Piqki Creative Testing System

*Max's operating manual. Last updated: 2026-02-20 13:20 PST.*

*Gin's role: approve once per week. Max's role: everything else.*

1. System Overview

The creative testing engine runs on a 4-week iteration loop: we launch 5 concepts, kill the bottom performers at $50 spend, scale the top performers by doubling budget, and by end of week 4 we know exactly which financial scenario we're operating in and what to spend in May. Max pulls data every Monday, writes the Tuesday brief to Gin with exactly 1-2 decisions, executes Wednesday-Friday, and reviews organic Sunday. Gin does not design tests, pick audiences, interpret metrics, or decide when to cut or scale — she approves the brief and makes the 1-2 highlighted calls. Every creative gets a score before it runs. Anything under 6 doesn't go live. The whole thing is documented so that if a contractor steps in, they execute from this doc without a briefing call.

2. The Creative Brief Template

Use this for every single creative. No exceptions. Fill in every field before anything gets shot, designed, or briefed to a creator.

PIQKI CREATIVE BRIEF

`

Brief ID: [e.g., PIQ-001]

Date: [MM/DD/YYYY]

Pillar: [ ] P1 — Demand Gen [ ] P2 — Demand Conversion [ ] P3 — Loyalty

Platform: [ ] Instagram [ ] TikTok [ ] Both

Format: [ ] All-text [ ] Lo-fi video [ ] Carousel [ ] Story

HOOK

(First 3 seconds of video / first line of caption. Must stop the scroll cold.)

→

MESSAGE / ANGLE

(What's the core point? One sentence. If you need two, it's too complicated.)

→

WHAT IT SHOWS

(For video: describe what happens on screen, beat by beat. For text: paste the full copy.)

→

CTA

(If P2: what do you want her to do right now? If P1: no CTA — seriously, leave it blank.)

→

SUCCESS METRIC

(Primary KPI we're tracking for this specific creative.)

→

KILL THRESHOLD

(At what spend/metric do we turn this off? Default: $50 spend + <0.8% CTR for P2 / <3% ER for P1.)

→

CREATIVE SCORE (pre-launch)

Hook Strength: /10

Message Clarity: /10

Emotional Resonance: /10

CTA Quality: /10

Brand Alignment: /10

TOTAL: /50 → /10 average:

[ ] Score ≥6 → approved to test

[ ] Score <6 → do not run — revise or kill

`

3. April Test Plan — Week by Week

Context: April is the pre-launch paid test. Goal is NOT to drive sales at scale — it's to find our winning creative angles and benchmark our real CAC before we put serious May budget in. We're spending ~$2K in April to avoid wasting $20K in May on the wrong message.

WEEK 1 (April 1-7): Setup

Objective: Launch 5 concepts, establish baseline metrics, let data breathe.

Budget: $2,000 total for the month. Week 1 allocation: ~$500-600.

Platform: Meta only (Instagram). No TikTok yet.

Optimization event: Link clicks (not purchases — not enough volume yet to optimize for conversions in week 1).

Audiences:

1. Broad 18-35F US (no interest targeting) — lets Meta's algorithm find her

2. LAL 1% — seeded from Piqki email list

3. LAL 1% — seeded from IG followers

4. LAL 1% — seeded from video viewers (25%+ watch time) — deploy after first 3 days of organic video

Run all 4 audiences simultaneously. Small budgets ($10-15/day per audience). We're not optimizing audiences in week 1 — we're optimizing creatives.

Daily budget per creative: $50-75 per ad. 5 ads running = $250-375/day. Compress to 3-4 days to gather initial signal without burning the whole month.

The 5 Creative Concepts (see Section below for full detail):

• Concept A: "The Salon Math" (P1, all-text)
• Concept B: "I Used To Say I Wasn't A Press-On Person" (P1, lo-fi video)
• Concept C: "3 Minutes. Seriously." (P2, lo-fi video)
• Concept D: "The PiqWheel Explained" (P2, lo-fi video)
• Concept E: "What Good Nails Actually Look Like" (P1, all-text/image hybrid)

What Max is watching in Week 1:

• CPM (target <$15 — if >$20, audience or creative is off)
• CTR link (baseline — don't kill yet, just observe)
• Hook rate (% who watch past 3 seconds) — if available in Meta Ads Manager
• Frequency (keep <2.0 in week 1 — if we're hitting the same person twice already, our audience is too small)
• Cost per link click (early CAC proxy — target <$3)

Max's week 1 output to Gin (Tuesday April 7):

> "All 5 concepts are live. Here's early CTR: A is at [X]%, B at [X]%. Too early to kill — data still thin. Watch what I send you Tuesday April 14."

WEEK 2 (April 8-14): First Cuts

Objective: Kill underperformers. Scale early winners. Concentrate remaining budget.

Kill rule: Any creative with <0.8% CTR after $50 spend → pause immediately. No exceptions. Don't give it more money hoping it turns around.

Scale rule: Any creative with >2.5% CTR after $50 spend → double daily budget. This is your signal.

Scaling floor: Don't scale below $50 total spend. CTR at $20 spend is noise.

What Max watches this week (beyond CTR):

• Cost per add-to-cart (if pixel is firing correctly)
• Audience performance: which of the 4 audiences has the lowest CPLC? Cut the worst 2, concentrate on best 2.
• Creative fatigue: if frequency is already >3 on any ad set → rotate creative or pull back spend

What Max tells Gin (Tuesday April 14):

> "Killed A and D — both under 0.8% CTR at $50 spend. C is our winner at 3.1% CTR — doubling budget. B and E still gathering. Budget reallocation attached. No decisions needed from you this week."

*(If there's a budget reallocation or a creative direction question, that's when Max surfaces the 1-2 decisions.)*

WEEK 3 (April 15-21): Optimize

Objective: Take the winning angles and iterate. Move from CTR to CAC.

Iterations: Take the winning creative format/hook from week 2 and test 2-3 variations:

• Same hook, different format (e.g., all-text → lo-fi video)
• Same format, different hook line
• Same concept, different proof element (different model/hands, different day count for wear)

Brief 3 new creatives based on what won. Use the same brief template. Score them before launching.

Audience expansion: If CAC is tracking ≤$30, add 2 new lookalike seeds:

• LAL from Shopify purchasers (once first test purchases come in)
• LAL from ManyChat leads (if ManyChat is generating volume)

What Max is tracking this week:

• First purchase CAC — not just CTR. This is the real number. CTR is a proxy; CAC is the truth.

- Formula: total ad spend ÷ number of first purchases attributed to ads

- Target: CAC ≤$25 (Base scenario)

• Add-to-cart rate (if people are clicking but not buying, the landing page has a conversion problem — flag for Gin)
• AOV (are people buying 1 unit or 2? This affects the CAC math significantly)

Mid-month checkpoint (Max, no Gin input needed):

If winning creative CTR >2.5% AND estimated CAC is trending ≤$30:

→ Keep scaling. Stay the course. Report to Gin Tuesday the 21st.

If CTR >2.5% but purchase data shows CAC >$40:

→ Flag immediately. Landing page or offer is the problem, not the creative. Escalate to Gin same day.

WEEK 4 (April 22-30): Decision

Objective: Deliver the May scaling decision with full confidence.

By April 30, Max has:

• 2-3 creative concepts with validated CTR and real CAC data
• Audience performance data across 4 lookalike seeds
• Enough purchase volume to calculate statistically meaningful CAC (target: 50+ purchases)
• A clear recommendation on which financial scenario we're operating in

CAC Decision Table:

Max's April 30 brief to Gin:

> "April results: CAC landed at $[X]. We're in [Scenario] territory. My recommendation: [specific action]. Here's what's working: [winning creative angle]. Here's what to fix before May: [1-2 items]. Your call: approve [May budget] or [alternative]. Need your answer by [date] to brief May creative in time."

This is the one decision Gin makes: the May budget. Max has done everything else.

The 5 April Creative Concepts

CONCEPT A — "The Salon Math"

• Hook: "the salon appointment funnel: book 2 weeks out → cancel → rebook → show up → wait 30min past your appointment → pay $95 → tip → chip on day 6"
• Format: All-text post. Static image (dark background, single text block). No logo. No CTA. Just the text.
• What it shows: The absurd ritual, listed out. Dry, deadpan, accurate. Nothing else.
• Why it should work: She's lived this. She sends this to her friend. It earns shares and follows from people who don't know Piqki yet. The comment section writes itself. P1 at its purest — establishes the problem without introducing the product.
• Pillar: P1

CONCEPT B — "I Used To Say I Wasn't A Press-On Person"

• Hook: (spoken or text overlay): "I used to say I wasn't a press-on person. Turns out I was just a bad-press-on person."
• Format: Lo-fi video, 30-45 seconds. Shot on iPhone, natural light. Model's hands, kitchen or desk setting.
• What it shows: Starts with bare nails. She opens the PiqWheel. We see the sizing wheel — she spins it, finds her size immediately. She presses on. Final shot: nails done, her going about her day (coffee cup, laptop, scrolling). Text overlay at end: "3 minutes. actually."
• Why it should work: Directly addresses the objection that killed the category for her. Reframes press-ons as a sophistication signal, not a shortcut. The PiqWheel visual is inherently interesting — no one's seen it before. Builds desire through reframe, not product feature.
• Pillar: P1 (with product present, but selling the idea, not the SKU)

CONCEPT C — "3 Minutes. Seriously."

• Hook: "3 minutes. I timed it."
• Format: Lo-fi video, 60 seconds. Real-time (or close to it) application. Clock visible in corner or timestamp overlay.
• What it shows: Timer starts. Model opens PiqWheel. Sizes on both hands (the wheel makes this fast — visible). Applies each nail. Timer stops. Final look. No voiceover needed — the timer does the work. CTA at end: "Comment WHEEL" text overlay.
• Why it should work: The claim "press-ons in 3 minutes" is unbelievable — which means showing it in real time is the only way to prove it. The clock creates tension and payoff. P2's job is to convert already-interested people into buyers. This is the proof video that closes the gap between "interested" and "buying."
• CTA: Comment WHEEL → ManyChat delivers link
• Pillar: P2

CONCEPT D — "The PiqWheel Explained"

• Hook: "why does this not exist everywhere" (text overlay on first frame of the wheel)
• Format: Lo-fi video, 45 seconds. Close-up product focus. No model face needed — hands only.
• What it shows: Opens on the PiqWheel in hand. Camera slowly rotates it — you see all the sizes arranged in the wheel format. Model spins to her size. Pulls it off the wheel. Presses on. Pull back to show both hands done. Text overlays explain: "finds your size in seconds" / "no more guessing" / "32 nails, zero waste." End with product name + "link in bio."
• Why it should work: The PiqWheel is Piqki's most differentiating physical feature before the application device launches. Showing it clearly, in a non-salesy way, triggers curiosity and saves ("I need to look at this later"). This is an explanatory P2 — it converts people who've seen P1 and want to understand what Piqki actually is.
• CTA: Link in bio
• Pillar: P2

CONCEPT E — "What Good Nails Actually Look Like"

• Hook: "not nail art. just nails." (standalone text post or photo caption)
• Format: Lo-fi image — close-up of model's hands with a Piqki set on. Real setting (book, coffee, steering wheel). No jewelry. No ring. Just the nails in a real moment. Single photo or carousel of 3.
• What it shows: The aesthetic. The "quiet luxury" angle. Nothing dramatic — just nails that look expensive and understated in a real environment. Caption: "not nail art. just nails. the [color name] set, link in bio."
• Why it should work: For the audience segment that's already quality-conscious and shopping on aesthetic alone, this is the content that makes them stop. It's the vibe before the product. Also serves as a benchmark for Piqki's visual identity — what we look like when we're not explaining anything.
• CTA: Link in bio
• Pillar: P1 (with subtle product presence — earns follows from aesthetic-driven audience)

4. Organic Content Calendar (April–May, Pre-Launch)

Goal: 1,000 followers before May launch. Establish the aesthetic. Seed the community. The account should look lived-in before we push hard.

Cadence: 3 posts per week. Tues / Thurs / Sat or similar. Consistent.

Rule: Every week covers at least 2 pillars. Never 3 weeks of the same pillar in a row.

APRIL — Week 1 (April 1-7): Establish the POV

APRIL — Week 2 (April 8-14): Establish the Problem

APRIL — Week 3 (April 15-21): Establish the Aesthetic

APRIL — Week 4 (April 22-28): Build Anticipation

MAY — Week 1 (April 29 - May 5): Pre-Launch Push

MAY — Week 2 (May 6-12): Proof Mode

MAY — Week 3 (May 13-19): Scale What's Working

MAY — Week 4 (May 20-26): Community Deepening

5. The Weekly Rhythm (Ongoing, Once Ads Are Live)

Max runs this every week without being asked. No exceptions.

Monday: Data Pull

• Pull from Meta Ads Manager: weekly spend, impressions, CTR per ad, CPM, ROAS, CAC.
• Pull from Instagram: engagement rate per post, saves, shares, followers gained.
• Pull from Shopify: sessions, conversion rate, AOV, new vs. returning customers.
• Flag any anomaly: CTR drop >20% week over week, CAC spike >$5, engagement rate collapse.
• Calculate MER (Marketing Efficiency Ratio = total revenue ÷ total ad spend). Target: ≥2.0.

Tuesday: Brief to Gin

Standard format, never more than 1 page:

`

WEEKLY PERFORMANCE BRIEF — [Date]

WHAT'S WORKING:

→ [Specific creative/angle] is at [X]% CTR / [X] shares. Keep running.

WHAT'S NOT:

→ [Specific creative] underperformed. Paused.

CAC THIS WEEK: $[X] — [vs. target]

MER THIS WEEK: [X] — [vs. target]

YOUR 1-2 DECISIONS THIS WEEK:

1. [Specific decision with recommendation]

2. [If needed — otherwise omit]

MAX'S ACTIONS THIS WEEK (no input needed):

• [What Max is handling]
• [What Max is handling]

`

Wednesday-Friday: Execute

• Implement any Gin-approved changes.
• Brief any new creatives (use brief template).
• Score new creatives before submitting to any creator or designer.
• Update ad sets: kill underperformers, scale winners, adjust audiences.
• If new UGC has come in: review, score, decide whether to run as paid or organic.

Sunday: Organic Review

• Review every organic post that published that week.
• Calculate engagement rate for each: (likes + comments + shares + saves) ÷ reach.
• Flag any post with ≥5% engagement rate → add to "boost queue" for Monday review.
• Note any comments or DMs that reveal insights about the audience (new objections, new language to test, new proof needs).

6. Creative Scoring System

Every creative gets scored before it goes live. No exceptions. Score assigned by Max, not Gin.

The 5 Criteria (each scored 0-10):

1. Hook Strength (0-10)

Does the first 3 seconds (video) or first line (text) make her stop? Would you stop scrolling if you didn't know this brand?

• 0-3: Generic, forgettable, or confusing
• 4-6: Decent, but could be anyone
• 7-9: Specific, true, and targeted — she feels seen
• 10: She's already sending it to a friend before it's over

2. Message Clarity (0-10)

After consuming this content, is it completely obvious what Piqki is and what it does?

• 0-3: She still doesn't know what was being sold
• 4-6: She could figure it out with effort
• 7-9: Clear without being explained
• 10: Product, value prop, and why-now are all immediately understood

3. Emotional Resonance (0-10)

Does it make her feel something? (Seen, validated, excited, amused, satisfied — any real emotion counts. "Meh" doesn't.)

• 0-3: Neutral, no reaction
• 4-6: Mild interest
• 7-9: Real reaction — she nods, laughs, screenshots
• 10: She shares it

4. CTA Quality (0-10) *(P1 posts: score 10 automatically if there's no CTA — absence is correct)*

Is the next step completely obvious? Is it low-friction?

• 0-3: No CTA or buried CTA
• 4-6: CTA present but passive ("check link in bio")
• 7-9: CTA clear and specific ("comment WHEEL to get the link")
• 10: CTA is part of the hook itself — the next step feels irresistible

5. Brand Alignment (0-10)

Does this feel like Piqki? Not a generic beauty brand, not Glamnetic, not Chillhouse — specifically Piqki.

• 0-3: Could be any brand
• 4-6: On-brand but generic
• 7-9: Distinctly Piqki — controlled chaos, dry confidence, nail-serious/self-not-serious
• 10: If she didn't know the brand and saw this, she'd want to find it

Score Calculation:

Total = sum of 5 scores ÷ 5 = average out of 10

Decision Rules:

Score tracking: Log every brief ID and its score in a running scorecard. Over time, this reveals which creative attributes consistently produce high scores AND high CTR — that's where the pattern is.

7. How to Accelerate CAC From $25 → $18

These are the actual levers. In priority order.

Lever 1: Hook Rate (Biggest impact — 80% of the battle)

The percentage of people who watch past the first 3 seconds of a video (or who stop scrolling on a text post) determines everything downstream. A bad hook means all your targeting, creative, and offer work is wasted on people who didn't engage.

• Test: A/B the first 3 seconds only (same video body, different opening). This isolates hook performance.
• What makes a Piqki hook land: specific truth + recognition + slight surprise. "The salon appointment funnel" works because it's specific and true. "Beautiful nails in minutes!" doesn't work because it's generic and expected.
• Benchmark: Hook rate (3-second video view rate) of ≥40% is good. ≥60% is exceptional. Below 30% → new hook, same video.
• Target: get hook rate from baseline to ≥45% on top 2 creatives before scaling.

Lever 2: Landing Page Conversion

If CTR is solid but CAC is high, the problem isn't the ad — it's the page. Common failure modes:

• Page loads slow on mobile (>3 seconds → significant drop-off)
• Product images don't match the ad aesthetic (incongruity kills trust)
• The sizing explanation is buried or missing
• No social proof above the fold
• The add-to-cart button isn't immediately visible

Max's job: after week 2 of paid, run a landing page audit. Check: mobile load time, first fold content, add-to-cart visibility, proof above fold. Flag any issue with a specific recommendation.

Lever 3: AOV Lift (From $50 → $60-65)

Every $10 increase in AOV drops effective CAC meaningfully. $25 CAC on a $50 AOV = 50% CAC/Revenue. $25 CAC on a $65 AOV = 38%.

How to raise AOV:

• Post-add-to-cart upsell: "Add a second set for $20 more" (if Shopify supports it — set up before launch)
• Bundle: "The Starter Two-Pack" as a hero offer instead of single unit
• ManyChat flow: after delivering the link, send a follow-up message with "most people grab two — here's the bundle"
• Don't discount to raise AOV — bundle to raise it.

Lever 4: Creator UGC Seeding

UGC from real creators consistently outperforms branded content on CAC, for one reason: it doesn't look like an ad. Target 10-15 creators before launch.

Criteria for Piqki UGC creators:

• 5K-100K followers (micro and nano creators — more trust, lower cost)
• Aesthetic match: Silver Lake/Bushwick vibe, lifestyle-forward, real-life settings
• No nail brand exclusivity agreements
• They post their actual life, not just sponsored content

How to seed:

• Send free PiqWheel set (no payment initially for organic-only seeding)
• Brief: "here's the one thing to show — the wheel and the timing. otherwise, make it yours."
• Do NOT give them a script. Give them the brief. Their authentic reaction is the creative.
• Ask for content rights in the outreach. Get written confirmation before running as paid.

Track: who posted, engagement rate on their post, whether their followers commented or asked about the product. This data feeds the creator lookalike audience.

Lever 5: Lookalike Seed Quality

Not all lookalike audiences are equal. The seed matters more than the lookalike percentage.

Best seeds for Piqki, in priority order:

1. Purchasers — highest intent, best conversion behavior. Build this fast.

2. ManyChat leads — people who commented WHEEL. High intent signal.

3. Email list — if it has real nail-interested people in it (verify quality)

4. Video viewers (25%+ watch time) — engaged but not bought. Use for P2 LAL.

5. IG followers — lowest quality (includes people who follow everything). Use last.

Rule: always test LAL against broad before assuming LAL wins. Sometimes broad outperforms. Let the data decide.

8. When to Add TikTok

TikTok is not in the April plan. It's earned, not assumed. Here's exactly when to add it:

All four criteria must be true:

When criteria are met:

1. Mirror the top 3 performing Meta creatives on TikTok — same hooks, reformatted for vertical/native TikTok feel

2. Start TikTok ads at 20% of Meta budget ($15K/mo if Meta is at $75K)

3. Track CAC on TikTok separately — do not blend with Meta numbers

4. Give it 30 days before making a scaling decision on TikTok

Why we wait:

TikTok before Meta CAC is proven wastes money and dilutes learning. Win on Meta first. The creative learning from Meta (what hooks work, what messages convert) transfers directly to TikTok. We're not learning twice; we're proving once and scaling the winner.

*Document ends. For brand strategy and content pillar detail, see: piqki-content-playbook.md*

*For financial scenario context, see: piqki-financial-updated.md*

Sources & Validation

[1] Hook rate benchmark — "≥40% 3-second view rate is good, ≥60% is exceptional, <30% → new hook"

Source: Meta Ads Manager published guidance + SBS workshop (Strong Brand Social, "F* The Algorithm 2.0," Feb 17 2026) + DTC practitioner consensus.

Status: Industry directional benchmark — not independently peer-reviewed. The 40%/60% thresholds are widely cited across DTC advertising circles and align with Meta's own creative best practices documentation. Piqki-specific hook rate benchmarks will be established in April testing. Until then, treat these as calibration targets, not hard rules.

[2] "$50 kill threshold for paid creative" (cut any creative with <0.8% CTR at $50 spend)

Source: DTC practitioner methodology — Meta advertising best practices. The $50 figure is a commonly recommended minimum for statistical signal in early creative testing.

Status: Directional — adjust based on Piqki's actual CPM. If CPM is high (>$15), $50 spend may be too few impressions to kill a creative. If CPM is very low (<$8), $50 generates strong signal faster. Max to recalibrate kill threshold in Week 1 based on actual CPM data.

[3] "CAC <$25 = Base scenario, CAC ≤$18 = Optimistic" — the April CAC decision table

Source: Piqki Financial Scenario Analysis (piqki-financial-updated.md) — scenario thresholds built by Max, Feb 2026. Informed by Northbeam DTC Benchmarks Report 2024 (northbeam.io/resources) and industry-observed ranges for DTC beauty brands.

Status: Internally consistent — not independently validated for press-on nail category specifically. The press-on nail market has limited published CAC benchmarks. The $18-35 range is reasonable for DTC beauty broadly. Piqki's actual CAC may fall outside this range — both better (strong identity-based content, high-intent audience) or worse (new brand, no social proof yet). April data is ground truth.

[4] "Non-founder content outperforms founder content on TikTok — Glamnetic: 244K vs. 124K avg views"

Source: Upwork social media dataset, Glamnetic TikTok, n=433 unique posts, scraped Feb 2026. Direct data.

Status: High reliability for Glamnetic specifically. This is actual platform data, not survey data. Caveat: Olive & June shows the opposite pattern. This is not a universal law — it reflects Glamnetic's specific brand-building approach. Piqki should monitor whether this holds true in their own early content, especially if any founder-adjacent content is tested.

[5] "UGC from micro/nano creators outperforms branded content on CAC"

Source: Multiple industry sources — Meta's own "Creators as Collaborators" research (2023), Smartly.io Creative Intelligence Report 2024, and SBS workshop [1]. Widely documented across DTC practitioner content.

Status: Well-established directional principle — not Piqki-specific. The mechanism is documented: content that doesn't look like an ad bypasses ad-fatigue and earned-trust filters. The specific CAC improvement magnitude varies widely by brand and product. Max to track UGC CAC vs. branded content CAC separately in April to establish Piqki's own numbers.

[6] "Lookalike audiences: purchaser seed > email seed > IG follower seed (quality order)"

Source: Meta advertising practitioner consensus — documented across multiple DTC case studies and Meta's own LAL audience guidance. SBS workshop reinforces prioritizing high-intent seed lists.

Status: Industry directional — validated by Meta's own published recommendations. URL: facebook.com/business/help/lookalike-audiences. Note: Meta's algorithm changes mean LAL performance fluctuates. Broad targeting (no LAL) sometimes outperforms LAL — Max to A/B test broad vs. LAL in Week 1 rather than assuming LAL wins.

[7] "MER (Marketing Efficiency Ratio = total revenue ÷ total ad spend), target ≥2.0"

Source: DTC practitioner metric — widely used as a blended efficiency measure. MER accounts for channel mix effects that single-channel ROAS misses (e.g., email, organic, direct). Popularized in DTC circles by operators including Taylor Holiday (Common Thread Collective) and similar agencies.

Status: Operationally sound metric — target threshold (≥2.0) is directional. For a new brand with no repeat purchase history, early MER may be below 2.0 as brand awareness builds. The threshold should be calibrated to Piqki's actual margin structure (see financial model) rather than applied as a universal rule. Max to set Piqki-specific MER targets by end of April based on actual COGS and contribution margin.

*Sources last reviewed: 2026-02-20 13:20 PST*