Hi there,
Thanks for reaching out!
It sounds like you're running into a really common, and very frustrating, problem with TikTok ads. That massive drop in performance after a brilliant first day is something we see a lot. It's not that you're doing something wrong necessarily, it's more about understanding the beast that is the TikTok algorithm and not falling into the traps it sets.
The short answer is that the algorithm is designed to give you that initial sugar rush. It finds the easiest, most likely buyers first to prove the ad can work. The real job, and where most people falter, is what you do on day two and beyond. I’ll give you some of my thoughts on how to build campaigns that have a bit more staying power.
TLDR;
- Your high day-one ROAS is misleading. The TikTok algorithm finds the 'low-hanging fruit' first, then performance normalises as it seeks a wider audience. Don't panic and kill campaigns on day two.
- Creative fatigue on TikTok is hyper-accelerated. An ad can burn out in 48 hours. You need a system for constantly creating and testing new ad concepts, not just finding one 'winner'.
- Stop relying on a single campaign structure. You should separate your testing from your scaling. Use a dedicated testing campaign to find winning ads and audiences, then move them to a separate scaling campaign to give them budget.
- The most important piece of advice is to shift your mindset from finding winning ads to building a winning *system* for producing and testing ads.
- This letter includes a unique 'Creative Burnout Calculator' to help you estimate how quickly your ads will fatigue based on your spend and audience size.
We'll need to look at The TikTok 'Day One' Trap...
First off, let's get one thing straight. That 8x ROAS on day one isn't real. Well, the sales are real, but the performance is an illusion created by the algorithm. When you launch a new ad, TikTok's primary goal is to spend your money effectively and get you a result as fast as possible to keep you advertising. To do this, it goes straight for the jugular.
It takes your ad and shows it to the tiny sliver of your target audience that is most likely to convert *right now*. These are the people who have been browsing similar products, who are in a buying mood, and whose behaviour screams "I'm about to purchase". This is the 'low-hanging fruit'. Of course your ROAS is going to be amazing. The system is basically cheating for you.
The problem comes on day two. That little pocket of hyper-engaged users is now exhausted. The algorithm, having proven it can get a result, now has to do its real job: find new customers from the rest of your much broader, much less-engaged audience. This is when the ROAS plummets to 3x, or even lower. This isn't the campaign failing; it's the campaign *starting*. Most advertisers see this drop, panic, and kill the ad, restarting the cycle and never leaving this frustrating loop. You need to let it run for at least 3-5 days to get a true picture of performance.
This process is so predictable you can almost set your watch by it. Here's a simple way to visualise the journey your ad goes on in its first few days:
Algorithm targets hyper-receptive "low-hanging fruit".
Easy wins are gone. Algorithm explores the broader, colder audience.
Performance finds its true baseline. This is your real number.
I'd say your creative is burning out, not your campaign...
The second part of this puzzle is creative fatigue. On platforms like Facebook, an ad might last for weeks. On TikTok, an ad can be completely burned out in 48-72 hours. The audience is scrolling faster, consuming content faster, and getting bored faster. That one brilliant creative you made isn't a long-term asset; it's a disposable entry ticket.
You can't solve this by trying to make the *perfect* ad. You solve it by building a *system* to produce a constant stream of *different* ads. Your goal shouldn't be to have one winning ad running, but to have 3-5 different ads testing at any given time. I remember one campaign we worked on for an app where we helped them get over 45k signups using TikTok Ads as a primary channel. They weren't running one ad; they were running dozens of variations.
Think about different angles, not just different visuals:
- Problem-Agitate-Solve: Start with the customer's pain point. "Hate that your clothes fade after one wash? It feels like you're throwing money away. Our new detergent locks in colour..."
- Before-After-Bridge: Show the transformation. "This was my skin before (show bad skin). This is my skin now (show great skin). The bridge was our new serum..."
- UGC-Style: Get someone to just talk to the camera as if they're a real customer. We had several SaaS clients see huge results with simple, unpolished user-generated style videos. They feel more authentic and less like an ad.
- Trend-Jacking: Use a popular TikTok sound or format, but adapt it to your product. This has a very short shelf life but can provide a quick burst of cheap traffic.
The speed of this burnout depends on your budget and audience size. The more you spend and the smaller your audience, the faster people will see your ad multiple times and tune it out. I've built a simple calculator below to help you get a feel for this. It's not an exact science, but it illustrates how quickly you can saturate an audience.
You probably should use a proper testing structure...
Given the issues above, just throwing a new creative into a campaign and hoping for the best is a recipe for inconsistent results. You need to separate your exploration (testing) from your exploitation (scaling). This means having at least two types of campaigns running at all times.
Campaign 1: Creative Testing (Ad Set Budget Optimisation - ABO)
The goal here is purely to find winning ads and audiences. You use ABO (setting the budget at the ad set level) to ensure each of your audience tests gets a fair amount of spend.
-> Ad Set 1: Broad Audience - Test Creative A, B, C
-> Ad Set 2: Interest Stack 1 - Test Creative A, B, C
-> Ad Set 3: Lookalike Audience 1% - Test Creative A, B, C
You run this with a small budget. After a few days, you look at the data. Did Creative B work best across all audiences? Did Interest Stack 1 outperform the others, regardless of creative? This gives you your winners.
Campaign 2: Scaling (Campaign Budget Optimisation - CBO)
This is where you make your money. You take the winning ad sets and winning creatives from your testing campaign and combine them here. By using CBO (setting the budget at the campaign level), you let TikTok's algorithm automatically push the majority of the budget to the best-performing combination of ad and audience in real-time. This campaign gets the majority of your total budget.
This structure gives you a perpetual motion machine. You are constantly feeding new ideas into the testing campaign, and graduating the winners into the scaling campaign, while turning off the losers. Your overall account performance becomes far more stable because it's not reliant on one single ad that's destined to fail.
You'll need to think beyond the ad itself...
Finally, no amount of ad optimisation can fix a broken offer or a website that doesn't convert. All the problems we discussed get worse if your underlying foundations are weak. When your ROAS drops from 8x to 3x, that's still a profitable result for many businesses. If a 3x ROAS is a disaster for you, it might mean your prices are too low, your conversion rate is poor, or your customer lifetime value is not high enough to support paid acquisition.
I've seen campaigns achieve dramatic turnarounds in profitability, but it rarely comes from just changing the ads. It's almost always a combination of better targeting, improved ad creative, AND working with the client to improve the landing page experience. The ad is only the first step. You need to analyse the entire funnel.
- Low CTR? -> Your ad creative or targeting is the problem.
- High CTR but no Adds to Cart? -> Your landing page, product photos, description or pricing is the problem.
- Lots of Adds to Cart but no Purchases? -> Your checkout process is too complicated, you have unexpected shipping costs, or your site doesn't feel trustworthy.
Be brutally honest with where the drop-off is happening. Sometimes the reason your ads 'fail' on day two is because the initial hyper-motivated buyers were willing to overlook flaws on your site that the general public simply won't tolerate.
This is the main advice I have for you:
To pull this all together, here is a table summarising the key problems and the recommended actions. This isn't a list of quick fixes, but a framework for a more sustainable approach to TikTok advertising.
| Problem | My Recommendation | Why It Works |
|---|---|---|
| Performance is great on Day 1, then crashes. | Don't react immediately. Let campaigns run for at least 3-5 days before judging performance. | This allows the campaign to exit the initial 'learning phase' and find a stable baseline performance after the low-hanging fruit is gone. |
| A good ad stops working after a couple of days. | Build a creative pipeline. Aim to test 2-3 new ad concepts every single week. | TikTok has hyper-accelerated creative fatigue. Constant testing is the only way to stay ahead of audience boredom. |
| Unsure which new ads or audiences to try. | Separate testing from scaling. Use an ABO 'Testing' campaign and a CBO 'Scaling' campaign. | This provides a structured, data-driven way to identify winners and allocate budget efficiently without disrupting your main campaigns. |
| Even stable campaigns aren't profitable enough. | Analyse your funnel. Look at your website analytics to find where users are dropping off (landing page, product page, checkout). | Often, the biggest gains in ROAS come from fixing conversion rate issues on your website, not from the ads themselves. |
As you can see, getting consistent results from TikTok isn't about finding a single 'hack'. It's about implementing a professional, disciplined system. It requires a constant effort in creative production, data analysis, and strategic testing which, frankly, can be a full-time job in itself.
If managing this kind of process sounds daunting, it might be worth considering getting some expert help. We specialise in building and managing these exact kinds of scalable advertising systems for our clients.
I'd be happy to offer you a free, no-obligation consultation where we can take a direct look at your ad account together. We can analyse your current setup, look at your data, and give you some more specific, actionable advice on what your next steps should be. Sometimes a second pair of expert eyes is all it takes to spot the opportunities you've been missing.
Regards,
Team @ Lukas Holschuh
