How can I tell whether a social post caused a wishlist spike?

Check whether the post has recorded wishlist attribution, whether the timing fits, and what else happened during the same period. Together, those checks can make a post a plausible explanation. They usually cannot prove how many wishlists it caused. Finish the review with an evidence statement and a next test, rather than assigning the whole spike to the most visible activity.

Separate three questions

A studio deciding whether to repeat a post needs to distinguish:

Question	Evidence you can inspect
Did people respond to the link?	Its recorded visits and attributed outcomes
Did wishlist growth change around the post?	Daily additions and the event timeline
Would those wishlists have happened without the post?	A credible comparison that accounts for other causes

The first two inform the third, but they do not settle it. A player might click your post and wishlist a game they already intended to follow. Another might see the post and later find the game through a separate journey.

Immutable's attribution guide similarly distinguishes recorded outcomes, campaign-period growth and incremental impact. Its treatment of overlapping activity is a useful starting point; a count attributed to a link is not necessarily a lower bound on the extra wishlists the post caused.

Make the timing comparison usable

Record when the post went live, its source URL and tagged link, and the reporting timezone. Check whether the spike began before or after publication. If the wishlist data is daily, avoid claiming an hour-by-hour response that the report cannot show.

Steam's UTM conversion window is 72 hours, and conversion totals finalize four days after a visit. That reporting window is not a prediction of how long a video's influence lasts. Valve: UTM Analytics.

Choose a response window before assessing a new experiment where possible. If you expand it afterward because the graph looks promising, record that change in the review.

A spike with more than one explanation

Consider this fictional example:

Observation	Result
Recent comparison period	Around 20 additions per day
New video	Published Monday
Other activity	Demo released Monday; paid ads still running
Monday–Wednesday	150 total additions
Video's tagged link	18 attributed wishlists

Using 20 additions a day as a simple baseline gives 60 expected additions over three days. The difference is 90.

That is 90 additions above the chosen baseline, not 90 proven video-generated wishlists. The demo, ads and other changes could contribute. The 18 attributed wishlists provide evidence of a recorded response to the link, but do not allocate the remaining growth.

Check how much the estimate depends on the baseline. At 15 per day, the difference becomes 105; at 25, it becomes 75. This range is a sensitivity check on assumptions, not a statistical confidence interval.

Write a conclusion someone can act on

For this example, a defensible review note would be:

The video's link received 18 attributed wishlists, and additions increased during the response window. The demo release and ongoing ads prevent us from isolating the video's contribution. We will repeat the creative format with a distinct link and record all overlapping activity.

A follow-up might schedule the next post away from another major beat when practical, keep the audience and link setup comparable, and define the review date in advance. Repetition can improve the evidence, but an uncontrolled before/after comparison still leaves alternative explanations.

When a spending decision requires a causal estimate, consider whether a suitable holdout or controlled experiment is feasible. A quieter calendar is helpful context; it is not a substitute for a control group.

How Coal helps

Coal brings recorded UTM outcomes, wishlist history and marketing events into the same investigation. Its user-assigned impact estimates preserve an interpretation of a selected period; they do not independently prove cause and effect.

Use the timeline to find the question worth investigating, then keep your conclusion tied to the evidence available.