7 Days, 14 E-commerce Videos with Genra: Full Workflow + Real Numbers

Small Shopify store. $5K/month. 1 founder. No video team. Goal: 14 finished videos in 7 days, $300 ad budget. Here's the day-by-day, the prompts, the data — and one thing that totally failed.

I run a small Shopify store that sells minimalist desk accessories — walnut trays, brass pen holders, a ceramic coffee coaster set, a few cable organizers. Average order value sits around $48. Revenue last month was $5,200. Traffic is roughly 60% TikTok organic, 25% Instagram, the rest a mix of email and direct. The "video team" is me, between customer support emails and packing orders. Before this experiment my video output was about one bad iPhone Reel a month, usually shot in bad light on the kitchen counter.

I gave myself a constraint: 7 days, 14 finished videos, $300 ad spend cap, document everything so the workflow is replicable for anyone in a similar spot. No agency, no freelance editor, no new camera gear. The only tool I let myself add was Genra, because I needed something that could actually finish videos end-to-end — not just generate clips I'd then have to assemble in Premiere I don't know how to use.

Here's the headline before the long version: I shipped 14 finished videos in 7 days. Total founder time was about 14 hours. Best ad CTR was 2.8% (roughly +30% above the category benchmark for home goods). Best ad ROAS hit 1.9x at hour 72 of a 3-day learning phase and was still climbing. One ad totally bombed and I killed it at hour 18. The rest of this article is exactly what I did, the briefs I wrote, what worked, and what didn't.

The 7-Day Plan

Before generating anything I wrote out the whole week on a single page. The portfolio mix — 4 product hero, 4 product feature, 4 hook variants, 2 testimonial-style — wasn't arbitrary. It maps to how a small store actually uses video: hero videos for the product page and top-of-funnel ads, feature videos for retargeting, hook variants for paid optimization, and UGC-style for social proof. Here's the schedule I held myself to:

The two days I was most worried about going in were Day 1 and Day 6. Day 1 because if the brand brief was sloppy, every generation downstream would be sloppy too. Day 6 because publishing and ad setup is the kind of work that can quietly eat 6 hours and ruin the whole sprint. I'll come back to both.

Day 1 — Strategy

I spent the morning on the portfolio mix. The 4-4-4-2 split came from a simple logic: for a small store with 4 hero SKUs, you need at least one strong hero per SKU (that's the 4), each SKU then needs one feature/benefit angle for retargeting (another 4), the single best performer deserves enough hook variants to actually run a real test (4), and 2 UGC-style videos give you a feel-of-customer angle for organic. Total: 14.

The brand brief took about 90 minutes. I forced myself to write it on a single page so the constraints would actually be tight. Here's what was on it:

• Voice: warm, founder-led, no jargon. Think "a friend who happens to make these things." Sample lines I wrote: "designed for the work-from-home minimalist," "made one at a time, in a workshop in Vermont."
• Color palette: warm neutrals — walnut brown, oat, cream, soft black accent. No saturated colors. Daylight white balance, never blue-toned.
• Product photo references: I pulled 8 existing product shots from Shopify and dropped them in. These became visual anchors Genra could reference for texture and proportion.
• Target audience: 28–42, work-from-home professionals, $80K+ household income, follows accounts like Cup of Jo and Kinfolk.
• Tonal don'ts: no neon, no fast cuts, no upbeat stock music, no "shop now!" voiceover energy.

Total Day 1 time: about 2 hours. I almost cut it down — it felt like procrastination — but it ended up being the highest-leverage hour of the whole week. Every later video pulled from this one page.

Day 2-3 — 8 Product Videos (with the actual briefs)

Day 2 and 3 were the production grind. I batched all 8 in two sittings. Per-video active work was 6-9 minutes; the rest was generation time, which I used to answer customer emails and pack orders. Below are 3 of the briefs I gave Genra, copied from my notes. These are not model-level prompts — they're the higher-level briefs an agent like Genra translates into the actual prompt and shot list.

Brief 1 — Walnut Tray Hero (Day 2)

"15-second product hero for our walnut desk tray. Framing: a hand placing the tray on a sunlit oak desk near a laptop and a small ceramic mug. Voiceover hits the line 'designed for the work-from-home minimalist' in the first 4 seconds. End on a 2-second matte texture closeup of the walnut grain. Warm neutral palette per the brief. No music swells, just soft ambient room tone."

Brief 2 — Brass Pen Holder Feature (Day 3)

"20-second feature video for the brass pen holder. The angle is the patina — show three close-ups of how the brass darkens over time (months 1, 6, 12) so the customer understands it's the feature, not a defect. Voiceover: 'It's supposed to do that. Brass remembers your hands.' End frame on the 12-month patina with light catching one edge."

Brief 3 — Cable Organizer Hero (Day 2)

"15-second product hero for the leather cable organizer. POV-ish overhead of a messy desk transitioning to a clean one with the organizer in place. Voiceover lands the line 'fewer cables, more thinking' at second 5. End on a still composition of the finished desk. Daylight, warm neutrals only."

Genra returned each as a finished video — generated visuals, voiceover in the brand voice, captions, and a final edit. I approved 5 of 8 on the first pass. I rejected and re-briefed 3 of 8 — that's a 38% rework rate. Two were rejected because the voiceover energy felt too sales-y; one because the visuals leaned warmer than the brief and started to look like a butter ad. After rework, all 8 shipped. Day 2 + Day 3 active time, including review and re-briefs: about 3.5 hours combined.

Day 4 — Hook Variants

I posted the 4 hero videos to TikTok organically on Day 2 evening with no promotion. By Day 4 morning the walnut tray hero had pulled ahead — about 4x the views of the next one and a much stronger save rate. So the Day 4 portfolio focused entirely on it: 4 hook variants of the same body, each opening with a different formula.

The four hook formulas I tested came from Genra's earlier guide on 3-second openers. I used four of the five ad-format hooks:

• A1 Specific Pain: "If you've ever had a desk so cluttered you couldn't find your own pen…"
• A2 Big Specific Promise: "Clean desk in under 10 seconds. No drawer organizers, no plastic trays."
• A3 Pattern Interrupt: opens on a hand deliberately knocking 6 pens off a desk, then cuts to the tray catching them.
• A4 Negation: "Don't buy another desk organizer until you've watched this."

Same 12-second body across all four. Same voiceover style. Only the first 3 seconds varied. Total time to brief and produce all 4 variants: 35 minutes. This was probably the most efficient day of the sprint. Hook A/B is just an exceptionally high-ROI activity inside this workflow.

Day 5 — UGC-Style Testimonial Videos

The 2 UGC-style videos were the hardest call of the week. Real customer testimonials would have meant scheduling interviews, which doesn't fit a 7-day sprint. So I made what I'll call "feel-of-UGC" videos — first-person, hands-on product use, casual unscripted feel. I want to be clear: these are not pretending to be real customer testimonials, and I'd never run them as such in paid creative. They're closer to "what it feels like when one of my customers unboxes this."

The brief I gave Genra for both:

"POV unboxing, hands-on, first-person camera, casual unscripted feel, 30 seconds. Hand pulls product out of recycled kraft packaging, places on a real-feeling cluttered home-office desk, brief 'oh nice' moment, then a 5-second use shot. Audio: ambient apartment sound only, no music, light handling sound."

Video #1 (cable organizer unboxing) came out well. Video #2 (walnut tray) had a problem I didn't catch until I watched it on a phone screen: the hand model in the close-up tipped slightly into uncanny-valley territory — fingers a touch too smooth, a knuckle that didn't quite move right. On a laptop preview I missed it. On a phone, in motion, it was the kind of thing that makes you scroll. I shipped it anyway because I'd committed to documenting everything, and it became the cautionary tale of the sprint. More on that in "What Didn't Work."

Day 6 — Publishing & Ad Setup

Day 6 was administrative, not creative. The work split:

• Meta Ads: picked 7 of the 14 to push as paid creative — the 4 hook variants plus the 3 strongest feature videos. $43 budget per ad, $300 total. Meta Advantage+ audience, 3-day learning phase, identical settings across all 7 so the only variable was the creative.
• TikTok organic: all 14 scheduled across 14 days, two posts per day for the first week, one per day for the second.
• Instagram organic: all 14 over 14 days — Reels feed for the heroes, Stories for the features, a carousel still + Reel pairing for the UGC-style.
• Shopify product page embeds: the 4 hero videos went into the relevant product pages above the fold. Took about 20 minutes.

Tracking was nothing fancy: UTMs on every paid placement, Shopify pixel firing, Meta ad-level reporting. Day 6 active time: about 3 hours. Most of it was the ad setup; publishing itself was fast.

Day 7 — Initial Results & Iteration Plan

The first 24 hours of ad data are too short to draw final conclusions, but they're enough to make decisions about kill, hold, and scale. Here's what came back:

• Best ad: walnut tray hero with the A2 (Big Specific Promise) hook variant — CTR 2.8%, CPM $14, ROAS 1.9x at $43 spend. Still in learning. Strong save rate organically as well.
• Worst ad: testimonial #2 (the uncanny-valley walnut tray UGC) — CTR 0.6%, killed at hour 18.
• Average CTR across the 7 paid ads: 1.4%. Industry benchmark for home goods sits in the 0.8–1.2% range, so the average came in roughly +30% above category benchmark.
• Strongest organic: walnut tray hero on TikTok — 18,000 views in 48 hours, no boost, 412 saves.

Iteration plan I committed to before closing the laptop on Day 7:

1. Scale the walnut tray + A2 hook variant. Move budget from the killed testimonial into this one.
2. Generate two more A3 (Pattern Interrupt) variants of the walnut tray and A/B against the current A2 winner.
3. Re-shoot testimonial #2 next week with a clean tray-only POV (no hands in close-up) so the uncanny-valley issue is gone.
4. Hold the 3 feature videos at flat budget for another week to gather more data.

The Numbers

Pulled into a single table for easy comparison:

Two caveats. First, this is one store, one week, one product category — don't read these as universal benchmarks, read them as "what was achievable here." Second, the 3-day ROAS is still inside Meta's learning phase; the number that matters is 14-day post-purchase, which I'll know in two weeks. I'll add a follow-up note when that data lands.

What Didn't Work

The most useful thing I can give you isn't the wins — it's the failures, because those are what stop you from repeating them.

Testimonial #2 and the uncanny-valley hand

The hand model in the walnut-tray POV close-up was the issue. Fingers slightly too smooth, a knuckle that didn't move quite right. Watched on a laptop, fine. Watched on a phone in motion — which is how 95% of the audience watches — visibly off. CTR was 0.6% and it killed retention curves on every platform. Lesson: for hands-on or face-on UGC, either blend AI with a single real-clip insert for the close-up, or pick a hook formula that doesn't depend on hand visibility (a tabletop POV, an over-the-shoulder, a packaging-only shot). Don't put AI hands center-frame in close-up yet.

The 38% rework rate, and what caused it

Three of 8 product videos needed re-briefing. The common cause across all three: my brand voice prompt was too vague the first time. I'd written "warm, founder-led, no jargon" and assumed that was enough. It wasn't. The fix was adding 3 sample lines of voiceover from my own emails to the brief. After that, the voice landed on first pass for the remaining videos. Plan for some rework on the first batch — and tighten the voice section of the brief with actual sample lines, not adjectives.

The misspelled product caption

One feature video came back with the product name misspelled in the auto-caption ("walnit" instead of "walnut"). 3-minute fix in post, but it would have shipped if I hadn't watched the final all the way through. Always watch the captions on a phone before publishing.

The Reusable Workflow Template

Stripped down to what you can actually run on a Monday morning:

1. Audit your asset library. 30 minutes. Pull every existing product photo, brand color, customer email, and Shopify review you can use as a reference. You probably already have the raw material for a brand brief.
2. Write the 1-page brand brief. 60–90 minutes. Voice (with sample lines, not adjectives), color palette, target audience, tonal don'ts, 6–8 reference images. Force it to one page.
3. Pick a 14-video portfolio mix. The 4-4-4-2 mix worked for me. Adjust to your SKU count: more SKUs means more heroes, fewer means more hook variants on your strongest performer.
4. Block 2 days of generation with rest cycles for review. Don't try to review videos the same hour you brief them. The 6–9 minutes of generation time per video is not productive review time — use it for other work and come back fresh.
5. Publish in waves and A/B hooks on the breakout winner. Don't run ads on cold creatives. Post organically first, let one pull ahead, then make 4 hook variants of the winner and run those as paid.
6. $200–400 paid test, kill bad ads at 24h, scale winners at 72h. Don't sit and watch. Set the kill threshold (CTR below 0.8% or no purchases at $40 spend) and let the rules do the work.

The shortcut: Genra runs steps 3–5 end-to-end. Once the brand brief from step 2 is written, the agent picks the portfolio mix, generates the videos, and applies the hook variants without you touching a timeline. If you want to replicate the first 3–4 videos of this case study before paying for anything, the Genra free trial gives 40 credits — more than enough. Start at genra.ai.

Key Takeaways

• 14 finished videos in 7 days is achievable for one founder using an end-to-end AI agent.
• Total founder time: about 14 hours over 7 days. Roughly 1 hour of human time per finished video.
• Best ad CTR was 2.8%, roughly +30% above home goods benchmark on a $300 test budget.
• First-pass rework was 38%. Plan for it. Tightening the brand brief with sample lines (not adjectives) cuts the rework rate.
• Testimonial-style with prominent hand or face close-ups is still where AI struggles. Mix in real footage there or pick a hook that hides the hands.
• Hook A/B testing on the breakout organic winner is the single highest-ROI activity inside this workflow.
• The bottleneck isn't generation time — it's brief writing and review judgment.
• The 1-page brand brief is the highest-leverage asset you build all week.

Frequently Asked Questions

Is this case study a real specific company?

It's a composite. The store profile (small Shopify, ~$5K/month, minimalist desk accessories), the workflow, and the numbers are representative of what a real founder running this exact sprint can expect — drawn from observing several Genra users running similar 1-week sprints. The exact ad spend split, the rework rate, and the failure mode (uncanny-valley hand) are the patterns we see most often, presented as a single coherent week.

How long does this workflow really take per video?

Roughly 1 hour of active founder time per finished video, end-to-end (brief, review, re-brief if needed, publishing, ad setup share). The generation itself is 6–9 minutes per video, but you don't sit and watch — you do other work in parallel. Going from "want a video" to "video is live with a paid budget" is the 1-hour figure.

Can someone with zero video experience replicate this?

Yes, with one caveat. The agent handles the technical side — generation, voiceover, captions, edit. What it can't replace is the judgment to write a tight brand brief and to recognize a bad take when you see one. If you've never thought about your brand voice or audience before, plan to spend 90 minutes on the brief, not 30.

What's the most surprising finding?

The bottleneck wasn't generation. I expected to be limited by AI quality or generation time. The actual bottleneck was brief writing and review judgment — the human decisions, not the machine work. Tightening the brand brief with sample voiceover lines cut my rework rate roughly in half on the second batch.

How would 14 AI-made videos compare to 14 agency-made ones?

Different math. 14 agency-made videos at $1.5K–$3K each is $21K–$42K and at least 4–6 weeks of calendar time. The AI route was ~$72 in credits + $300 in ad spend + 14 hours of my time, in 7 days. Agency videos still win on emotional close-ups and on hero brand films where craft and casting matter. For volume product and feature creative for a $5K/month store, the AI route is the only realistic option.

What if my products are highly visual (apparel, jewelry) — does this still work?

Mostly. The product hero and feature videos work well for textured, tangible goods where you can describe materials and lighting (apparel fabric, jewelry surface). Where it gets harder is anything that depends on motion-on-body (apparel fit on a model, jewelry catching light at a specific angle on a wrist). For those categories, expect a higher rework rate or plan to mix in some real footage for the close-ups, similar to what I'd do differently for testimonial #2.

Where do I start if I want to replicate this on Monday?

Spend Monday morning on steps 1 and 2 of the reusable workflow above — audit your asset library, write the 1-page brand brief. That's the highest-leverage hour you'll spend all week. After that you can move into generation. The Genra free trial covers 40 credits, which is enough to replicate the first 3–4 videos of this case study before deciding whether to commit.