting both numbers side by side in your campaign reporting gives clients an honest view and positions you as an operator who measures what actually persists.

Build the retained CPM calculation into your standard reporting template. It takes one extra column in a spreadsheet and it is the single fastest way to differentiate your reporting from agencies that only screenshot the delivery notification.

A repeatable promotion analytics workflow has four components: a delivery snapshot at the midpoint of the order window, a final delivery confirmation at close, a 30-day retention check, and a cost reconciliation that updates CPM based on retained volume. None of these steps requires custom tooling if your promotion dashboard exports timestamped data at the order level.

The midpoint snapshot is the most skipped step, and it is the most operationally valuable. Catching a delivery shortfall at hour 36 of a 72-hour order gives you time to escalate or adjust volume through the scaler. Catching it at hour 72 means you are writing a make-good request after the fact. Build the midpoint check into your calendar as a hard event, not an optional audit.

For clients who receive monthly reporting decks, structure the promotion section as: ordered volume, delivered volume, retained volume, delivered CPM, retained CPM, and a single-sentence assessment of pacing quality. Six fields. That is a complete picture. Anything more is noise; anything less is an incomplete record.

If your delivery rate is under 80 percent at the two-thirds time mark, open a delivery adjustment or escalation through your dashboard before the order window closes. Do not wait. Most fulfillment issues that are caught mid-window can be corrected; most that are caught post-window become credit disputes.

If your survival rate drops below 60 percent at the 30-day check, the root cause is almost always source quality, not pacing. The corrective action is to test a different package tier or a different content category on the next order — not to increase volume on the same configuration. More volume through a low-quality source accelerates the problem.

If your retained CPM is more than 40 percent above your delivered CPM, you are overpaying for the effective audience you are actually building. That gap is your negotiating number for the next order cycle and your benchmark for evaluating whether a package tier is worth repeating.