Cost per purchasedown 54.70%in a market that kept closing.
Multikino is the largest cinema chain in Poland and part of Vue International. In 2022 the network operated 44 cinemas and 316 screens across 37 cities. The work started in 2017 with no advertising accounts, no historical performance data and no documentation of what had been run before. Below is what that account looked like by 2022, and which figure covers which window.
- -54.70%
- cost per ticket purchase
- -68.78%
- average cost per purchase on Google Ads
- +17.46%
- growth in ticket purchases
- 62.33%
- of account purchases from Performance Max
An account with nothing to learn from, in the years the venues kept closing.
Nothing in the account to learn from
No advertising accounts, no historical performance data, no clear documentation of what had run before. Every assumption had to be earned from a live test instead of read off a past result. That is the slowest way to start an account, and it was the only one available.
Marketing that was measured in recognition
Activity up to that point had been built to keep the brand known. Campaigns were fragmented and inconsistent, with no clear division of roles inside the funnel and no precise targeting. Ticket sales, online and offline, were not what the setup was shaped to move.
A market that kept closing
Cinemas closed and reopened at irregular intervals through the pandemic, while attendance rules, restrictions and public mood changed with them. A plan written in one month was often worth very little in the next, and the account had to be steered rather than scheduled.
The first attempt missed. Everything after it is the case.
The first campaigns underperformed, in the open
Early campaigns did not deliver what was expected. Instead of a quiet swap, the account structure was reworked in front of the client team, with brainstorms, new campaign settings and repeated changes to the build. The cinema side of the knowledge came from their team, the account side went back the other way as training. A test that fails in private leaves the client with a number and no reason for it.
Keyword groups regrouped from SKAG to STAG
Search and Display carried the Google Ads side, with targeting methods combined to reach people choosing a film rather than everyone nearby. Keyword performance was reviewed continuously and the structure moved from single keyword ad groups to single topic ad groups. Not every film was promoted: the mix of evergreen, seasonal and one-off campaigns was a decision each time.
Performance Max and dynamic search ads took the rest of the schedule
Performance Max was introduced and by June 2022 accounted for 62.33% of all purchases generated through the account. Dynamic search ads covered new releases that had no dedicated campaign behind them, so a title did not have to wait for someone to build one before it could be found.
Organic posts turned into paid delivery on Meta
From June 2019 post boosting was added and the whole Facebook advertising side came under the same plan. Boosted posts were aimed at reach and sold tickets anyway, but the part that compounded was quieter: bigger remarketing audiences and finer segmentation for the campaigns that came after them.
The cost of a purchase, in złoty rather than in percentages.
The study gives one pair of figures in absolute terms, from the same account and the same metric. In November 2020 the average website purchase cost PLN 16.32. By June 2022 it was PLN 2.25. That pair is the chart below. Everything else in this case covers a different window, and those figures are named after it.
Two bars, not four, because the other figures do not share a scale with these. The 54.70% fall in cost per purchase is a Meta figure comparing May 2019 with April 2022. The 68.78% fall in average cost per purchase is a Google Ads figure comparing January 2018 with April 2022. Both stay in the strip at the top with their own labels. A bar chart cannot show direction, so a falling cost drawn next to rising purchases would read as two movements the same way, which is the opposite of what happened. One rough edge is worth naming rather than smoothing: the study reports 17.46% growth in ticket purchases and, in the same list for the same comparison, an overall increase of more than 401.53%. Those cannot both describe the same metric over the same months, the study does not say which is which, so the larger one is not used here at all. A last pair from the source, because it complicates the picture: between October 2017 and April 2022 the share of purchases coming from paid campaigns moved from 3.82% to 5.10%, while their share of users moved from 4.39% to 12.41%. Traffic share grew faster than purchase share, and that gap is in the study too.
“I was impressed with the effectiveness of campaigns. They’re organized and accommodating. Their team adapted their workflow, centering it around the variable cinema market in Poland.”Elżbieta Polkowska, Manager, Multikino
The part you can copy is the year before the numbers moved.
The first campaigns here missed, and that is the part worth copying. The response was to rebuild the structure with the client in the room, rather than quietly changing the creative and hoping nobody asked. The study puts a year between that start and the point where performance held steady. If your account cannot survive a year of being visibly wrong, the thing to fix first is the reporting line, not the campaigns.
The second thing worth taking is the footnote on that 17.46%. It was measured while the attribution model moved from 28+1 to 7+1, a shorter window that credits fewer sales to the same ads. If your own window narrows part way through an engagement, your reported numbers drop without your account doing anything differently, and somebody upstairs will read that as a performance problem. Agree how you will report the change before it lands, not after your dashboard does.
This work was delivered by MTA Group, which Zero Fluff Digital is part of, by the same specialists who would work on your account. Every figure here comes from the case study MTA published, republished here with their consent. The reviews behind it are public and verified on Clutch, where clients rate the work rather than the agency describing itself.
A purchase costs you something other than what the panel reports.
30 minutes, no deck. You find out what your accounts pay per sale, and which of your reported numbers would survive a change to the attribution window.
Book a callNext case studyRoltomrolety: purchases up 57.68% on Google Ads →