Roblox Ads: Plays vs Engagement Campaigns (What We Learned)

Roblox ads are the fastest way for a new game to get its first few hundred players. They are also the fastest way to spend money and learn nothing. This post covers what we saw running sponsored campaigns in Roblox Ads Manager for two of our games, Shopping Cart Mayhem and Car Wash Tycoon: how the Plays and Engagement goals behaved, why clicks are a misleading number, and the settings we now change on every new campaign.

One note before we start. Roblox changes its ad products, audience rules and eligibility thresholds often. Everything below is what we observed in October 2026. For the current rules, check the Creator Hub documentation and your own Ads Manager screens before you spend.

Why a new game needs ads at all

A new experience has no history, so Roblox's discovery has nothing to rank it on. Without some first traffic, you have no funnel data, no retention numbers and no idea whether your first five minutes work. A small, controlled ad run fixes that. We treat ads as a measurement tool first and a growth tool second.

That framing matters for the rest of this post. If you buy ads to "get players", you will optimise for the cheapest player. If you buy ads to learn, you will optimise for players who behave like real fans.

Plays vs Engagement: what the goals do

When you create a campaign, Ads Manager asks for a goal. The two we tested:

Goal	What it optimises for	What we saw
Plays	Getting people to click and join	Lots of impressions and clicks, cheap joins, very few people who stayed
Engagement (Beta when we used it)	People who are likely to spend real time in the game	Fewer clicks, but the joins looked more like real players

Our first campaign for Shopping Cart Mayhem used Plays. On paper it was a success: tens of thousands of impressions and around fifteen hundred clicks. Then we looked at highly engaged players, the metric that actually mattered to us, and the campaign had brought in single-digit numbers. Almost all of those clicks were people who joined, looked around and left.

We paused it and started an Engagement campaign with the same thumbnails.

The click trap

High click-through looks like a good thumbnail. It can also mean your thumbnail promises something the game does not deliver in the first minute, so players bounce. Or it can mean the Plays goal found the audience that clicks on everything. Either way, clicks tell you about the ad, not the game.

The numbers that tell you about the game:

How many joiners reach your first real milestone (for us, the first escape from the store).
How long a session lasts.
Whether they come back the next day.
The engaged-player count in the experience dashboard.

If you only have budget to watch one, watch the one closest to "this person liked the game."

Highly engaged players and the audience gate

This is the part that surprised us, and the part to double-check against the Creator Hub for your own game.

Both games launched with a restricted audience setting. The experience's audience reach page showed progress toward a target number of highly engaged players, and reaching it was the route to a wider audience. For a kid-focused game that is a big deal, because a restricted audience shuts out most of the players you built the game for. There was also a paid expedited review offered as an alternative; check the current terms in your dashboard if you are weighing it.

Two things we learned about this gate:

The maturity questionnaire is separate. Filling in the questionnaire (Configure, then Questionnaire) gives your game its content label. It did not by itself lift the audience restriction. The engaged-player progress was the gate in our case.
Plays campaigns barely move it. Thousands of low-intent clicks produced almost no engaged players. That is why we switched both games to the Engagement goal.

This is also why ads and retention work go together. An Engagement campaign can only find engaged players if the game gives them a reason to stay. When we fixed the first-session leak in Shopping Cart Mayhem (see our retention case study), the same ad traffic turned into far more players who reached the good part.

Settings we change on every campaign

Small budgets are easy to burn by accident. The create form has defaults that suit big advertisers, not a two-person studio testing a new game.

Turn off Auto-Reload

In our experience the create form defaulted to automatically reloading credits when the balance ran low. For a test campaign, turn it off. You want the campaign to stop when the test budget is gone, not keep spending while you sleep.

Use a lifetime budget with an end date

The form also defaulted to running continuously. We switch to a lifetime budget with fixed start and end dates, a few days long. A test with an end date is a test. A campaign with no end is a subscription.

You cannot change the goal later

We found that a campaign's goal could not be edited after creation. Switching from Plays to Engagement meant pausing the old campaign and creating a new one. Decide the goal before you launch.

Expect a review step

New campaigns sat in a scheduled or moderation state for a while before serving. Plan your test window around that, especially if you are timing it with an update.

Thumbnails that show the game

Your ad thumbnail is the first five seconds of your game, so it should look like the first five seconds of your game.

What worked for us:

Real gameplay, real place. The Shopping Cart Mayhem ads showed actual heist moments, like the airport map and outrunning the guards. Car Wash Tycoon showed the neon street the player actually builds.
One action, one line of text. "Build Your Empire" over a street at night. "Ride a Dino" over a dinosaur. Not a feature list.
Readable on a phone. Most impressions are small. If the text needs zooming, it is too small.

What did not work: art that looked more polished than the game. It may lift clicks, but players who join expecting that art leave fast, and Engagement campaigns will learn to avoid your game.

Read results after the reporting delay

Ad dashboards are not real time. Reporting on our campaigns lagged by about two days. If you judge a campaign on day one, you are judging partial data.

Our routine:

Launch with a fixed budget and end date.
Wait at least two days after the campaign ends.
Compare campaigns on engaged players per credit spent, not clicks per credit.
Cross-check against your own analytics funnel, filtered to the build that was live during the campaign.

Step four is the reason we tag every analytics event with a build label. Without it, you cannot tell whether a better result came from the ad or from the update you shipped the same week.

A simple playbook

Fix the first session before you spend. Ads amplify whatever your first five minutes do.
Choose the Engagement goal if your aim is real players.
Turn off Auto-Reload and continuous running. Set a lifetime budget and end date.
Use gameplay thumbnails with one clear action.
Wait for the reporting delay, then compare engaged players per credit.
Change one thing per campaign, so you know what moved the number.

Ads did not make our games good. They showed us, quickly and honestly, where the games were not good yet. That is worth paying for. You can see the games these campaigns pointed at on our games page, and how we design the first minute in the 5-second hook.