Marketing

Segmentation ─► Targeting ─► Positioning ─► Value proposition ─► Messaging ─► Channels ─► Assets
who exists     who first     vs. what       outcome + proof      by funnel    where they   ads, pages,
                             alternative                         stage        already are  emails

Feature:  "Bank feeds sync every 15 minutes."
Benefit:  "See today's cash position before your 9 a.m. call."
Proof:    "Syncs with 40 banks every 15 minutes."

┌───────────────────────────────────────────────┐
│ HEADLINE   the benefit, matching the ad/query  │  above the fold
│ SUBHEAD    who it's for + how, one sentence    │  (on mobile too)
│ [ PRIMARY CTA ]   risk reversal under it       │
│ hero image/video showing the product or result │
├───────────────────────────────────────────────┤
│ social proof: logos, a number, one testimonial │
│ 3 benefits, each with a feature as proof       │
│ how it works: 3 steps                          │
│ objections: price, switching, security (FAQ)   │
│ deeper proof: case study, reviews              │
│ [ REPEAT CTA ]                                 │
└───────────────────────────────────────────────┘

CAMPAIGN BRIEF
Objective         one measurable goal + date (e.g. 200 trials by 30 Jan)
Audience          segment, awareness level, what they believe now
Insight           the truth about them the campaign is built on
Single-minded     the one thing they should take away
  proposition
Reasons to        2-3 proofs we actually have
  believe
Offer + CTA       what we're asking them to do, and why now
Channels          where, with rough budget split
Mandatories       legal lines, brand rules, disclosures
Budget, timeline  total, flighting, key dates
KPIs              primary metric, guardrails, how attributed
Owner, approvers  who decides

<context>
I'm launching [launch, e.g. a new reporting feature in our bookkeeping app] and need a launch campaign: a messaging matrix, a channel plan, a campaign brief, then the assets.
Launch size: [launch tier: major launch | feature launch | changelog entry and email]. I am not a marketer.
</context>
 
<task>
Interview me first: ask the questions a product marketer asks before planning a launch.
</task>
 
<requirements>
- At most 10 questions, grouped under: the product and offer, segments and targeting (STP), jobs to be done and customer language, proof, goal and metric, budget and timeline, current funnel numbers (AARRR), constraints.
- For each question: why it matters to the campaign, and 2-3 typical answers I can pick from.
- Ask what proof I have (numbers, customers, quotes) and where each comes from.
- Do not write messages, plans or copy yet.
</requirements>
 
<process>
Ask the questions, then stop and wait for my answers.
</process>

<materials>
[paste the interview questions and my answers]
</materials>
 
<task>
Propose 2-3 materially different launch strategies for this product.
</task>
 
<requirements>
- For each: the target segment and its awareness level, why this segment first (STP), the single-minded proposition, the 2 lead channels and why, the split between demand capture and demand generation, the primary metric, and the main risk.
- Mark any proposition that relies on proof I haven't given [NEEDS PROOF].
- Recommend one and say what result in the first 2 weeks would make you switch.
</requirements>
 
<process>
Give the strategies as a table, then the recommendation, then stop and wait for my choice.
</process>

<materials>
Interview answers and the strategy I chose: [paste]
Customer language: [paste interview quotes, reviews or support tickets, verbatim]
Current numbers: [paste, e.g. 10k visits a month, 2% sign-up, 15% trial-to-paid]
</materials>
 
<context>
Product: [what it is and its price, e.g. Ledgerly Reports, included in the $49/month plan]
Audience: [segment, e.g. owners of US bookkeeping firms with 2-10 staff] · awareness: [awareness: unaware | problem-aware | solution-aware | product-aware | most aware]
Target: [number and date, e.g. 200 trials by 30 January]
Budget and timeline: [amount and dates, e.g. $3,000 over 6 weeks, launch on 15 January]
Proof we can use: [paste numbers, customers and quotes; real ones only]
The plan is for [team, e.g. me, a part-time designer and a contractor who runs our ads].
</context>
 
<examples>
[optional: 1-2 launch emails, ads or pages you like and why; 1 you don't, and why]
</examples>
 
<task>
A launch campaign whose only goal is [goal: trials | demos | upgrades | purchases]: a messaging matrix, a channel plan, a one-page campaign brief, then first drafts of the core assets, so the team can execute from one document.
</task>
 
<requirements>
- Messaging matrix: per segment, the job to be done, top pain, awareness level, the single message, 3 supporting points (benefit first, feature as proof), the objection to pre-empt, and the CTA.
- Channel plan: the 2-3 channels that fit where this audience already is, labeled owned, earned or paid, each with budget, a weekly metric, and the result that would make us stop it.
- Campaign brief fields: objective, audience, insight, single-minded proposition, reasons to believe, offer and CTA, channels, mandatories, budget and timeline, KPIs (primary plus guardrails), attribution plan, owner and approvers.
- Assets: a landing page outline with message match to each ad and email, 3 paid ads (one PAS, one AIDA, one Before-After-Bridge), a 3-email launch sequence with subject lines, and a UTM naming convention.
- One audience, one message and one action per asset.
- Voice: [traits, e.g. plain-spoken, warm, no hype]. Mandatories: [paste legal lines, disclosures and brand rules].
- Never invent numbers, names, quotes or sources: write [NEEDS PROOF] where a claim isn't backed by the proof above.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every asset carries the single-minded proposition and the same CTA.
- Each landing headline repeats the promise of the ad or email that sends traffic to it (message match).
- Every number and claim traces to my materials or is marked [NEEDS PROOF].
- The KPIs can be measured with what we have, and the attribution caveats are stated.
- The plan fits the budget and the team: nothing needs a channel we can't run.
</quality_bar>
 
<output_format>
1. Messaging matrix (table). 2. Channel plan (table: channel | owned, earned or paid | budget | weekly metric | stop rule). 3. Campaign brief (one page, the fields above). 4. Assets, each labeled with its channel and funnel stage. 5. A week-by-week timeline. 6. The assumptions.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
Write the messaging matrix first and check it against the quality bar, then build the rest on it.
</process>

<materials>
The brief: [paste the stage 3 prompt]
The campaign: [paste]
</materials>
 
<context>
This is a fresh chat. The campaign below was written in another conversation; review it as someone seeing it for the first time, before anything ships.
</context>
 
<task>
A review of this launch campaign. Do not rewrite it.
</task>
 
<requirements>
- Judge against: specific audience, one clear message, benefit-led and feature-proved, awareness fit, honest claims, clear CTA, channel fit, measurable, testable.
- Read each asset as a skeptical member of the audience seeing it for 3 seconds on a phone: what they understand, what they doubt, whether they would click.
- Check message match between every ad or email and its landing page.
- Highlight every claim and say whether the brief's proof supports it; recheck all arithmetic.
- Quote the line behind every finding.
</requirements>
 
<quality_bar>
Before answering, check that every finding names a criterion, quotes the line, and proposes a specific fix.
</quality_bar>
 
<output_format>
A table: rank | finding | criterion | asset and quoted line | fix, most damaging first. Then one line: ship, fix first, or rethink the strategy.
</output_format>
 
<process>
Rank the issues by likely effect on the goal; do not produce revised assets.
</process>

Here are [source: interview notes | reviews | support tickets | sales-call notes]: [paste]. Extract the exact phrases customers use for the problem, the outcome they want, their objections and the alternatives they name. Group and count them, quoting verbatim.
Then suggest 5 headlines in their words, not ours. Never paraphrase a quote.

<materials>
[paste interview notes, reviews, support tickets or sales-call notes, each labeled with its source and date]
</materials>
 
<context>
Product: [what it is, e.g. bookkeeping software for small firms]
Segment: [segment, e.g. owners of firms with 2-10 staff]
What I'll use it for: [use, e.g. landing page headlines and ad hooks for a launch]
</context>
 
<task>
A customer-language swipe file: the words customers actually use, organized so I can write copy from it.
</task>
 
<requirements>
- Extract verbatim phrases for: the problem, the moment it hurts (the "when..." of a job to be done), the outcome they want, their objections, and the alternatives they name.
- Group similar phrases, count how often each appears, and keep the source label on every quote.
- Write 3-5 job statements in "When I..., I want to..., so I can..." form, each backed by quotes.
- Suggest 5 headlines that use their words, and say which quote each comes from.
- Never invent numbers, names, quotes or sources: quote only what is in the materials, and write [NEEDS SOURCE] where a theme has too few quotes to trust.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every quote appears word for word in the materials.
- Counts are counts, not impressions, and single mentions are labeled as such.
- The headlines make sense to someone who has never heard of us.
</quality_bar>
 
<output_format>
A table per group (phrase | count | sources), the job statements, then the headlines with their source quotes.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Write a landing page for visitors arriving from [paste the exact ad, email or keyword], who are [segment, e.g. owners of 2-10 person bookkeeping firms] and [awareness: problem-aware | solution-aware | product-aware]. Its only goal: [action, e.g. start a free trial].
The headline repeats the source's promise; under 500 words; mark missing proof [NEEDS PROOF].

<context>
Product: [what it is and its price, e.g. bookkeeping software, $49/month per firm, 30-day free trial]
Visitors arrive from: [paste the exact ad, email or keyword]
Audience: [segment, e.g. owners of US bookkeeping firms with 2-10 staff] · awareness: [awareness: unaware | problem-aware | solution-aware | product-aware | most aware]
Proof we can use: [paste numbers, customers and quotes; real ones only]
Objections we hear: [objections, e.g. migration pain, price, data security]
</context>
 
<examples>
[optional: a landing page you like and why, e.g. "one button, the proof right under the headline"; one you don't, and why]
</examples>
 
<task>
A landing page whose only goal is [action, e.g. starting a free trial].
</task>
 
<requirements>
- Message match: the headline repeats the promise of the ad, email or keyword above, close to its words.
- Structure: headline (the benefit); subhead (who it's for and how, one sentence); primary CTA with the risk reversal under it; social proof; 3 benefits, each proved by a feature; how it works in 3 steps; an FAQ that answers the objections; the CTA repeated.
- For unaware or problem-aware visitors, lead with the problem; for product-aware or most aware visitors, lead with the offer.
- Under 500 words; no navigation or secondary offers.
- Voice: [traits, e.g. plain-spoken, warm, no hype].
- Never invent proof: no numbers, testimonials, logos or quotes beyond those above; write [NEEDS PROOF] where proof is missing.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The headline-only test: someone who reads only the headline and the button knows what they get and what happens next.
- Every claim is backed by the proof above or marked [NEEDS PROOF].
- One CTA, repeated, that says what happens after the click.
- Every listed objection is answered.
</quality_bar>
 
<output_format>
The page as labeled sections in order (headline, subhead, CTA and so on), with the word count at the end.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
Then give 3 headline angles with their trade-offs, and stop for my choice. After I choose, write the page, then the assumptions.
</process>

Write 6 [channel: paid search | paid social] ads for [segment, e.g. owners of small bookkeeping firms] who are [awareness: problem-aware | solution-aware | product-aware]: 3 angles, 2 ads each, within these character limits: [paste the platform's current limits]. Use only claims from this proof: [paste].
End with which 2 to test first and why.

<materials>
Proof we can use: [paste real numbers, customers, quotes]
Platform limits: [paste the current character limits for each field from the platform's help pages]
Keywords, for paid search: [paste]
</materials>
 
<context>
Product and offer: [what it is, e.g. bookkeeping software, $49/month, 30-day free trial]
Channel: [channel: paid search | paid social | retargeting]
Audience: [segment, e.g. owners of US bookkeeping firms with 1-10 staff] · awareness: [awareness: unaware | problem-aware | solution-aware | product-aware | most aware]
Landing page headline: [paste it, so the ads can match it]
</context>
 
<examples>
[optional: 1-2 ads you like and why, e.g. "names the pain in the first five words"; 1 you don't, e.g. "Revolutionize your workflow!"]
</examples>
 
<task>
A set of ad variants whose only goal is [goal: trials | demos | purchases | sign-ups], built as a test: distinct angles, each traceable to the proof, ready to upload.
</task>
 
<requirements>
- 3 angles, 2 ads each; within each angle, one ad uses PAS and the other Before-After-Bridge.
- Paid search: group the keywords by search intent (learn, compare, buy), drop the informational ones and say why, repeat the searcher's words in headlines, and add 10 negative keywords.
- Paid social: a hook for the first line or the first 3 seconds (under 12 words), and an image or video idea per ad.
- Every field within the limits above; count the characters.
- No exclamation marks and no hype words (revolutionary, seamless, game-changing).
- Never invent proof: write [NEEDS PROOF] where a claim goes beyond the proof above.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The two ads within an angle differ in framework; the three angles differ in promise.
- Each ad matches the landing page headline closely enough that a visitor knows they're in the right place.
- Every character count is within its limit.
</quality_bar>
 
<output_format>
A table: angle | framework | each field with its character count | CTA | creative idea. Then the 2 ads to test first and why.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Write a [sequence: welcome | nurture | trial conversion | win-back] sequence of [number, e.g. 5] emails for [segment, e.g. trial users who haven't connected a bank]. Goal: [action, e.g. connect a bank account].
Per email: day, its one job, 3 subject lines under 50 characters, a body under 150 words, one CTA. Include an exit condition, and no fake urgency.

<materials>
[paste the proof we can use, any current emails, and the product's onboarding steps]
</materials>
 
<context>
Sequence: [sequence: welcome | nurture | trial conversion | win-back | launch]
Who gets it: [segment and trigger, e.g. firms that started a trial and haven't connected a bank within 48 hours]
Sender: [sender, e.g. the cofounder by name, plain-text style]
</context>
 
<examples>
[optional: 1-2 emails you were glad to receive and why; 1 you deleted unread, and why]
</examples>
 
<task>
An email sequence whose only goal is [action, e.g. connecting a bank account, then converting to paid], with timing and branching, ready to load into our email tool.
</task>
 
<requirements>
- Per email: the day sent, the one job of the email, 3 subject lines under 50 characters, preview text, a body under 150 words, one CTA.
- Branching: what happens when someone completes the goal, and an exit condition so nobody gets the rest after converting or unsubscribing.
- Plain and specific, no hype, and only real deadlines.
- Compliance: a visible unsubscribe link and a postal address in every email (CAN-SPAM); one-click unsubscribe if we are a bulk sender.
- Voice: [traits, e.g. plain-spoken, warm, no hype].
- Never invent numbers, names, quotes or sources: write [NEEDS PROOF] where support is missing.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Each email can be summarized in one sentence, and no two have the same job.
- Every subject line says something specific; none is clickbait.
- The sequence stops the moment the goal is reached.
</quality_bar>
 
<output_format>
A summary table (email | day | job | CTA | exit rule), then each email in full, then what to A/B test first.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Turn these questions prospects ask on sales calls into an SEO content plan: [paste]. Per question: target query, search intent, funnel stage, the page type that tends to rank, a working title, and the page it links to.
Don't invent search volumes; I'll check them in an SEO tool.

<materials>
[paste questions from sales calls, support tickets or community threads, verbatim]
[optional: paste keyword data exported from your SEO tool]
</materials>
 
<context>
Product: [what it is, e.g. bookkeeping software for small firms]
Audience: [segment, e.g. owners of firms with 2-10 staff]
What the content should drive: [goal: trials | demo requests | newsletter sign-ups | awareness]
Existing pages: [paste the URLs and titles of pages we already have]
Capacity: [capacity, e.g. 2 articles a month, written by me]
</context>
 
<task>
An SEO content plan built from real buyer questions, ordered so I write the pages most likely to bring in buyers first.
</task>
 
<requirements>
- Per piece: target query, search intent (learn, compare, buy, navigate), funnel stage (TOFU, MOFU, BOFU), the page type that tends to rank for it (guide, comparison, template, tool), a working title, and the existing page it should link to.
- Group the pieces into topic clusters, each with a pillar page.
- Follow Google's guidance on helpful, people-first content: each piece must answer the question better than what ranks now, from our own experience.
- Never invent search volumes, difficulty scores or rankings: write [VERIFY IN SEO TOOL] where a number would go.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every piece traces to a question in the materials.
- Intent matches page type: no product page for a "how to" query, no guide for a "pricing" query.
- The first 3 pieces are the closest to a buying decision, not the broadest topics.
</quality_bar>
 
<output_format>
A table (query | intent | stage | page type | title | links to | source question), the clusters as a short list, then the order to write them in.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Page: [page, e.g. the trial sign-up page]. Weekly visitors: [visitors, e.g. 2,000]. Current conversion: [rate, e.g. 3%]. Change: [change, e.g. a headline that matches the ad]. Estimate visitors per variant for 80% power and 5% significance to detect [effect, e.g. 3% to 3.6%], showing the formula and each step.
Then how many weeks that takes and, if it's too long, a bigger change to test instead.

<context>
Page or email: [page, e.g. the trial sign-up page]
Weekly visitors: [visitors, e.g. 2,000]
Current conversion: [rate and definition, e.g. 3% of visitors start a trial]
Change to test: [change, e.g. a new headline that matches the ad]
Primary metric: [metric: sign-up rate | trial starts | purchases | revenue per visitor]
Guardrail: [guardrail, e.g. trial-to-paid rate must not fall]
</context>
 
<task>
An A/B test plan with an honest sample size, so I know before starting whether the test can give an answer.
</task>
 
<requirements>
- Sample size per variant for 80% power and 5% significance, using Lehr's rule of thumb (16 × p × (1 − p) ÷ δ², where δ is the absolute change), showing each arithmetic step; then the duration in whole weeks.
- If it takes more than about 6 weeks, say so and propose a bigger change to test instead.
- One change, one primary metric, one guardrail; the duration fixed in advance; no stopping early when it first looks significant.
- How traffic is split, and exactly what counts as a conversion.
- Never invent numbers, names, quotes or sources: write [VERIFY] beside the sample size so I check it in a calculator such as Evan Miller's.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The arithmetic is right: recompute it a second way.
- The effect is stated both as an absolute change and as a relative lift.
- The plan says which decision each outcome leads to, including "no difference".
</quality_bar>
 
<output_format>
Hypothesis, metric and guardrail, sample size with the working, duration, decision rules. Under 300 words.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Review this [asset: ad | landing page | email | social post] for [segment, e.g. owners of small bookkeeping firms]: read it as someone seeing it for 3 seconds on a phone, then highlight every claim and say whether this proof supports it: [paste proof]. Rank the fixes; don't rewrite.
[paste the copy]

<materials>
The copy: [paste]
Proof we have: [paste]
Where traffic comes from: [paste the ad, email or keyword that sends people to it, if any]
</materials>
 
<context>
Asset: [asset: paid ad | landing page | email | social post | pricing page]
Audience: [segment, e.g. owners of US bookkeeping firms with 2-10 staff] · awareness: [awareness: unaware | problem-aware | solution-aware | product-aware | most aware]
Goal: [action, e.g. start a free trial]
</context>
 
<examples>
[optional: paste a past asset that performed well, with its numbers, as a benchmark]
</examples>
 
<task>
A review of this copy before it ships. Do not rewrite it.
</task>
 
<requirements>
- Judge against: specific audience, one clear message, benefit-led and feature-proved, awareness fit, honest claims, clear CTA, channel fit, measurable, testable.
- Read it as a skeptical member of the audience seeing it for 3 seconds on a phone: what they understand, what they doubt, whether they would click.
- Highlight every claim and say whether the proof above supports it.
- The competitor test: list the lines that would still work with a competitor's name.
- Check message match with the traffic source.
- Never invent numbers, names, quotes or sources; mark every claim the proof doesn't support [NEEDS PROOF].
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every finding quotes the line and names the criterion.
- Fixes are ranked by likely effect on the goal, not by ease.
- Nothing in the review is a rewrite in disguise.
</quality_bar>
 
<output_format>
A table: rank | quoted line | criterion | problem | fix direction. Then one line: ship, fix first, or rethink.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Campaign results by channel: [paste spend, impressions, clicks, sign-ups, customers, revenue]. Our own database shows [count, e.g. 41] new customers. Compute CTR, CPC, CVR, CPA and ROAS per channel, showing the arithmetic.
Then: the biggest drop-off, which numbers attribution may inflate, and 3 changes to test next.

<materials>
Platform-reported results by channel: [paste a table: spend, impressions, clicks, sign-ups, customers, revenue]
New customers in our own database: [paste the count and any source data]
Funnel numbers: [paste visitors, sign-ups, activated, paid, still paying at month 3]
</materials>
 
<context>
Campaign: [campaign, e.g. a 6-week launch, $3,000 across paid search and paid social]
Attribution model in each platform: [model: last click | data-driven | platform default | don't know]
Decision this feeds: [decision, e.g. where to put next month's $2,000]
</context>
 
<task>
A diagnosis of these results: what worked, where the funnel leaks, how far to trust each channel's numbers, and what to do next.
</task>
 
<requirements>
- Compute CTR, CPC, CVR, CPA and ROAS per channel, showing each step.
- Find the biggest drop-off in the funnel (AARRR) before recommending more traffic.
- Reconcile platform conversions with our database: explain the gap (attribution windows, view-through, double counting, model choice) and give a range for each channel's real contribution.
- Design one holdout or lift test to measure the largest channel's incrementality.
- Never invent numbers, names, quotes or sources: write [NEEDS SOURCE] where support is missing, and [VERIFY] beside any benchmark you cite.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The arithmetic is right: recompute each total a second way.
- No recommendation rests on platform ROAS alone.
- Every "cut" or "scale" call says what result would reverse it.
</quality_bar>
 
<output_format>
A metrics table, the funnel with the biggest leak marked, a reconciliation table (channel | platform-reported | estimated real range | why), the holdout test design, then 3 next steps.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

Write a Facebook ad for my bookkeeping software.

<context>
Product: Ledgerly, bookkeeping software, $49/month per firm, 30-day free trial, imports from QuickBooks in one click.
Audience: owners of US bookkeeping firms with 1-10 staff · awareness: problem-aware. Month-end close takes them about 6 days and eats weekends; they haven't heard of us.
Proof we can use: 300 firms use Ledgerly; in our survey of 85 customers, the median close fell from 6 days to 2.
Channel: paid social, feed placements.
</context>
 
<task>
6 paid social ads whose only goal is free-trial starts, built as a test.
</task>
 
<requirements>
- 3 angles (time back, fewer errors, calm month-end), 2 ads each: one PAS, one Before-After-Bridge.
- Per ad: a hook for the first line (under 12 words), primary text under 125 characters, a headline under 40 characters, a CTA button, and an image idea.
- No exclamation marks, no hype words.
- Never invent proof: no claims beyond the proof above; write [NEEDS PROOF] where you'd want one.
- End with a list of the assumptions you made.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every number in an ad appears in the proof above, with the same meaning.
- The hook names the problem in the reader's terms.
- The six ads differ in ways a test can measure.
</quality_bar>
 
<output_format>
A table: angle | framework | hook | primary text | headline | CTA | image idea, with character counts. Then which 2 ads to test first and why.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change your answer; otherwise go on.
</process>

1. Time back · PAS
   Hook: Month-end shouldn't cost you a weekend.
   Primary (101): Six days of matching bank lines, every month. Ledgerly matches them for you. Try it free for 30 days.
   Headline (25): Close month-end in 2 days
   CTA: Sign up · Image: a laptop shut on a Saturday morning
 
2. Time back · Before-After-Bridge
   Hook: Picture Friday with the books already closed.
   Primary (80): Before: 6-day closes. After: 2. 300 firms already close in 2 days with Ledgerly.
   Headline (32): Join 300 firms closing in 2 days
   CTA: Sign up · Image: a calendar with 4 days crossed out
 
3. Fewer errors · PAS …
 
Test first: ad 2 (strongest proof) against ad 1 (strongest pain).
Assumptions: feed placements only; the trial needs no credit card.

Ad 2 says "300 firms already close in 2 days", but the proof is two separate facts: 300 firms use Ledgerly, and in a survey of 85 customers the median close fell from 6 days to 2. Rewrite only the lines that state or imply a 2-day close for everyone (ad 2's primary text and headline, and ad 1's headline) so each claim keeps its real meaning, e.g. "Customers' median close: 2 days (survey of 85)". Keep everything else, recount the characters, and recheck that the suggested test still compares pain against proof.