Most founders start paid ads the expensive way: launch first, learn later. They write creative from instinct, set a budget from vibes, and treat the first few months of spend as unavoidable tuition. It is not unavoidable. The tuition has already been paid, publicly, by every competitor in your category, and their receipts are sitting in the Meta Ad Library. Ad intelligence for founders is the discipline of reading those receipts before you spend, and this is the 90 days plan for doing it in the right order.
The structure is deliberate: thirty days of looking, thirty days of borrowing, thirty days of building. Spend money in month one and you are paying to learn things you could have read for free. Skip straight to copying in month two without the mapping and you will copy the wrong things. Do it in sequence and by day 90 you have something most founders never build: your own evidence base.
Days 1-30: Map the field, spend nothing
The first month has one rule: zero ad spend. Everything is observation.
Verify who actually advertises, not who you assume
Every founder carries a mental list of competitors. That list is usually wrong for advertising purposes, because it is built from press, Twitter, and pattern-matching, not from who is actually buying reach. Some of your "biggest competitors" run no paid social at all. Some brand you have never heard of is quietly the category's heaviest advertiser.
So start with verification: search your category in the Meta Ad Library and Google's Ads Transparency Center and list who has live ads today. The active advertisers are your real competitive set for paid, whatever the market map on your pitch deck says.
Ingest the top five rivals' libraries
Pick the five most serious active advertisers and pull their complete ad libraries, not just the first screen of results. You want everything live, per brand, logged with first-seen dates. This is also the moment to start snapshotting: the Ad Library shows you today, and only today. History is something you build by checking daily, and the earlier you start, the more the next sixty days of this plan can lean on it.
Read the longevity distribution
For each rival, look at how long their ads survive. Brands do not keep paying to run losing ads, which makes survival time the honest performance proxy in public data, a principle we unpack fully in the days-running post. The distribution tells you the strategy:
- A few ads alive for months: concentrated spend behind proven winners. Study those specific ads hard.
- Constant churn of short-lived creative: an aggressive testing culture. Study their rate of iteration, not any single ad.
- A stale handful running forever with nothing new: a neglected account, not a strategy to emulate.
Build the hook and format inventory
Now catalog what the survivors have in common. For every long-running ad across your five rivals, log the hook type (problem-first, customer story, founder story, us-versus-them, offer-led), the format (static, video, carousel, UGC-style), the tone, and the offer. Patterns emerge fast: maybe every survivor in your category opens on a customer's words, or every winner is a specific video style, or nobody is running a particular angle at all, which is either a gap or a graveyard.
By day 30 you should be able to answer, from notes rather than memory: who advertises, what survives for them, and what the category's proven structures look like. Total spend so far: zero. If you want the deeper manual playbook for this phase, the cheap competitor analysis guide walks through it tool by tool.
Days 31-60: Test with borrowed conviction
Month two is where money enters, in small amounts, pointed by month one's map.
Model tests on proven formats, not your guesses
Your first campaigns should be structured like the category's survivors: same hook family, same format class, your product and voice. This is not copying an ad; it is borrowing a validated structure. If three rivals' longest-running ads are all short customer-story videos, your first test should probably be a short customer-story video, not the clever concept you dreamed up in the shower. Your instincts get their turn in month three, after the borrowed structures have set a baseline. The line between modeling a structure and ripping off a creative matters, and the clone-and-launch post draws it explicitly.
Set up honest tracking before the first dollar
Before anything goes live: pixel and Conversions API on Meta, Google tag with a real conversion action, one verified test conversion end to end. Broken tracking makes every test you run this month unreadable, and unreadable tests are just spend. Do the boring setup once.
Define kill criteria before launching
Decide, in writing, before launch: how much each test gets to spend, how long it runs, and what result keeps it alive. The specific thresholds depend on your price point and margins, and any number I put here would be an invented benchmark, so I will not. The point is not the exact threshold; the point is that the decision is made before the emotional attachment forms. A test that hits its kill criteria dies, even if you love the creative. A test that survives earns another cycle. Without pre-committed criteria, every founder finds a reason to keep feeding their favorite.
Days 61-90: Build your own evidence
Month three shifts the source of truth from rivals' data to yours.
Scale what survived your kill criteria
Whatever passed month two's filter gets more budget, incrementally, while you watch whether performance holds. Whatever died gets an autopsy: was it the hook, the format, the offer, or the audience? Your longevity data on rivals now has a private counterpart: your own survival record, which beats any competitor signal because it carries real cost and conversion numbers only you can see.
Start the swipe file and the weekly ritual
Two habits start now and never stop:
- A swipe file of category winners. Every long-running ad you find in your niche goes in, with dates and notes on why it likely survives. Six months from now this file is your creative team's brief-writing engine.
- A weekly monitoring ritual. Once a week, check your tracked rivals for changes: new ads launched, old winners retired, offers shifted. Twenty minutes. A rival killing a long-term winner, or suddenly doubling creative output, is strategic information you want within days, not quarters. Alerts make this near-automatic; the competitor alerts workflow covers the setup.
Learn your category's creative lifespan
By day 90, your snapshots cover a full quarter, enough to see roughly how long a winning ad lives in your category before fatigue sets in. Some categories burn through creative in weeks; others run the same ad for a year. Knowing your category's rhythm tells you how fast you need to produce, which is a budgeting fact most founders learn only after their first winner dies unexpectedly.
What NOT to do in the first 90 days
- Do not buy expensive spy-tool subscriptions before you know what to look for. Month one costs nothing but attention. A paid tool amplifies a workflow; it cannot replace knowing which signals matter.
- Do not copy creatives pixel-for-pixel. It is legally risky, brand-damaging, and strategically empty: the surface of an ad is the least of why it works. Borrow structures, never assets.
- Do not chase every rival launch. Competitors test constantly and most of their tests fail. React to what survives, not to what appears. An ad that shipped yesterday is a hypothesis; an ad still alive at day 60 is information.
- Do not skip the tracking week. Everything in months two and three assumes your numbers are real.
The one-page 30/60/90 summary
| Phase | Goal | Actions | Spend |
|---|---|---|---|
| Days 1-30 | Map the field | Verify active advertisers, ingest top 5 rivals, read longevity, build hook/format inventory | Zero |
| Days 31-60 | Test with borrowed conviction | Launch small tests on proven structures, verify tracking, pre-commit kill criteria | Small, capped per test |
| Days 61-90 | Build your own evidence | Scale survivors, autopsy losers, start swipe file, weekly rival check, learn creative lifespan | Scaling winners only |
FAQ
Can I really spend nothing for the first 30 days?
Yes, and you should. The Meta Ad Library and Google's Ads Transparency Center are free, and every signal month one needs (who advertises, what survives, which structures dominate) is public. Founders who skip the mapping month do not skip the learning; they just pay ad platforms to teach them the same lessons slower.
How many competitors should I actually track?
Five serious active advertisers is the sweet spot to start. Fewer than three and you cannot tell category patterns from one brand's quirks; more than seven and the weekly review stops happening because it takes too long. You can always rotate brands in and out as you learn who is actually worth watching.
What if nobody in my niche is running ads?
That is a genuine finding, not a dead end, and it means one of two things: you have discovered an under-advertised channel, or others tried it and the economics failed. Widen the search to adjacent categories that share your customer and read their survivors instead. Then test carefully with your own kill criteria, because you are the category's pioneer either way.
Is it copying to model my ads on competitors' formats?
Modeling a structure is not copying a creative. Every category converges on proven patterns (hook types, formats, offer framings) because those patterns survive contact with real spend, and using them with your own product, footage, and voice is simply learning from the market. Reproducing someone's actual assets or script is a different act entirely, and it is the thing to never do.
Run the whole plan from one chat
Every step above works manually with browser tabs and a spreadsheet, and the plan is worth running even that way. AdWhispr just collapses the mechanical parts. Day one, find_competitors verifies who in your niche is actively advertising right now, which is the entire mapping question answered in seconds. Week one, paste each rival's Facebook URL and AdWhispr ingests their full ad library, snapshots it daily so the longevity distribution is real history, and AI-classifies every ad by hook, format, tone, and offer, which is your inventory built for you. By day 45 you can clone a proven structure for your own brand and launch it on Google Search, PMax, or TikTok from the same conversation. The Free plan needs no credit card and covers the whole mapping phase; Pro at $39/month carries you through testing and scaling. Connect from Claude, ChatGPT, or Cursor at https://adwhispr.com/api/mcp, or run npx adwhispr-mcp-server config.
Stop paying tuition your rivals already paid, start citing your inputs: run your first 90 days with AdWhispr.