We launched a new campaign for a local medspa. Budget: $200/day. Goal: leads under $80. After 72 hours, we had spent $600 and generated exactly two leads. The CPL was $300. The client, looking at the same dashboard we are, sends the inevitable email: “Is this working?”

This is the moment that separates experienced operators from novices. The novice panics, starts swapping out creative, tweaks audiences, or—worst of all—kills the campaign. The experienced operator knows the first few days aren't about performance. They’re about diagnostics.

Your job in the first two weeks isn't to hit your target cost-per-lead. It's to build a foundation for a system that can. It’s about patiently gathering clean data, validating your setup, and finding the first flicker of signal in the noise. Here’s what we actually look at, day by day.

Days 1-3: Is the Plumbing Working?

In the first 72 hours, I care about one thing: data integrity. I am not looking at CPL. I am not looking at conversion rates. Staring at those numbers now is like judging a cake by tasting the raw flour. It's pointless and you’ll come to the wrong conclusion.

During this phase, we are just asking, “Is this thing on?” It's a technical shakedown, not a performance review.

Here’s my checklist:

• Spend & Delivery: Is the campaign actually spending its daily budget? Or is it stuck at $5 spent by midafternoon? If it’s not spending, we might have an issue with the payment method, a low bid cap, or an audience so small it can’t find anyone to serve impressions to. If one ad set is hogging 90% of the budget in an Advantage Campaign Budget (ACB) setup, that’s also a data point, but not one I’d act on yet.
• Impressions & Clicks: Are we getting impressions? Are they translating to link clicks? The numbers will be small, but they need to exist. Zero clicks from 5,000 impressions tells me something is fundamentally wrong with the ad creative or the CTA. Maybe the image is broken or the headline is incomprehensible. It's a check on the most basic function of an ad.
• Pixel & CAPI: Is the tracking firing correctly? This is the most critical check. I use the Events Manager testing tool to submit a test lead. I want to see the Lead event fire in the tool, correctly deduplicating the browser (pixel) and server (CAPI) events. Then, I check our ad reporting. Does the lead show up in the Results column? If not, we have a tracking problem, and every other metric is garbage until it's fixed.
• CRM Integration: For leads passing to a client's CRM via a webhook or Zapier, we confirm the test lead we submitted actually arrived. Did it map to the correct fields? Did it trigger the client's automated follow-up sequence? A $50 lead that gets lost in a CRM black hole is a $50 waste.

At the end of 72 hours, we've hopefully spent a few hundred dollars and maybe have a lead or two. The CPL will almost certainly be horrifying. This is normal. The algorithm is just flailing around, trying to learn. Our only job was to ensure the pipes are connected and data is flowing. No optimizations. No panic.

Days 4-7: Looking for First Signals

We’re entering the second phase. The campaign is still in the “Learning Phase,” and it will be for a while. Patience is still the keyword. We are not making changes unless something is catastrophically broken. However, we can start looking for the first whispers of actual performance.

I’m moving from pure technical diagnostics to directional analysis.

• Engagement Metrics (with a grain of salt): Now I’ll glance at Click-Through Rate (CTR). I prefer CTR (Link) because it shows intent to leave the platform. I'm not optimizing for it, but it's a good health metric. If our CTR is 0.4%, our creative is not resonating with our audience. If it’s 2.2%, we have a good match. This is one of the earliest indicators of whether your core ad hypothesis is sound.
• Funnel Drop-off: We now have enough clicks to check the next step. Let's say we have 150 link clicks but only one lead. It's tempting to blame the ad or the audience. Don't. The problem is almost certainly on the landing page or in the lead form. The ads successfully got 150 interested people to click—their job is done. Your landing page or Instant Form is failing to convert them. A landing page conversion rate (Leads / Landing Page Views) below 5% is a red flag. An Instant Form completion rate below 10% is also a sign of a problem—maybe you're asking for too much information, or your form is confusing.
• CPL Trend, Not Value: The CPL is still not an actionable number, but I do look at its trajectory. If on Day 3 it was $300, on Day 5 it was $220, and on Day 7 it’s $175, that is a fantastic sign. The algorithm is learning. The system is working. If the CPL started at $150 and is now $280, I’m concerned. It suggests the initial conversions were flukes and the algorithm is struggling to find more like them.

By the end of the first week, I still haven't touched anything inside the running campaign. The goal is to let it exit the learning phase with a clean, uninterrupted dataset. The only action I might take is outside the campaign—like telling the client their landing page needs work based on a 2% conversion rate from 200 clicks.

Days 8-14: The First Real Decisions

Welcome to week two. By now, we've hopefully spent enough to get a meaningful number of conversions—ideally close to the 50 conversions Meta recommends to exit the learning phase. Now, and only now, can we start making informed decisions.

This is where we introduce a number-based example. Let’s stick with the medspa client. Goal: leads under $80. Budget: $200/day. By Day 14, we’ve spent roughly $2,800.

Our campaign has three ad sets inside an ACB structure:

• Ad Set 1 (Broad): Targeting women 35-60 in a 15-mile radius. It has spent $1,500 and generated 12 leads. CPL = $125.
• Ad Set 2 (Lookalike 1% - Leads): Based on a past lead list. It has spent $1,000 and generated 20 leads. CPL = $50.
• Ad Set 3 (Interest Stack): Targeting interests like “Luxury Goods” + “Skincare.” It has spent $300 and generated 1 lead. CPL = $300.

A novice looks at this and immediately kills Ad Set 1 and Ad Set 3 and shoves all the budget into Ad Set 2. This is a mistake.

Here’s what a practitioner does:

1. Analyze Lead Quality, Not Just Cost: First, we talk to the client or check the CRM notes. How many of the 20 leads from the Lookalike audience actually booked a consultation? What about the 12 leads from the Broad audience? It's common to find that the cheaper leads from a Lookalike are lower-intent, while the more expensive leads from a Broad audience are from net-new customers who become high-value clients. You might find that the cost-per-booked-appointment from the Broad audience is actually lower. You don't know until you look past the CPL.
2. Make Incremental Changes: Ad Set 3 is not performing. With only 1 lead for $300 in spend, it's safe to pause it. This will automatically redistribute its potential budget to the other two ad sets. I am not going to double the campaign budget because Ad Set 2 is working. That can shock the system and throw everything back into learning. Ad Set 1 is over our target CPL, but it’s generating volume and might have good downstream quality. I’ll leave it running for another week to gather more data.
3. Plan the Next Iteration: We’ve learned that a Lookalike of past leads works well from a cost perspective, and Broad audiences can work but are more expensive. Our next test, likely duplicated into a new campaign to preserve the learning of this one, might involve a new creative angle for the Broad audience or a test of a 3% Lookalike to see if we can get more scale.

Decisions in week two are about pruning the clear losers and validating the winners against business outcomes, not just platform metrics. They are small, deliberate moves that set the stage for long-term scaling.

The first 14 days of a new Meta account are an exercise in restraint. Your primary job is to be a diagnostician, ensuring the technical foundation is flawless. Then, you become an observer, patiently watching for the first signs of life without disturbing the system. Only in the second week do you put on the hat of an optimizer, and even then, your moves should be small, informed by data, and validated against the client's actual business goals. Rushing this process to appease a nervous client or chase an early CPL target is the fastest way to kill a campaign that might have eventually succeeded. Let the data accumulate, validate the plumbing, and trust the process. The time for aggressive optimization will come, but it's not in the first two weeks.