TikTok Ads Data Analysis & Optimization

SCOPE: $300 overall budget Offer $50 upfront, another $100 if completed, another $150 if the job is done extremely well and thoroughly BACKGROUND: - Every day for the last two months, we’ve launched a series of 1-6 campaigns with similar targeting, all of which has relevant data available for download. Every campaign only runs for one day, the next day 6 more are launched, those are turned off, so on and so forth. - The ad strategy that we employ has shown that TikTok sets certain targeting for specific ad groups at the beginning of the day, and stays with them throughout the day. (For example, targets Women 45+ in Florida in one ad group, Women 24-34 in New York in another) - Per ad group, not all of the audience that TikTok chooses at the beginning of every day perform well, but I have been waiting to spend $6 before cutting it off if there’s not a purchase, and then scaling the ones that have a CPA that are really low. - However, this is not the best ad strategy, as I’m 100% sure there are other early indicators that an ad will either perform throughout the day, or it won’t. This the first task you’ll need to do, and metrics like this are available for download (CPA, Cost per LP view, 25% VV views, etc) you can also make custom formulas within TikTok, and it will automatically populate (Ex, if you want to do metrics acquired per a certain amount spent. For example, TikTok has two second video views total, however, if you do a custom formula for total spend / total video views, you can find total video views per $1 spent, which is much more effective for trend analysis) - We're using a dynamic budget adjustment strategy called 'surfing.' Essentially, throughout the day, we're tweaking our advertising budget to see what works best, after these losers are killed off. The second task for you would be to find the perfect balance where we're spending just enough to get the most conversions at the lowest cost. - In other words, as we increase the budget during the day, we've noticed that the cost per acquisition (CPA) tends to go up. However, there's a point where increasing the budget further doesn't bring us more profit; instead, it starts eating into our gains. We're trying to figure out exactly where that point is by analyzing our data to find the budget level that gives us the highest return on investment. - Finally, I’ve noticed trends in performance not only vary by ad group, but also time of day, and it’s pretty consistent in drops of performance, increases of performance per hour per day by week. This is the final variable you will need to take into consideration when determining your rules GOALS: 1. Discover clear trends of metrics that signify early promise of being a “winning” adsets, as well as metrics that signify the ad will be a losing adset within $2-3 spend, so that if the ad hits $2-3 spend with these metrics above or below a certain amount, we can either know to cut it or scale it These metrics can be things such as looking at the CTR, Cost per LPV, 6 second focused views per $1 spent, etc of losing adsets and seeing clear trends across all of them indicating lousy performance in the future 2. When we find the optimal adsets within a campaign, we want to scale them to get the maximum ROA for our budget. This involves looking at trends not only per day of the week across our dataset, but also hour per day and seeing where there was top performance. Adding budget will make the CPA increase, but also brings more results in. This means you’ll also want to take how much we get per purchase into consideration as a factor for your decision. 3. You’ll want to take into consideration the budget set, CPA, and purchases as well, and see which budget is optimal at certain CPAs with a certain amount of purchases 4. Finally, looking at all of these rules, you will create custom rules for our automatic scaling program, reveal bot, which can cut and scale ads per time of day, metrics acquired, and virtually anything else you can think of- if you spot a trend and want to make a rule- there’s a way to do it! MUST HAVES: - Know how to create formulas for Ads accounts specifically to find trends (past experience doing this. Facebook ads or Snapchat, even twitter is okay) - Experience analyzing LARGE datasets and being able to make significant contributions to clients- must have at least two reputable vouchers - Must speak English - Must be able to pay high attention to detail and general intellect - Friendly, easy to work with - Experience in Excel, Chat GPT, Power BI/Tablaeu or another data analytics software BONUS POINTS: Revealbot or other automatic scaling demonstrable results