This is a generic preview — prepared for a fictional company, with the specifics redacted. Your copy is written for your company and market.A generic preview for a fictional company — specifics redacted. Yours is written for your market. What's inside →
The structure is real. The words are generic — your copy is written for your company, your market and your competitors. One real element from each of the five parts:
Built from 6,730 real computer software ads on LinkedIn's public Ad Library
prepared by adsignal · human-checked · adsignal.eu
Steps 1–7 are setup, about two hours. Steps 8–11 come after. Two, verbatim:
One decision, thirty seconds. Then stop thinking about it.
Build all 3 ads as a single image, so the only thing you test is the offer and the words.
| Format for all ads | Single image |
|---|---|
| Image spec | 1200 × 627 px. One focal point. Under six words of text. |
| Do not | Do not mix formats for six weeks. Do not run video because it feels modern. |
| Week 7 experiment | One carousel, same offer, same audience, same form. |
Why: 24 of the 60 longest-surviving computer software ads use single image measured — the format your category already paid to learn.
Written against your own best ad, never an outside benchmark.
| Leads | KILL if… | KEEP if… | SCALE if… |
|---|---|---|---|
| 0–9 not readable |
Zero leads after €84 on that ad arithmetic | Everything else. Do not edit copy or narrow the audience. | Nothing — budget here buys noise faster. |
| 10–29 loosely readable |
Over 2× your best ad's cost per lead, or above €48 rule of thumb | Within 2× of your best — it pays rent as the comparison. | At or below €10 and it is your best: +50%. Never double. |
| 30+ act with confidence |
Over 1.5× your best ad's, two weeks running rule of thumb | Within 1.5× — the variant pool for next week. | Best ad at or below €17: +50%/week while it holds. |
Why: only 1.8% of computer software ads reach 180 days on air measured — plan on ~55 attempts before one compounds.
Three slots, one shared audience, one shared form. Each prints its role and settings sheet:
Its job: names the problem your buyer already has, then offers the shortest route out. Highest intent, highest cost per lead — the baseline.
Why this slot: problem-first openers recur across your category's longest survivors measured.
| Objective | Lead generation, native form |
|---|---|
| Ad format | Single image |
| Campaign | Core — Single image (from step 1) |
| Audience | The step-2 audience, identical on all slots |
| Geography | Lorem ipsum geo set for your territory |
| Company size | 11–50, 51–200, 201–500 lorem |
| Seniority | Lorem Manager · Ipsum Director · Dolor VP |
| Audience Expansion | OFF |
| Audience Network | OFF |
| Placement | LinkedIn feed only |
| Bid type | Maximum delivery |
| Daily budget | €12/day |
| Lead-gen form | The 5-field form from step 4 |
| CTA button | Lorem ipsum |
| Tracking URL | https://lorem-ipsum.example/?utm_source=linkedin&utm_medium=paid&utm_campaign=playbook&utm_content=ad1 |
No ad copy is shown here — generic words would tell you nothing about the copy you are buying.
8 complete prompts, pasted into Claude or ChatGPT with your playbook attached. The model reads it first, so what comes back follows your audience, offer and thresholds:
Use it when your audience estimate is under 50,000. The full prompt is printed in your copy.
I am pasting my Ad Playbook, which contains the audience I am currently targeting on LinkedIn. My audience estimate is too small. Propose exactly 3 ways to widen it, ranked by how little they damage lead quality: one adjacent job function, one neighbouring geography, one company-size band. For each, say what kind of person newly enters the audience and what the risk is. Then tell me which single one to do first, and what I should expect to see in the numbers within 14 days if it worked. --- MY PLAYBOOK --- PASTE YOUR PLAYBOOK HERE
41 questions in seven groups, including the uncomfortable ones:
What should the thank-you screen actually do?
One job: give them the next step while intent is still hot. If the offer was a document, deliver the file right there rather than only by email. If the offer was a demo, put your booking link there. One button, never your homepage.
What is a good cost per lead on LinkedIn?
For B2B lead-gen forms, €10–€24 is a healthy reference band and anything under €17 is genuinely good rule of thumb. LinkedIn publishes no cost data, so treat that band as a craft frame, not a measurement. After 30 leads your own median replaces it — and if your deal size is €50,000, a €40 lead may be excellent.
Can I use the same form for all three ads?
You must. Three ads with three different forms cannot be compared — a difference in cost per lead could be the ad or could be the form, and you will never know which. One form, versioned, changed only between test cycles.
This is the one part we don't show.
The follow-up system behind 14,000+ leads and ≈€500,000 of our own LinkedIn spend — what happens after someone fills in the form, step by step, including the versions that failed. It is where most advertisers lose the money they just spent, and publishing it here would hand it to people who never paid.
It arrives in full with your playbook.
Your company, your market, your competitors — in your inbox within 24 hours.
30-day money-back guarantee · what's inside · the free study
Start at the floor, not at what you can afford. The number below is the smallest daily spend that still buys enough impressions for the first read to mean anything in your audience size.
Leave it untouched for the first fourteen days. Every budget change restarts the learning period, and a campaign that is restarted three times has never actually run.
| Daily budget, campaign 1 | set in your copy |
| Bid type | set in your copy |
| Minimum before you judge it | set in your copy |
| Change budget after | set in your copy |
Three numbers decide everything and they are read in this order. Anything outside these three is noise at this spend level and will make you change the wrong thing.
Write the numbers down at the end of the first week and again at the end of the second, before you look at them together. Reading them live, one refresh at a time, is how good campaigns get killed early.
| Look at first | set in your copy |
| Healthy range | set in your copy |
| Fix before touching creative | set in your copy |
| Kill threshold | set in your copy |
Paste your playbook, then this prompt. The model already knows your audience, your offer and your thresholds, so what comes back is written against them rather than against a generic idea of your industry.
Ask for the variants ranked by which claim they lean on, not by which one sounds best. You are choosing an argument, not a sentence.
| Use it when | printed in full |
| Variants returned | printed in full |
| What to change first | printed in full |
Almost always one of four causes, and they are checked in a fixed order because three of them are free to fix and the fourth costs money. The order below is the one that finds it fastest.
If all four check out and impressions are still flat after 72 hours, the cause is the audience size rather than the campaign, and the widening prompt in Part 3 handles it.
| Check 1 | answered in full |
| Check 2 | answered in full |
| Check 3 | answered in full |
| Check 4 | answered in full |
The sample stops here. Your copy keeps going — written for your company, human-checked, in your inbox within 24 hours.