Generate and refine ad copy

Create copy from your product

1. 1Choose the productCopywriter loads the saved product context.
2. 2Select the format and goalChoose the platform or output type and the message you need.
3. 3Choose audience and angleUse saved strategy or provide focused context. A saved Custom audience supplies the same audience text used by Image Ads, while Copywriter applies it to the writing rather than visual casting.
4. 4Choose the intensityLeave Go aggressive off for the default on-brand treatment, or turn it on for sharper hooks, stronger emotional contrast, more vivid stakes, and a more direct close.
5. 5Review and generateConfirm the displayed credits, then wait for the result in Copywriter history.

On-brand versus Go aggressive

Go aggressive changes the rhetoric, not the factual foundation. It can push a coherent desire, pain, notion, identification, and cost of inaction harder, but it cannot invent or strengthen product capabilities, outcomes, comparisons, numbers, timeframes, testimonials, guarantees, discounts, deadlines, scarcity, or proof. Audience research describes what readers think and want; it does not prove that the product delivers a claim. Both modes use the same selected product evidence, reference structure, and IVOC combination when those inputs apply.

• On-brand is the default. It keeps emotional intensity and claim strength close to the saved product and brand context.
• Go aggressive uses bolder pattern interrupts, sharper contrast, more explicit stakes, and a firmer close without changing product facts.
• Aggressive mode still builds one coherent selling idea. It does not force unrelated or conflicting audience datapoints into one script.
• Review factual, regulated, and platform-policy-sensitive claims before publishing in either mode.

Clone a transcript structure

Use a transcript when its pacing or persuasion structure is useful. Tartol adapts the same narrative device, beat order, progression, and closing method to your product instead of replacing them with a generic framework. When a newly transcribed spoken source contains multiple detected speakers, the cloned reference and generated script preserve the visible Speaker 1, Speaker 2, and subsequent turn labels; a single detected speaker remains unlabeled. Transcripts containing sung vocals also remain unlabeled so one singer’s changing pitch is not mistaken for several people. It does not add HOOK, PROBLEM, CTA, timestamp, or other visible section labels unless the source already uses them. Reference clones skip unrelated customer-voice seed research, while still loading the product and audience context needed for accurate copy. The reference-text card stays compact: choose Read more to open the source in the full-screen copy reader, where reference details include only length, read time, and Copy, or use Copy text on the card to copy the untouched source. Review every factual and legal claim before publishing.

When customer research is still running

The displayed estimate adds a four-minute research allowance only when the job actually schedules new product, audience, or customer research. When saved research is ready and reused, the estimate shows only the writing allowance of about one minute for every started 1,500 characters requested. Actual timing can vary with the provider and output length. Tartol waits for a terminal research result and reuses that same run if a worker must resume, so leave the job processing and check Copywriter history rather than submitting another generation.

When the IVOC research option appears

The IVOC research system is a controlled production experiment visible only to the designated Tartol super admin. It builds and caches a separate direct-Reddit profile for each product and selected awareness level: Unaware, Problem aware, Solution aware, Product aware, or Most aware. Each profile separates materially different audience Identifications into tabs. The normal Copywriter research system remains the default for every other account.

When IVOC is selected, Generate opens a selector before copy starts. Choose the awareness level, one evidence-backed audience Identification, and at least one atomic Desire, Firm Notion, Shakeable Notion, Identification, Characteristic, or Other truth. You can choose at most one item per category, and empty categories stay empty. If that awareness profile is missing, start its 10-credit research phase in the modal; copy credits are not charged until research is ready and you confirm the message. If a bounded provider attempt fails, the modal shows a readable cause and an explicitly priced Retry IVOC research action. That action charges one new 10-credit research attempt, atomically archives the failed attempt in the audit history, and unlocks only that product-and-awareness profile; page reloads and worker retries never spend again automatically.

Product Intelligence and landing-page fields are seller-page orientation, not customer evidence. They identify the offer and set the factual claim boundary, but their saved audience, pains, desired outcomes, benefits, differentiators, objections, proof, scanner demographics, and wording are not forced into IVOC copy. Query planning uses five provisional situation-based readers across three awareness-specific search zones. Reddit evidence must validate, merge, replace, or reject those hypotheses. IVOC supplies the customer psychology and message; product context keeps the result truthful.

• Every selected comment receives a salvage pass for desires, causal beliefs, trust preferences, objections, alternatives, identity, situation, and vivid wording. Advice and troubleshooting can be strong evidence. A comment is marked No signal after salvage only when it genuinely contains no usable customer truth; no category is forced.
• Comment selection uses evidence-quality signals, upvote percentile within each thread, provisional-reader and source diversity, near-duplicate penalties, and diminishing returns for repeated comments from one thread. Lower-awareness query text rejects later-stage shopping language, but discovered threads receive a soft awareness-fit adjustment instead of a hard keyword exclusion, so one incidental price or review mention cannot erase otherwise useful customer truth. A thread found through several searches retains every matching reader and search-zone path for source planning. Sampling stops after the 800 candidates the profile can use, and raw upvote totals are never compared across communities.
• Tartol selects up to 40 primary and 24 backup threads, requests up to 24 comments per thread inside one 1,000-item scrape, and analyzes up to 640 primary comments in 40-comment batches. It can append up to 160 more retained-overflow or backup comments when evidence remains thin. Independent extraction batches run in bounded five-call waves so the larger profile remains time-bounded.
• Research runs before copy rather than making a writing job wait blindly. Simultaneous requests reuse the same product-and-awareness build. Structured research reserves output space for visible data, and if the one semantic thread plan is empty or unusable, Tartol continues from the already-discovered sources with a diverse local ranking instead of buying a repair call.
• Query planning also remains inside one bounded AI call. If that response has fewer than five provisional readers, missing reader-by-zone searches, duplicate wording, invalid keys, or malformed JSON, Tartol preserves every usable field and deterministically repairs the missing cells into a verified five-reader by three-zone matrix before Reddit discovery begins. The profile audit records whether repair was used and which structural rules initially failed; an imperfect query-plan format no longer discards the entire generation or purchases a second planning call. A profile stopped by the old query-plan error before any Reddit search or scrape can enter this repaired pipeline once when a newly paid matching IVOC job starts; other failed paid research remains locked against automatic retries.
• Distinct desires, notions, identifications, and characteristics remain separate atomic ideas; empty categories are valid. Direct one-off ideas remain visible as emerging signals, while cautious inferred Identifications or Characteristics need repetition before they become ranked summaries.
• On the Brands page, use the IVOC icon on a product row to switch between available awareness profiles and audience tabs, then inspect cross-category evidence-strength rankings, source breadth, exact datapoint quotes, recurring phrases, no-signal raw evidence, and the sampling audit.
• In a generated copy preview, File Details shows a clickable Target Audience field for IVOC output. It records the exact Identification selected for that copy and reopens the complete cross-category rankings for that audience.
• Cloned ad text and video scripts keep the reference structure while using the selected primary IVOC idea and only its compatible supports. Go aggressive can intensify that coherent message, but it cannot add unsupported facts or force unrelated profile items into the script.
• Image Ads and the public REST API or MCP server do not use this experimental system while the comparison is running. Tartol Agent can explain the visible workflow and audit field, but it cannot run, reset, or inspect internal IVOC research.

Refine without losing the original

• Favorite the strongest result so it is easier to find.
• Create variants with one controlled difference.
• Use Edit for a direct instruction such as shortening the hook.
• Use Translate for supported languages and review the output in context.
• Set the best result as the main version for its copy family.