Case studies

Subida AI case studies: six projects and the numbers they left behind

Six projects where we were responsible for customer acquisition: over 500K Mentorist users acquired through Facebook Ads, around 1,000 copies of “Cisza w świecie dziecka” sold in the first month, 350 pre-orders for Silne Dziecko, profitable scaling of Neurony Szaleją's digital product sales, and user acquisition with web2app funnels for the English Series and Lighthouse apps.

Every write-up follows the same structure: the context, what we actually did, and what the result was. No "increased engagement" and no charts without axes. Where we don't have a number we can show in the data, we describe the outcome instead of inventing a metric.

Written by Grzegorz Świerad, founder of Subida AI.

Case studies

  1. Mentorist

    B2C app · Facebook Ads · funnel optimisation

    500K+users acquired through Facebook Ads
    Context
    A flagship B2C startup we worked on for several years, in the highly competitive self-development app market — a market where you bid for the same user against brands with budgets many times larger.
    What we did
    Optimising every stage of the user journey — from the first ad creative, through the landing page, to the end of the sales funnel. Quizzes, reworked messaging, systematic analysis of competitors' strategies and more than 500 A/B experiments to separate what really lifts conversion from what merely looks good.
    Result
    Over 500K users acquired through Facebook Ads and funnel optimisation — mostly via ads on Facebook.
  2. Neurony Szaleją

    Digital products · growth partnership · custom sales pages

    Digital product sales scaled while staying profitable — practically every new digital product was sold online at a profit.

    Context
    A digital brand whose audience wasn't naturally inclined to buy digital products at first. We came in as a growth partner, with the goal of building a repeatable sales system rather than just switching on ads.
    What we did
    We started with Facebook Ads and then took over the sales pages: we designed and built dedicated pages ourselves so we could test different sales approaches precisely and learn faster. Today the project runs on a mix of our creatives and the brand founder's own.
  3. Silne Dziecko (Kamil Kajak)

    Book · pre-sale · custom funnel and store

    350pre-orders, or 500 copies sold
    Context
    Kamil came to us looking for someone to set up online sales from scratch — he had no prior experience in this area. The project started with no sales history and no funnel in place.
    What we did
    We took responsibility for practically the whole funnel: the store page design, most of the creatives, and a series of experiments to find which sales approach delivered results fastest. Instead of copying off-the-shelf templates, we built a custom setup designed for rapid testing and quick iteration.
    Result
    350 pre-orders, which came to 500 copies of the book sold — before the product had even reached customers.
  4. Cisza w świecie dziecka

    Digital product · ad-driven launch

    ~1,000copies sold in the first month
    Context
    A digital product for parents, brought to market with a launch driven by Facebook ads.
    What we did
    Running the ad campaign around the launch: messaging built around a specific problem the audience has, and creatives tested while the campaign was live.
    Result
    Around 1,000 copies sold in the first month.
  5. English Series App

    Mobile app · user acquisition · web2app funnel

    User acquisition from ads built on a web2app funnel and analytics, which we used to test new versions of the funnel and shift budget to where conversion was strongest.

    Context
    An iOS app for learning English from TV series and films, aimed at intermediate and advanced learners. Before you watch, it shows the harder vocabulary that will come up and lets you practise it.
    What we did
    Ad campaigns and creatives, messaging that explains how the app differs from ordinary flashcards, a web2app funnel from the ad, through a web page, to install and subscription, and analytics showing at which step users drop off.
  6. Lighthouse

    Subscription app · campaigns · web2app funnel · CRO

    A web2app funnel and analytics that let decisions about new funnel and creative versions rest on data rather than gut feeling.

    Context
    A life-tracker app: weekly summaries of mood and areas of life, daily challenges, and charts showing what actually affects how you feel. A subscription model (Lighthouse PLUS) in a category where it's easy to over-promise.
    What we did
    Ad campaigns and creatives, a web2app funnel from the ad, through a web page, to install and subscription purchase, analytics measuring every step, and A/B tests of successive funnel versions.

Numbers that recur across our projects

The projects above are specific brands and specific launches. Below is what we see on average in campaigns run on our system — first signals usually appear within 7–14 days, and a measurable impact on campaign economics after about 60 days.

average drop in CPA in the first 60 days
-30%average drop in CPA in the first 60 days
ROAS increase after rolling out AI creatives
+150%ROAS increase after rolling out AI creatives
more creatives tested on the same budget
10xmore creatives tested on the same budget

Common questions about these projects

They come from projects run personally by Grzegorz Świerad and from clients' sales dashboards. We don't publish metrics we can't show in the data — which is why Neurony Szaleją has no single number, just a description: its digital product sales scaled while staying profitable.

The first signals usually show up within 7–14 days, and a measurable impact on CPA and ROAS typically after about 60 days. The results above are the outcome of months of work and hundreds of tests, not the first week of a campaign.

We're strongest with digital products: courses, ebooks, apps and books sold online. There, producing one more unit costs close to nothing, so the margin can absorb the acquisition cost and you see faster whether a campaign makes sense.

The volume of testing and control over the entire journey, not just the ad. In every one of them, the result came from testing more variants of messaging, creative and funnel than the competition — at Mentorist that meant more than 500 A/B experiments.

In these particular projects, Facebook and Instagram ads carried the weight of acquisition. We also run Google Ads in our current campaigns, but the results described above come primarily from Meta Ads.

On two models: a fixed retainer, or revenue share — a cut of the revenue we help generate. The choice depends on margin, scale and whether sales can be clearly attributed to the campaign. We've described both models, along with the situations where neither pays off.

Want to see a number like this for your brand?

Show us your product, your price and what's happening in your campaigns today. We'll come back with an assessment of where the biggest potential lies — and an honest answer if now isn't the right time.

Book a free consultation