Case Study: Down $718 in Week 3, Net $2,286 After 30 Days

Case Study · Issue 9

Hi, I’m Peter.

This case comes from the ProPush official blog, which collected several posts on the AffLIFT forum where people tested Constructor with real money.

Issue 6 looked at what it’s like when a landing page edit goes wrong. This one looks at people who ran it with real money: how they lost, how they hung on, and how much they finally made.

About the Threads

The data comes from the ProPush official blog (September 17, 2026). This article is published with ProPush’s authorization.

01 How It Differs from a Smartlink

AffLIFT is a very active forum in the industry. A Follow Along thread means posting while you run: dashboard data, screenshots, wins and losses, updated every day.

Recently several people started their own threads testing Constructor. In those threads, they don’t quite agree on what it is.

Luke, the owner of AffLIFT, treats it as a layer of arbitrage: you buy traffic on an ad platform, send it through a Constructor page, then sell it through ProPush and keep the spread.

dood says it’s a bit like a Smartlink, similar to Zeydoo offers: the user goes to the end of the page, and which offer they land on isn’t known in advance.

The official line is that both descriptions are right, but there’s one difference, and it directly affects the final result.

A quick recap first: a Constructor page can stack up to 9 monetization slots. Whatever action the user takes on the page, they leave through the matching exit, and even if the offer doesn’t convert, the impressions can still earn money.

Karina explained it this way: with a Smartlink, the user lands directly on the advertiser’s offer. With Constructor and Zeydoo offers, the user first plays with the landing page, and only then gets to pick an offer.

According to Karina, playing with the page first makes users more engaged, so the CPM ends up higher than sending them straight to a Smartlink.

02 Building the Page Is the Easy Part

You can edit the page yourself: change the headline, button copy, icons and CTA, and the same flow gets a different angle. You can also just tell the AI assistant what you want changed. You can even download the template as a package and edit it on your own server. If you don’t want to change anything, the default version runs as it is.

Almost nobody in the threads had complaints about building the page:

• Drinko said building the page is fast and simple, “honestly very impressive”;
• Yordan said he creates a new Constructor for every campaign he opens;
• Before launching, Johnny_eX used ProPush’s AI assistant to rework the landing page.

Luke’s thread put it most clearly: it simply generates a landing page that can take traffic. Outside the page, he also set up a Vultr server, bought a domain, and configured a tracker and postback, and he had AI do all of that too.

The page takes a few minutes to build. What takes time is everything outside the page.

03 5 Pitfalls from the Threads

Karina started with one line:

Like any other offer in this industry, Constructor is not a magic pill that hits 100% ROI on day one. In some ways it takes more work than running a Smartlink directly, but the extra effort up front pays off, so don’t give up too quickly when problems come up.

1. Closing too early: By day three, Drinko had set a rule: if there's no conversion for two days, shut it off. The campaign that eventually worked happened to break that rule: it sat on a small budget, he almost forgot about it, and on day 12 its ROI turned positive. The ones he shut off had no conversions at all, while the one he left running had income the whole time, just not enough yet.
2. Changing too much at once: L1Ght started out running two landing page designs, two monetization networks and several countries at the same time. Every group had too little data to tell which was better. Later he pulled his own data and saw that the second Clickbox design kept people and had a low exit rate, so he ran only that one, and after that changed one thing at a time.
3. Spreading the budget too thin: In the same thread, each of L1Ght's campaigns got about $2 and 115 impressions a day, which looked more like noise. He stopped the weakest countries and put the money into the three most promising ones. After his account manager looked it over, the verdict was the same: too little traffic.
4. Trusting whitelists too much: L1Ght split his best-performing zones out into their own campaigns. One ended up stuck at -76%, and a zone in another died at around -90%. He later admitted that the same zone in a different campaign, even in the same country, can perform completely differently, so you can only optimize inside each campaign separately.
5. Testing only one traffic source: Someone asked Varun which platform and format worked well. He said Push, Pop, Direct and Social all work, wouldn't say which exact one, and told them to keep testing. A source that dies for you may be very profitable for someone else, and one failed combination is just one data point.

These pitfalls have little to do with Constructor itself. You can step in them with any campaign.

04 The Results

It was dood’s first test, and he picked this country because the traffic was cheap. His ROI matches Desha’s. Desha is a case ProPush’s official blog covered earlier: $31k earned with Constructor in one month, on a budget 20 times dood’s.

In week 3, Anthony was still down for the month, but the last 7 days were already at +52.8%. Two more weeks later, the 30-day total came to $2,286 net profit.

He wrote in the thread: 30 days, losses at the start, and it kept getting better.

Two days without traction is only the first chapter of the story.

05 5 Questions Karina Answered

Nearly every thread leaves a few questions behind:

1. How long is the warm-up: Three posters heard three different numbers from their own account managers: eCPM starts to climb after about 1,000 impressions over 3-5 days, 5,000 per zone, or 50,000. The mechanism is the same, and Anthony sums it up in one line: "The zone is new, the algorithm has to judge quality first, and junk traffic doesn't monetize, which is a good thing." So warm-up is measured in impressions per zone. A zone that gets 10 impressions a day will never warm up, however long you leave it on. Get volume first, then talk about patience.
2. Where did the traffic go: The impressions you buy and what you see in Constructor will never match 1:1. The user has to do something on the page before an exit fires: some don't move, some close it right away, some leave halfway through. The question to ask is why they left: weak traffic quality, or a page that isn't engaging enough. Just look at final ROI and the eCPM of each exit zone, and if one zone drops badly, shut it off. There's no need to report a bug.
3. When to shut it off: There's no standard answer. If the traffic is running normally, two days without a conversion is a reasonable line, since two days is enough for a normal zone to monetize and ramp up. Shutting off after an hour or two is worse, because you may have just hit the wrong time of day. Whatever you run, set a spend cap on the campaign.
4. The tracker shows less than the dashboard: Most likely the tracker is set to count only one conversion per user. In Constructor, the same user can subscribe to push first and then go to the main exit, and some versions have two main exits. If the tracker doesn't accept duplicate conversions, these get merged into one. Before judging the page, turn on multiple conversions. In Desha's case, the shortfall came from some leads not being recorded at all, so check both sides before blaming the page.
5. Is a whitelist worth doing: Yes, but don't run only whitelists. Put verified sources on the whitelist and you can use a higher budget and bid, but also keep a few blacklist-only campaigns, because new placements keep appearing in a traffic source and buying only the whitelist means you never see them. The next improvement mostly still comes from new creatives, new angles and landing page changes.

06 My Take

In these threads, nobody made money on day one. The ones who did make money got through the losing stretch at the start first.

That includes my own Constructor runs, which also took a month to get working. I lost quite a bit along the way too.

In this project, patience, a testing method and luck are all essential.

The two points I agree with most in this issue: warm-up is measured in impressions per zone, and change only one thing at a time.

If you have Push/Facebook traffic, pick a ready-made page from Constructor in the ProPush dashboard and get it running first. Set a spend cap, warm up by impressions per zone, and don’t draw conclusions after an hour or two.

Note: the income and ROI figures in this article are personal or case-specific historical results and do not represent typical outcomes. Paid traffic carries a risk of loss, so test with a small budget first.