Blog / Affiliate marketing
Autonomous Campaign Manager: AI Runs Your Campaigns While You Sleep
Affiliate campaigns don't sleep. Traffic flows around the clock, results shift every hour, and a poor campaign burns through budget just as efficiently at 3 AM as it does at 3 PM. Meanwhile, you're asleep, in a meeting, or simply living your life - and you can't react to every anomaly in real time.
An autonomous campaign manager is a system of rules and AI tools that does this for you. It monitors performance, pauses campaigns that stop being profitable, scales the winning ones, and responds to anomalies - without your intervention. In this article, I'll show you how to build such a system for your MyLead campaigns, which tools to use, and what rules to set so it runs safely.
What you'll learn from this article
How much manual campaign management really costs and why that cost stays invisible
How an autonomous campaign manager works at the level of monitoring, rules, and system learning
What specific thresholds to set for pausing rules, scaling rules, and alerts
How Meta Automated Rules, Revealbot, and Madgicx differ and when to choose each
How to configure the entire system step by step, without risking your budget
Which mistakes most often ruin the results of campaign automation
The cost of manual campaign management
Manual campaign management has three hidden costs that most publishers never calculate.
Cost of delayed reaction. A campaign starts losing profitability at 10 PM. You notice it at 8 AM. For 10 hours, budget flowed in a direction that should have been stopped long ago. With a daily budget of 200 PLN, that's 83 PLN burned outside the optimal window.
Cost of missed opportunities. A campaign starts scaling organically in the middle of the night - results are 40% better than average. By morning, the peak has passed. If the system had automatically increased the budget by 20%, you would have earned more without any additional effort.
Cost of attention. Checking several campaigns every few hours is work that takes time without generating value. An autonomous system frees that attention for tasks that actually require your thinking: strategy, new niches, new creatives.
How an autonomous campaign manager works - layers of automation
An autonomous campaign manager works across three layers that together form a complete system:
Layer 1: Real-time monitoring. The system monitors key metrics (CTR, CPC, CPL, ROAS, conversion rate) and compares them against established thresholds and each campaign's historical baseline. Any deviation above a defined threshold triggers an evaluation.
Layer 2: Decision rules. Based on the evaluation, the system applies predefined rules: pause, scale, change bid, send alert. Rules can be simple ("if CPL > X, pause") or complex ("if CTR drops >20% with stable budget and CPL rises >15%, pause and send notification").
Layer 3: Learning and adaptation. Advanced AI systems (like Madgicx or Revealbot) learn from historical campaign data and adjust decision thresholds to account for seasonality and traffic patterns. A system that has been running for a month makes better decisions than one launched a week ago.
Rules for automatic pausing and scaling - concrete thresholds
The rules below are a starting point - adjust the thresholds to your niche, margin, and historical campaign results.
Pausing rules
CPL exceeds the profitability threshold - if the cost per lead exceeds 80% of the lead commission value for 6 consecutive hours, pause the campaign and send an alert. Example: commission 50 PLN, pause threshold CPL = 40 PLN.
CTR drops >30% relative to the 7-day average - with a stable budget, this indicates a burned-out creative or a saturated audience. Pause the ad set (not the entire campaign) and launch a new creative.
Zero conversions for 48 hours with spend >2x the commission value - the campaign isn't converting despite sufficient budget. Pause and investigate the cause before relaunching.
Sudden CPM increase >50% without a conversion increase - may indicate ad delivery issues or changes in the auction. Pause and check account diagnostics.
Scaling rules
ROAS above target for 3 consecutive days - increase the daily budget by 20%. No more - larger jumps reset the learning phase of the platform's algorithm.
CPL dropped >20% relative to baseline with a growing number of conversions - the campaign is entering its optimal phase. Increase budget by 30% and monitor for 48 hours.
A creative has CTR >2x the campaign average for 72 hours - duplicate the ad set with that creative and increase its budget by 50% at the expense of weaker ad sets.
Alert rules (no automatic action)
Daily spend exceeds 120% of the planned budget
Conversion rate drops >40% compared to the previous week
Campaign reaches 80% of monthly budget cap before the 20th of the month
Tools: Meta Automated Rules, Revealbot, Madgicx
Meta Automated Rules - native rules at no extra cost
Facebook Ads Manager has a built-in automated rules system that lets you set conditions and actions directly in the panel. Free, native, no integrations needed - the best starting point for publishers just beginning with automation.
What it enables: Pausing, activating, changing budget and bid based on campaign metrics. Rules checked every 30 minutes or every hour.
Limitations: No support for complex conditions (OR, NOT), no machine learning, no cross-platform support. Works only within the Meta ecosystem.
When to choose it: As a starting point, for campaigns running only on Facebook/Instagram, when the budget for SaaS tools is limited.
Pro tip: Set separate rules for the campaign testing phase (first 48-72 hours) and the scaling phase. In the testing phase, thresholds should be wider - the campaign needs data, not intervention.
Revealbot - advanced automation with conditional logic
Revealbot extends Meta's and Google's native rules with complex conditional logic, rule scheduling (e.g. "check only between 6:00 AM and 10:00 PM"), integrations with external data, and notifications via Slack or email.
What it enables: AND/OR conditions, rules based on historical data (e.g. "compare with the previous week"), automated reporting, management of Facebook and Google campaigns from one place.
When to choose it: When you need more complex rules than Meta's native system offers, or when you're running campaigns on Facebook and Google simultaneously.
Pro tip: Set up Slack notifications as the first layer - before enabling automatic actions. For the first 2 weeks, observe which rules would have triggered and whether the decisions would have been correct. Only then enable automatic actions.
Madgicx - AI with machine learning
Madgicx goes a step further than threshold-based rules. ML models analyze historical campaign data and independently recommend (or automatically implement) budget, targeting, and schedule changes. The built-in Autonomous Budget Optimizer recalculates budget allocation in real time.
What it enables: Automatic budget optimization across campaigns, performance prediction, identification of audience segments with the highest conversion potential, automated creative testing.
When to choose it: With higher budgets (minimum 5,000-10,000 PLN per month on campaigns), when the volume of historical data is sufficient for the ML model to learn from.
Pro tip: Madgicx requires a minimum of 2-3 months of historical data for its predictions to be reliable. Don't run it on new ad accounts - not enough data for learning.
How to configure the system step by step
Define profitability thresholds for each campaign. Before setting any rule, you need to know: what is the commission value, what is the maximum acceptable CPL, what ROAS is the target. Without these numbers, rules will be arbitrary.
Start with pausing rules, not scaling rules. Pausing rules protect the budget. Scaling rules increase spend. Start by configuring only pausing - for the first 2 weeks you want to protect the budget, not automatically increase it.
Set up notifications before automatic actions. Every rule should first run as an alert (notify, but don't act) for a week. Check whether the decisions the system would have made were correct. Only then enable automatic actions.
Configure rule time windows. Some rules shouldn't run at night (e.g. budget scaling - better to do this during hours when you can monitor the effects). Others should run 24/7 (pausing campaigns burning through budget).
Set safety caps. Every scaling rule should have a maximum daily budget limit above which the system doesn't scale further without your manual approval. This prevents uncontrolled spend increases.
Set up a weekly system review. Once a week, check the action log - which rules triggered, what decisions the system made, and whether they were correct. Optimize thresholds based on your observations.
Sample configuration for a CPL campaign in the finance niche
The following scenario is hypothetical and serves as an illustration of system configuration. The numbers are for example purposes only.
A publisher is running 5 CPL campaigns, commission per lead: 45 USD, total daily budget: 300 USD.
Configured rules:
Pause ad set if CPL > 36 USD (80% of commission) for 8 hours → Slack alert
Increase ad set budget by 20% if CPL < 20 USD for 3 days with >10 conversions per day
Pause campaign if no conversions for 24 hours with spend >90 USD
Alert: if daily spend exceeds 360 USD (120% of budget)
Result after one month of operation: the system triggered 23 pausing actions (mainly at night and on weekends), 8 scaling actions, and sent 12 alerts. The publisher saved an estimated 4-5 hours per week on manual monitoring and reduced wasted budget by approximately 15%.
Most common mistakes in campaign automation
Thresholds set too aggressively from the start. A system that pauses a campaign after 2 hours without a conversion destroys the learning phases of platform algorithms. Give campaigns time to collect data - at least 48-72 hours before the first automatic intervention.
No safety caps. A scaling rule without an upper budget limit is a risk. A rule malfunction, a condition loop, or a configuration error can lead to the budget being multiplied in a short time. Always set a cap.
Automating bad campaigns. An automated system manages good campaigns efficiently and scales bad ones just as efficiently. Before introducing automation, make sure the campaigns it manages have solid historical data and a positive ROI.
Ignoring action logs. An autonomous manager is not "set it and forget it" forever. Market conditions change, seasonality affects results, new campaigns require different thresholds. A weekly log review is the minimum.
Not testing rules before enabling them. Run every rule in "notification only" mode for a week first. Check whether the system's decisions would have been correct. Only then enable automatic actions.
Summary - configuration checklist
I've defined profitability thresholds for each campaign (commission value, max CPL, target ROAS)
I've configured pausing rules as the first layer - before scaling rules
For the first 2 weeks, rules run in "notification only" mode
Every scaling rule has a maximum daily budget cap
Pausing rules run 24/7; scaling rules run only during monitored hours
I have Slack or email notifications configured for all key actions
I conduct a weekly review of action logs and optimize thresholds
Campaigns covered by automation have at least 30 days of historical data
FAQ
Will campaign automation work on an ad account with no data history?
We don't recommend it. Simple pausing rules (e.g. based on a CPL threshold) can be set up right away, but advanced learning systems like Madgicx require a minimum of 2-3 months of historical data for their predictions to be reliable. On a new account, start with Meta's native rules and give campaigns time to collect data.
Is there a risk that automated rules will increase my budget uncontrollably?
Only if you don't set safety caps. Every scaling rule should have an upper daily budget limit above which the system doesn't act without your manual approval. This is a basic configuration element, not an optional add-on.
Which rule should I start with if I've never automated campaigns before?
Start with pausing rules, not scaling rules. Pausing rules protect your budget from losses, while scaling rules increase spend - a riskier first step. Also run them initially in "notification only" mode for a week before enabling automatic actions.
Do I have to choose just one tool, or can I combine several?
You can combine them, but thoughtfully. A typical path is to start with native Meta Automated Rules, move to Revealbot when you need more complex logic or multi-channel management, and add Madgicx once budgets and historical data are large enough for ML models.
How often should I review how the autonomous system is performing?
At least once a week. An autonomous manager isn't a "set it and forget it" solution forever - market conditions, seasonality, and the performance of new campaigns change, so decision thresholds require regular review and optimization.
Want to know which MyLead campaigns are best suited for autonomous management? Log in to your account and contact your Affiliate Manager - they'll help you choose offers with the right data history and scaling potential that are ideal material for an autonomous campaign manager.
Have any questions? Feel free to reach us through our channels.