# Incrementality for game marketing on a small budget

> How indie developers can measure the wishlists a campaign really added with baseline, holdout and on/off tests, using Steamworks data, not lift studies.

By Yos Riady. Updated 2026-11-19. Published 2026-11-18. Canonical URL: https://manalens.dev/blog/game-marketing-incrementality

Incrementality is the number of wishlists a campaign added that would not have happened without it. Platform lift studies need scale: Google's Conversion Lift asks for 1,000 observed conversions and a $5,000 minimum budget (Google Ads Help). Indies can run baseline, country holdout and on/off tests with Steamworks wishlist data instead.

## Key points

- Google Conversion Lift needs 1,000 observed conversions and a $5,000 minimum budget, and it cannot see Steam wishlists at all.
- For a baseline test, subtract the average daily wishlist adds of the 7 to 14 days before from the wishlists in the campaign window.
- In the Glyphica test, $500 of Reddit ads gave a lift of 159 wishlists at $3.14 each, against $5.49 on 91 UTM wishlists.

## What is incrementality?

The difference between what happened with a campaign and what would have happened without it. The second part is the counterfactual. You cannot observe it directly, so you estimate it from a baseline period or a group of regions that did not see the campaign.

Meta's open source GeoLift tool describes it the same way: the incremental effect is the observed result minus the result in a world where the campaign did not take place.

Sources: GeoLift documentation, Methodology.

## Why does incrementality differ from UTM attribution?

UTM counts clicks that Steam could follow. Incrementality counts every extra wishlist, including players who saw the ad and searched later. It can also be lower than UTM, if the people who clicked would have wishlisted anyway.

Two published cases show the gap. In the Glyphica Reddit test, Steam attributed 91 wishlists to the ads, while the week had 159 more wishlists than the control week before it. In a Green Man Gaming campaign, UTM saw 24% of the wishlists that the lift showed.

Sources: Squeaky Wheel (21 June 2024); Green Man Gaming (24 February 2026).

## Can I use platform lift studies?

Usually not for a Steam game. Google Conversion Lift needs at least 1,000 observed conversions, a $5,000 minimum campaign budget and a conversion action that Google can track. A wishlist on Steam is not a conversion that Google can see, because Steam runs no ad tags.

Meta's GeoLift is open source and works on regional data you supply, such as wishlists per country per day. It uses synthetic control: it builds a weighted mix of regions that matches your test regions before the campaign. It runs in R and includes power calculators that tell you if a test can detect the lift you expect. It is a good tool if you have the data and the time.

Sources: Google Ads Help, Conversion Lift; GeoLift documentation; Steamworks Google Analytics documentation.

## Which incrementality tests work on a small budget?

Three simple designs cover most indie campaigns. Each needs daily wishlist adds, which Steamworks updates once a day for the day before.

*Low-budget incrementality tests and their main risk.*

| Test | How it works | Needs | Main risk |
| --- | --- | --- | --- |
| Baseline comparison | Compare daily wishlists during the campaign with the 7 to 14 days before | Daily wishlist adds | Anything else that changed in the same days |
| Country holdout | Run ads in some countries only, compare with similar countries without ads | Wishlist adds per country per day | Too few wishlists per country to see a change |
| On/off switchback | Turn ads on for a week, off for a week, and repeat | Daily wishlist adds over several weeks | Effects that carry over into the off weeks |

Sources: Steamworks Wishlists documentation; Squeaky Wheel (21 June 2024).

## How do I run a baseline test?

Measure a quiet week, then the campaign week, and keep everything else still.

1. Pick a campaign window with no festival, sale, demo update or planned creator video.
2. Average the daily wishlist adds of the 7 to 14 days before. Use the same weekdays if you can.
3. Run the campaign with tagged links, so you also have the UTM count.
4. Sum the wishlist adds in the window and subtract the baseline for the same number of days. This is the lift.
5. Divide spend by the lift. Compare it with spend divided by UTM wishlists.
6. Write down anything unplanned that happened, such as a viral post.

Sources: Squeaky Wheel (21 June 2024); Steamworks Wishlists and UTM Analytics documentation.

## What does a real baseline test look like?

Glyphica ran Reddit ads from 27 May 2024 and compared with 19 to 26 May. The control period had 227 wishlists, an average of 28 a day. The ad period had 386, an average of 55 a day. The team counted 159 more wishlists than the control, of which 91 were attributed to Reddit by UTM.

Our arithmetic: on the $500 spent in that week, the lift cost is $500 / 159 = $3.14 per wishlist, and the tracked cost is $500 / 91 = $5.49. The team stopped using later days because a showcase overlapped them. That is the right call.

*Wishlists in a control period and a campaign period.*

| Period | Wishlists | Per day |
| --- | --- | --- |
| Control, 19 to 26 May 2024 | 227 | 28 |
| Ads, 27 May to 2 June 2024 | 386 | 55 |
| Difference | 159 | 27 |

Sources: Squeaky Wheel (21 June 2024).

## What spoils an incrementality test?

Other beats in the same window. Steam sales, festivals and wishlist emails all move the daily count. Steam emails wishlisters at launch and for discounts of 20% or more that last over 8 hours. Melos Han-Tani's Danchi Days log shows how much can overlap: in one month of Reddit ads, he split 2,185 wishlists into ads, sales events, baseline and an unaccounted share.

Small numbers also spoil tests. If you get 10 wishlists a day, a lift of 5 a day is hard to tell from noise. Run longer, or accept a range instead of one number.

Sources: Steamworks Wishlists documentation; Melos Han-Tani (12 April 2026).

## How do I run a country holdout?

Pick countries that behave alike, run ads in half of them, and compare. This design handles a Steam sale or a festival better than a baseline test, because both groups see the same event.

1. Export daily wishlist adds per country for the last 4 to 8 weeks.
2. Choose pairs of countries with similar daily wishlists and a similar trend, for example two Nordic or two Latin American countries.
3. In each pair, run ads in one country and none in the other. Keep everything else the same.
4. After the campaign, compare the rise in daily wishlists in the test countries with the rise in the holdout countries.
5. Divide spend by the extra wishlists in the test countries. With few wishlists per country, report a range.

Sources: GeoLift documentation; Steamworks IPartnerFinancialsService documentation.

## Where do I get the daily data for a lift test?

Manalens is the analytics and growth engine built for games. Connect a Steamworks Financial API key (the automatic Steam sync is in early access) to sync wishlist adds and deletes per day and country, and import the UTM report for the tracked count. The product does not compute lift for you; it puts both numbers side by side.

- [Wishlist velocity calculator](https://manalens.dev/tools/wishlist-velocity-calculator)

Sources: Manalens product documentation.

## Frequently asked questions

### What is incremental cost per wishlist?

Spend divided by the wishlists that would not have happened without the campaign. You estimate those wishlists as the wishlist adds in the campaign window minus a baseline. It differs from the UTM cost per wishlist, which counts only tracked clicks.

### Can a small indie run a geo holdout test?

Yes, if enough wishlists come from each country. Run ads in some countries, keep similar countries without ads, and compare daily wishlist adds per country. With few wishlists per day, the change can be too small to see, so use a longer test.

### How long should an incrementality test run?

At least one full week, plus a baseline of 7 to 14 days before. Google recommends 14 days or more for its own Conversion Lift studies. Avoid windows with Steam sales, festivals or planned creator videos, and run longer if you get few wishlists a day.

### Is UTM attribution useless if I measure incrementality?

No. UTM rows compare ads, creators and posts under the same counting rules, which a lift test cannot split. Lift tells you how big the whole effect was. Use UTM to choose between options, and lift to decide how much to spend.

## Sources

- [GeoLift documentation: Methodology (Meta Open Source, read 18 November 2026)](https://facebookincubator.github.io/GeoLift/docs/Methodology/)
- [Google Ads Help: Set up Conversion Lift based on users (read 18 November 2026)](https://support.google.com/google-ads/answer/12005564?hl=en)
- [Squeaky Wheel: How many wishlists can $500 worth of Reddit ads get you? (Ryan Sumo, 21 June 2024)](https://www.squeakywheel.ph/blog/how-many-wishlists-can-500-worth-of-reddit-ads-get-you)
- [Green Man Gaming: The Steam attribution black hole (Adele Simpson, 24 February 2026)](https://www.greenmangaming.com/company/news-category/latest-news/steam-utm-marketing-attrbition-black-hole/)
- [Steamworks documentation: Wishlists (Valve, read 18 November 2026)](https://partner.steamgames.com/doc/marketing/wishlist)
- [Steamworks documentation: Google Analytics (Valve, read 18 November 2026)](https://partner.steamgames.com/doc/marketing/google_analytics)
- [Melos Han-Tani: Danchi Days, 3600 wishlists through Reddit Ads (12 April 2026, edited 31 October 2026)](https://buttondown.com/melodicambient/archive/danchi-days-3600-wishlists-through-reddit-ads-was/)

## Related guides

- [Game marketing measurement guide](https://manalens.dev/blog/game-marketing-measurement-guide)
- [Cost per wishlist benchmarks by channel and genre](https://manalens.dev/blog/cost-per-wishlist-benchmarks)
- [Game marketing attribution explained](https://manalens.dev/blog/game-marketing-attribution-explained)
- [Paid ads for indie games: which platform gives the cheapest wishlists](https://manalens.dev/blog/paid-ads-for-indie-games)
- [Wishlist velocity calculator](https://manalens.dev/tools/wishlist-velocity-calculator)
- [Gamesight alternatives: 6 tools for PC game attribution](https://manalens.dev/compare/gamesight-alternatives)
- [Game marketing guides](https://manalens.dev/blog/topic/game-marketing)
- [Marketing mix modelling vs attribution for indie studios](https://manalens.dev/blog/marketing-mix-modelling-vs-attribution-games)
- [Best game marketing attribution tools for PC and web games in 2026](https://manalens.dev/blog/best-game-marketing-attribution-tools)
- [Best Steam attribution and wishlist tracking tools in 2026](https://manalens.dev/blog/best-steam-wishlist-tracking-tools)
- [Devlogs: do they sell games? What the numbers show](https://manalens.dev/blog/do-devlogs-sell-games)
