How to get your first ads-agency clients with teardowns
You do not need a fake logo wall to land client one. You need a narrow niche, a list of brands already running ads, a specific public teardown, and a paid next step.
This post is the client-acquisition playbook: ICP, list build, teardown craft, outreach, paid audit, pilot, referrals. The start an AI ads agency guide owns the operating model (niche → offer → delivery). If you are still deciding what to sell and how to deliver it, read that first. If the shop model already exists and you need clients #1–5, stay here.
Cluster chooser: Packaging retainers and delivery → sibling Post 1. What an AI media buyer is → that hub. Fee math vs % of spend → flat fee Google Ads. This URL = first clients via teardowns and outreach.
Why first clients stall (and what replaces spray-and-pray)
Most new shops stall because they pitch services to people who are not spending, or they pitch with empty proof. Replace volume spam with public-signal targeting and a teardown that names one real observation.
Spray lists of “marketing managers” produce polite silence. Brands with live ads, a clear offer page, and a reachable buyer are a different game: you can point at something checkable before you ask for time.
Warm network still matters for client #1–3
Export LinkedIn connections. Ask ex-coworkers, alumni, and past clients for intros or a 15-minute “what’s broken in X” chat. Warm paths often close the first retainer before cold does — then cold becomes insurance, not the whole pipeline.
You are not “cheating” by starting warm. You are buying learning cycles on real conversations while the cold motion matures. Ask for introductions to people already spending, not to “anyone who might need marketing.” The same spend filter that makes cold lists work also makes warm intros convert.
Cold works when the opener is checkable
Cold email and LinkedIn reply when line one references a live ad, landing friction, or a tracking smell — not “I help brands scale with AI.”
The buyer should be able to open a tab and verify your observation in under a minute. That single constraint kills most generic agency templates.
Step 1 — Sharpen who you sell to (before you write one email)
One vertical, one buyer role, one spend signal. A narrow ICP makes teardowns sharper and outreach shorter. Niche definition lives in the operating-model post; do not re-derive the whole model here. Lock the worksheet below so every teardown and email points at the same buyer.
ICP worksheet (one sentence each)
- Vertical: Who you serve in one clause (e.g. local service businesses with booking calendars; DTC brands already on Meta who want a second network).
- Buyer: Founder, marketing lead, or owner-operator — name the role that can approve a paid audit.
- Spend signal: Active ads in Ad Library, clear paid landing presence, or other public proof they already buy media.
- 30-day offer: One network you can deliver end-to-end in a pilot without inventing capacity you do not have.
If you cannot fill those four lines, stop drafting emails. Fix the ICP first.
Step 2 — Build a tight list of brands that already run ads
Build 40–60 accounts from public sources, not a 5,000-row spray list. Prefer brands with two or more active ads and a clear offer page.
Quality beats quantity here. Forty named accounts with notes beat five hundred rows you will never open.
Public sources (no paid-tool requirement)
- Meta Ad Library — active ads, creative count, landing destination.
- Prospect site — offer, booking, or checkout path; headline vs ad claim.
- Optional LinkedIn — decision-maker title plus recent posting activity (useful for the outreach step, not required for the teardown itself).
You can add Google presence and other public creatives when your niche lives there. Start with whatever library actually shows spend in your vertical.
Qualification filters
Keep a row only if most of these are true:
- Already advertising (budget signal).
- Small-to-mid (reachable buyer, not enterprise procurement theater).
- Visible creative or funnel issue you can name without account access.
- Fit your delivery niche and the one network you offer in the first 30 days.
Drop anyone who fails the spend signal. “Might advertise someday” is not a prospect for teardown outreach.
Capture a few columns as you go: brand, URL, Ad Library link, buyer name if known, one candidate observation, and status (teardown / sent / replied / audit / dead). A spreadsheet is enough. Fancy CRM comes later.
Step 3 — The public teardown (your outbound proof)
A teardown is a short, specific write-up or Loom of what you can see in public — creative patterns, funnel mismatch, obvious tracking gaps — plus two tests you would run. Stick to public signals you can verify.
Loom here means a short screen-share walkthrough of their public ads and landing page — a format buyers already know. It is not a product pitch.
Teardown template
Fill these six fields. Keep the whole piece skimmable in under five minutes of reading (or a few minutes of video).
- Brand + offer in one line.
- Where you looked (Ad Library, site, or both).
- One observation — creative fatigue, hook–offer mismatch, landing friction, or a pixel / attribution smell you can see without login.
- Why it matters in plain language (what a buyer would feel: wasted spend, soft conversion, unclear tracking).
- Two concrete next tests you would run if hired.
- Soft ask — paid audit, or a 15-minute review of the written notes.
Ship the template as a short doc, a one-pager, or a Loom with the same spine. Specificity is the product; length is not.
Free teardown vs paid audit (do not conflate)
Free teardown = outbound hook: time-boxed, public signals only, one observation plus two tests. Paid audit = first SKU: deeper structure, priorities, and a retainer roadmap. Free work (running campaigns for free) is a different — usually worse — motion. Charge for diagnosis depth; give away only enough specificity to prove you can see.
What good specificity looks like
Name the ad angle, the landing headline versus the ad claim, or the form friction. Vague “your Meta could be better” emails die in the archive. If a stranger cannot verify the observation in a public tab, rewrite it before you hit send.
Examples of the right grain (illustrative patterns, not claims about a named brand):
- “Your Ad Library shows three nearly identical UGC ads with the same hook; the landing headline promises a different outcome than the ad claim.”
- “Checkout asks for phone before price — that friction is visible without account access.”
- “The only active creative is a static that has been live long enough that creative fatigue is a fair hypothesis to test.”
Wrong grain: “I can 3× your pipeline with AI.” That line cannot be checked, so it cannot open a cold conversation.
Step 4 — Outreach that earns a reply
Short message, one observation, one ask. Email plus LinkedIn support beats email-only for many buyers. Follow up with new value, not “bumping this.”
Email spine (structure, not invented reply rates)
- Subject — human and specific (brand + one cue), not “Quick question” spam.
- Line 1 — checkable observation.
- Line 2 — implication (why a busy buyer should care).
- Line 3 — what you are offering (teardown attached, or paid audit SKU).
- One soft CTA — reply, book a short review, or accept the audit invoice.
Keep it under a screen on mobile. Attach or link the teardown; do not paste a pitch deck into the body.
A workable shape (adapt, do not copy-paste as a “proven” script):
Subject: [Brand] Ad Library note
Saw [one checkable observation]. That usually means [implication].
Wrote a short teardown with two tests I’d run — happy to send the paid audit scope if useful.
Either way, no hard sell; reply “send it” if you want the notes.
The point of the spine is discipline: observation first, ask last, one CTA.
LinkedIn variant
Comment or connection note that references the same observation. Send the teardown after they accept. Do not paste a pitch into the connection request. Same observation across channels feels intentional; two different pitches feel automated.
Sequence length (operator default)
Plan roughly four to seven touches over two to three weeks across email and LinkedIn. Stop on reply or a hard no. Change the value in each follow-up (second observation, a competitor creative pattern, a one-line test idea) — do not send empty bumps. No open-rate benchmarks here; run your own notes after a few dozen sends.
Step 5 — Productize the paid audit (first money)
Price a time-boxed audit so tire-kickers self-select. Deliver a written doc they can use even if they never hire you — then the call is about implementation, not proving you did homework.
Pick a fixed fee and a fixed turnaround (for example, a few business days after access or screenshare). Publish the deliverable list on the invoice. You do not need a market-wide “standard audit price” — you need a number that covers your hours and filters people who only wanted free consulting. If the buyer pushes for free forever strategy, hold the line: the free teardown already did the hook; depth is the SKU.
Scope boundaries
- One primary network for the first audit (often Meta or Google — whichever your niche already uses).
- Inputs you need versus what you can do from public signals plus a screenshare.
- Deliverable: prioritized findings plus a 30-day test plan.
- Non-goals: no live spend changes in the audit SKU unless separately contracted.
Write the scope into the invoice. Ambiguous audits become unpaid consulting with a new name.
Audit → pilot conversion
A pilot is one network, a budget range, weekly notes, and a clear end date. Win process trust before multi-network retainers. If they already have campaigns running, importing existing campaigns can shorten the ramp — still keep the pilot scoped. Approvals before live spend are hygiene, not hesitation; see the AI ad approval workflow when you productize that step.
Step 6 — From pilots to retainers (and when delivery breaks)
Retainers stick on boring reliability: access hygiene, approvals, spend caps, reporting cadence. When client count grows, delivery — not sales copy — is the bottleneck.
Write the retainer like an operator, not a slogan: which network(s), who owns creative, how spend caps are set, when you report, and what happens when something underperforms. Clients stay when those answers are dull and consistent.
Ask for referrals after a clean pilot close, not after the first discovery call. A short “who else in your circle runs paid on this network?” lands better when they have already seen weekly notes and spend discipline.
Soft product note (Growth / Scale)
Lazy Ads is a fixed monthly fee for AI media buying on the client’s own accounts — not a cut of ad spend. When you are running multiple client accounts, compare Growth ($199/$249) and Scale ($399/$499) annual/monthly on /pricing. Networks available today: Meta, Google, TikTok, LinkedIn, Reddit, Apple, Bing, ChatGPT, Snapchat. For what the buyer layer is (and is not), see AI media buyer; for fee framing vs % of spend, see flat fee Google Ads. Multi-network delivery detail lives on multi-platform AI ad manager when that is the constraint.
A 60–90 day motion (checklist, not a miracle timeline)
Treat this as a work calendar, not a guaranteed client count.
| Window | Focus |
|---|---|
| Days 1–20 | ICP worksheet + 40–60 prospect list + five public teardowns |
| Days 21–45 | Outbound sequence + sell paid audits |
| Days 46–90 | Pilots live + ask for referrals from clean closes |
Adjust the calendar to your capacity. The sequence matters more than the day numbers: list before spam, teardown before pitch, paid audit before free forever strategy, pilot before multi-network retainer.
FAQ
Do I need case studies before outreach?
No. You need specific public observations and a clear paid next step. Case studies come after pilots — with client permission, and only with real numbers.
Should the first teardown be free?
Often yes as a short hook; the deeper work should be a paid audit. Do not confuse free diagnosis with free campaign management.
Which network should I pitch first?
The one your niche already uses and you can deliver. Expanding networks is a retainer conversation, not day-one spam. If your niche is Google-heavy, Google Ads automation is the adjacent read for delivery shape — still keep first outreach on one network.
How is this different from starting an AI ads agency?
Sibling Post 1 (start an AI ads agency) = operating model. This post = how you land the first engagements once that model exists.
Close
Pick ten prospects this week, write three teardowns, send the sequence. When delivery for more than one client is the constraint, look at Growth or Scale on pricing — not another empty pitch template.