Provide a URL, campaign goal, and target audience. Get five LinkedIn ad copy variants following best practices, labeled for A/B testing in Campaign Manager.
Paste in the client's URL, the campaign goal, and the target audience on LinkedIn. Juma pulls information from the client's site and returns five complete LinkedIn ad copy variants, each built around a different hook angle. Every variant includes intro text, a headline, a CTA, and character counts ready to paste into Campaign Manager.
After reviewing the first set of variants, use this step to replace the weakest angles before running a paid test. The context from your brief carries over automatically, so there is no need to re-enter the client URL, campaign goal, or audience. Ask Juma to swap in data-driven angles, such as customer outcome stats or specific product claims.
This step is most useful when the first round feels directionally correct but two or more variants are too similar to produce differentiated test results. Each new variant replaces a specific hook angle from the original set, so the final five are clearly distinct before any budget goes live.
What you can swap in:
Use this step when the first round is approved and the campaign needs creatives for a different funnel stage. The context from the previous step carries over, so the Flow returns five new variants without re-briefing the client. Name the funnel stage and the Flow rewrites the angles, CTAs, and format recommendations to match.
Awareness and conversion campaigns require fundamentally different copy. Awareness ads address people who do not yet know the brand exists, so they lead with problem recognition. Conversion ads address people already evaluating options, so they lead with proof and a specific next step. The same offer needs different framing at each stage.
What changes between stages:
Once copy is confirmed, generate an ad visual matched to the approved variant. The copy context carries over automatically, so there is no need to describe the brand or audience separately. Describe visual preferences in the prompt and the output applies them alongside the confirmed copy context.
LinkedIn feed images perform best when they are clean and professional rather than high-contrast or consumer-style. The visual should stop the scroll without competing with the copy. Describe the subject, composition, and color palette and Juma generates an image ready to upload directly into Campaign Manager.
What to include in the visual prompt:
LinkedIn ad copy gets written on repeat: new campaigns, new offers, same client voice and audience. A Juma project stores the brand context and performance history so every round of variants starts from what's already known about the client, not from scratch. If the client project already exists, add these items to it.
Add a short description to each knowledge item in the project's info field so Juma knows what each file contains and when to use it. For example:
Each set of LinkedIn ad copy examples includes five variants, each testing a distinct angle: statistic-led hook, pain-point opener, social proof, curiosity gap, and direct benefit statement. The mix adapts to the client's product and audience, and every variant follows LinkedIn ad copy best practices for hook structure.
For a B2B product targeting operations teams, a statistic-led hook opens with a time-saving claim while a pain-point opener leads with the friction that exists before the product is in use. Every variant carries a labeled testing hypothesis so the team knows what it is designed to prove before allocating budget. If the first round does not produce five clearly distinct angles, a follow-up step adds data-driven or customer outcome variants to replace the weakest performers.
LinkedIn truncates intro text after 150 characters on mobile, placing a "see more" break before the reader reaches the rest of the ad. The first two sentences carry the full weight of the hook. Copy that does not earn attention before that cut will not earn the click.
This constraint separates LinkedIn from other ad platforms. On Meta, a strong visual compensates for a slow-opening text block because the image loads before the copy. On Google Search, intent already exists and the user is actively looking for what the ad offers. LinkedIn ads appear in a passive scroll feed populated by organic content, so they compete for attention rather than intent, and the 150-character break is the primary performance variable in the intro text.
LinkedIn intro text supports up to 600 characters. Headlines are capped at 70 characters. The Flow writes within these linkedin ad copy character limits by default and includes per-element character counts in every variant so teams can verify limits before uploading to Campaign Manager.
Character limits on LinkedIn function differently from other platforms. The 600-character intro limit is generous enough for two or three substantive sentences, but the 150-character mobile truncation means only the first portion displays without expansion. The 70-character headline sits below the visual and carries less weight than the intro in scroll environments. The Flow accounts for all three constraints by default, so every variant is platform-ready on delivery.
Providing the client URL aligns variants with the brand's actual positioning, product language, and audience signals. Variants reflect what the client actually says and sells, not just what you brief in. A brief with no URL produces more generic output that requires more revision rounds to match the brand voice.
The site research covers product descriptions, value propositions, and outcome claims the brand makes publicly. A statistic-led hook draws on a claim already on the homepage rather than inventing a figure. A pain-point opener uses the exact friction language the client uses in their own messaging. This alignment reduces revision rounds before approval and makes variants consistent with other brand touchpoints the audience has already encountered.
The run-first recommendation identifies which two to three variants to prioritize in Campaign Manager's first test, based on the campaign goal and audience segment. Running all five variants simultaneously splits budget too thin to produce a clear signal. The recommendation concentrates spend where data is most likely to confirm a winner quickly.
The recommendation accounts for both the testing hypothesis and the audience type. Variants built for cold audiences receive different prioritization than those built for retargeting segments. After the first test concludes, the remaining variants rotate in for the second test. This structured sequencing accumulates learning across rounds rather than producing inconclusive data from a single over-split test.
