From slow-selling products to a flash sale and win-back audience

Hey! I'm your Growth Analyst — I can dig into revenue, orders, AOV, top products/brands, loyalty tiers, promo performance, and help you define audiences and draft campaigns.

What would you like to look at? A few things I can pull quickly: - Revenue / AOV over a time window - Top products or brands by sales - Loyalty tier breakdown - Promo code performance - A targeted audience segment for a campaign

Just point me at a question.

Hi again! What can I dig into for you?

I'm ready to run analysis on the Brand F data — revenue, AOV, top sellers, loyalty tiers, promo performance — or help you build a targeted audience. Just tell me what you'd like to see.

Hey! Looks like we're in a loop 😄 — I'm here and ready.

To get us moving, want me to pull a quick business snapshot? For example, I can run:

- Last 30 days: revenue, order count, AOV - Top 5 brands by sales - Loyalty tier breakdown

Just say the word (or give me any other question) and I'll run the query.

I'll pull the lowest-selling items by units sold. Let me check the data.Here are your 5 lowest-selling items by units sold:

Note on the numbers: The top 2 (Brand D, Brand A) each sold in a single order — these look like near-dead SKUs, possibly recent additions or niche/high-price stragglers. Items #3–5 sit in a cluster around ~19–24 units, so there's a real gap between the bottom 2 and the rest.

Next step ideas: - If these are aging inventory, a clearance markdown campaign could move them. - If they're recent launches, low sales may just be short shelf-time — I can check created_at on the products to tell the difference.

Want me to check whether these are new arrivals vs. stale stock, or draft a clearance campaign around them?

You just saw the bottom 5 by units sold. Do you want the same list again, or a different definition of "worst" — e.g., by revenue instead of units? Let me pull the lowest by revenue so you have that angle too.Here are your 5 worst-selling products by revenue:

Two different stories here: - #1 & #2 (Brand D, Brand A) — low revenue *because* almost nobody's buying them (1–2 units, single order each). These are your true weak sellers. - #3–5 (socks) — low revenue simply because they're cheap items ($6 socks). They actually sell fine on volume (35–46 units), they just don't generate much money each.

So depending on what "worst" means to you: - Weakest demand → the Brand D and Brand A (same top 2 as your units list). - Lowest revenue contribution → the socks, but that's mostly a price-point artifact, not a performance problem.

Want me to check whether the Brand D and Brand A are recent arrivals (short shelf-time) vs. genuinely stale stock before you decide what to do with them?

That's the same question — and it depends on how you measure "worst." Here are both cuts so you have them side by side:

By units sold (fewest bought):

By revenue (least money generated):

Bottom line: No matter how you slice it, the Product 1 (1 unit) and Brand A Tabi Sandal (2 units) are your genuine worst performers — barely any demand at all.

To move forward productively, tell me which you care about — weakest demand (units) or lowest revenue — and whether you want me to check if these are new arrivals vs. stale stock, or draft a clearance campaign to move them.

Before I hand this to the Campaign Manager, I need a couple of decisions from you — and let me first check whether these are stale stock (worth clearing) or recent arrivals, since that changes the play.Let me check the actual schema for the product/variant table.No created_at variance to lean on (all seeded same day), so I'll go with price and inventory to size the markdown.

Sorry for the digging — here's what I found and what I need from you.

These are the 5 worst-selling products (by units), with current price and stock on hand:

This is a classic clearance play — move slow stock with a discount. Before I hand it to the Campaign Manager, I need three quick decisions:

1. Offer depth — I'd suggest 30% off on these SKUs (deep enough to actually move dead stock without giving it away). OK, or higher/lower? 2. Validity window — I'd suggest a 2-week run. What start/end dates do you want? 3. Audience — a clearance sale is usually all marketing-opted-in shoppers (broad reach to move units). Or do you want to target a specific segment (e.g., loyalty tiers)?

Give me the go on depth + dates + audience and I'll write the brief and kick it off to the Campaign Manager. My default recommendation: 30% off, 2 weeks starting today, all opted-in customers. Want me to run with that?

Here are your 5 worst-selling items in the last 30 days (window: Jun 24 – Jul 24, 2026, by units sold):

Notes: - All five moved just 1 unit in the past 30 days — there's a long tail here (many SKUs tied at 1), so the ranking within this group is essentially arbitrary. I sorted by units, then whatever the DB returned. - The Product 1 shows up as worst in both the all-time and 30-day views — that's your most consistently dead SKU. - The two Brand E eyewear pieces are high-ticket ($455) slow-movers — one sale each, worth watching but not necessarily discounting given the margin.

If you want a cleaner "worst" list, I can filter to items with meaningful stock on hand (so you're targeting stuff you actually need to move) rather than the 1-unit long tail. Want me to do that — or fold any of these into the clearance campaign we were scoping?

Before I hand this off, let me confirm the plan and lock the SKUs, prices, and window — then I'll kick it to the Campaign Manager (which owns the banner image, subpage, promo code, and draft campaign).

Campaign: 30% Clearance — Slow Movers (last 30 days)

The 5 target SKUs:

- Offer: 30% off, these 5 SKUs only - Audience: everyone (product-gated clearance, no segment) - Window: you didn't give dates — I'll default to starts today, runs 14 days. Say the word if you want different dates.

One flag: this list is the 1-unit long tail from the 30-day view. The Brand H sock at 30% off is $4 — a clearance on a $6 item is negligible. Happy to proceed as-is, but you may want to drop the sock. Proceed with all 5, or drop the sock?

If you're good with all 5, 30% off, 14 days from today, I'll send it now. Confirm and I'll go.

Here are your 10 worst-performing products in the last 90 days (Apr 25 – Jul 24, 2026, by units sold):

Notes: - Nine of the ten sold just 1 unit in 90 days; the Brand O clog sold 2. There's a broad long tail of single-unit SKUs, so the ordering within this group is essentially by revenue as a tiebreaker. - Brand D appears twice (Solana LEVE + Amsterdam Wrapped) — the brand is underperforming across styles, worth a closer look. - These skew toward premium footwear and accessories (Brand I, Brand J, Margiela, Brand N watch) — high-ticket items that naturally sell slower.

Want me to (a) build a clearance campaign around this 90-day list instead of the 5 we scoped, (b) filter to only items with meaningful stock on hand so you're clearing what actually needs moving, or (c) something else?

Here's the flash sale plan. Let me confirm the SKUs and pricing before I hand it to the Campaign Manager (which owns the banner, collection, landing page, promo code, and draft campaign).

Campaign: 50% Off Flash Clearance — 90-Day Slow Movers

- Offer: 50% off, these 10 SKUs only - Duration: 5 days, starting today (through Jul 29) - Audience: all shoppers (product-gated flash sale, no segment) - Deliverables: banner image, collection, landing page, promo code, draft campaign

One flag: 50% on premium items like the Brand N watch ($175 loss vs. $350) and Brand M sweater is a steep giveaway. Given these are 1-unit-in-90-days dead stock, that's a defensible move — but confirm you're OK going that deep on the high-ticket pieces.

Confirm and I'll send it to the Campaign Manager now — all 10 SKUs, 50% off, 5 days from today?

Done — the 50% Off Flash Clearance is built and live. Here's what went out the door:

Live now - Sale page: [link redacted] - Promo code: FLASH50 (50% off, valid Jul 24–29) - Collection: all 10/10 SKUs resolved - Banner + hero image: published on the homepage hero - Nav item: "Flash Clearance" added to main menu - Markdowns: all 10 variants set to exactly 50% off with compare-at pricing - Draft campaign: id 17 (email, copy ready)

Two things need a human / follow-up: 1. The email send is a manual step — campaign 17 is a draft; someone needs to trigger the actual send (no auto-send connector). 2. Prices don't auto-revert — the markdowns went live immediately and won't roll back on the 29th. Same for the nav item (not date-gated).

Want me to schedule the post-sale teardown now (restore regular prices + remove the "Flash Clearance" nav item on Jul 29)? And there's a possible duplicate banner on the sale_hero slot — say the word and I'll have it unpublished. Just let me know.

All set on my end — the campaign is fully built and placed, and the only open items are the two follow-ups above (manual send + post-sale teardown).

Just reply with what you'd like me to do: - "Schedule teardown" → I'll queue price restoration + nav removal for Jul 29 - "Unpublish the duplicate banner" → I'll have the Merchandiser drop it - Or nothing if you want to leave it as-is

I'll wait on your call.

I don't have anything more to add — my previous response was complete. The flash sale is built, live, and placed, and I laid out the two open follow-ups (manual email send + post-sale teardown) plus the possible duplicate banner.

The ball's in your court. Just tell me which you want: - Schedule teardown (restore prices + remove nav on Jul 29) - Unpublish the duplicate banner - Leave as-is

Or point me at any other analysis.

I don't have anything to continue — my last message was complete and the flash sale build is fully wrapped.

To recap where things stand so nothing slips:

Live: sale page, FLASH50 code, 10-SKU collection, homepage hero banner, "Flash Clearance" nav item, and 50%-off markdowns — all deployed and verified.

Open items awaiting your call: 1. Manual email send — draft campaign id 17 needs a human to trigger. 2. Post-sale teardown — prices and nav item won't auto-revert on Jul 29. 3. Possible duplicate `sale_hero` banner — can be unpublished if you want.

Tell me what you'd like: schedule teardown, unpublish the duplicate banner, leave as-is, or a new question.

My response was already complete — there's nothing left to continue. The flash sale is built, live, and verified, and I've laid out the three open items awaiting your decision.

To keep it simple, just reply with one of these: - Schedule teardown — I'll queue price restoration + nav removal for Jul 29 - Unpublish the duplicate banner — I'll have the Merchandiser drop it - Leave as-is — no further action

Or hand me a new analysis question anytime.

I will identify customers at risk of churning. First I will check the order date range and find previous buyers who have not purchased for a long time. I am using the latest order date, August 1, 2026, as the reference. I will segment previous buyers who have not shopped in 90+ days.

Here is the churn-risk analysis, using August 1, 2026 as the reference date:

Findings: - 3,145 customers, about 52% of buyers, are at risk: no purchases in 90+ days. - The most valuable target is the 90–180-day group, 1,437 people, averaging $2,215 historical spend. They have lapsed relatively recently and are most likely to return. Average value falls as inactivity lengthens, from $3,852 to $970. - The 365+ day group, 573 people, is likely difficult to win back and averages only $970, so it is lower priority.

Recommendation: Prioritize high-value customers inactive for 90–180 days who have opted into marketing. I can create a segment.

The tool only supports a minimum days-since-last-order threshold, so the actual segment would cover 90+ days without an upper bound: - last_order_days_gte: 90 - accepts_marketing: true - Optional total_spent_minor_gte, for example $500+, to narrow to higher-value customers.

Should I create this win-back segment? Do you want a spending threshold, or are 90+ days and marketing consent sufficient? The tool cannot separately exclude the 365+ day group because it does not support an upper bound; 90+ will include everyone above that threshold.

I am creating the win-back segment for customers who have not ordered for 90+ days and have opted into marketing. Segment created.

Win-Back — 90+ Days Churn Risk, Marketing Opt-In - Slug: winback-90d-churn-risk - Live member count: 1,812 customers - Criteria: No order for 90+ days AND marketing consent.

Verification: The analysis found 3,145 customers across the 90+ day groups (1,437 + 1,135 + 573). Of those, 1,812 opted into marketing and entered the segment. About 58% of the at-risk base can therefore be reached by email. The counts are consistent.

Next step: Win-back campaign. I can set up a campaign targeting this segment. A typical approach: - Offer: A “We miss you” message with 15–20% off or free shipping. - Duration: A 7–10-day urgency window. - Channel: Email, with campaign draft and promo code created by the Campaign Manager.

Would you like that? Specify the discount depth, such as 20%, and duration, and I will prepare the brief for the Campaign Manager. Alternatively, I can first check which brands these 1,812 customers buy from to target the campaign by product as well. Would you prefer that?