Your weekly Shopify action plan.

Every Monday morning we scan your Shopify store and surface a ranked list of specific actions to take that week: restock alerts, win-back campaigns, pricing nudges, ad creative briefs, cash-on-delivery risk (if you take COD), and on Plus, funnel-aware insights powered by a privacy-safe Web Pixel. How many you get depends on your plan: 1 a week on Free, 5 a week on Pro, 10 a day on Plus. Each recommendation has a clear next step, a sized estimate of what it's worth per month, and a deep link into your Shopify admin. We never write to your store; you make the change yourself.

Each rule does its own arithmetic against your real store numbers, and we show you that arithmetic on the card. Three guardrails keep it honest. First, every estimate is a range with a confidence level, never a single figure. Second, no single opportunity may claim more than 20% of your trailing 30-day revenue, so a small store never gets told it's leaving five figures on the table. Third, when a signal is real but we can't size it defensibly, the recommendation ships without a number rather than with an invented one. Low-confidence estimates appear on their own card but never roll into your headline total.

A single 0–100 number backed by seven sub-scores: revenue, conversion, products, customers, marketing, inventory, and delivery. Anything we can't genuinely measure on your store is marked unrated with the reason (for example, marketing and conversion need visitor tracking; customers needs at least 10 buyers with an order), and your total is the weighted mean of whatever is actually measurable. Movement is capped at 5 points per snapshot so one sold-out product can't swing the number.

Yes. When you mark something done, we re-read the metric behind it 14 and 28 days later and compare it against an equal-length window before you acted. Slower fixes get 28 and 56 days instead: restock (supplier lead time), win-back (a lapsed customer returning isn't a same-week behaviour), and cash-on-delivery (orders have to travel out and back before they settle). Windows are always whole weeks so day-of-week seasonality cancels out. We report this as what happened after you acted, never because you acted, we can't run a holdout on a single store and won't pretend otherwise.

Shopify gives you the data. We give you the to-do list. Instead of 50 charts, you get 5 ranked things to act on this week, phrased in plain English, with the math already done.

No. Our architecture splits responsibility: a deterministic rules engine reads your store data and decides what to recommend; the LLM only rewrites the language. Numbers and product names always come from your store, never from the model.

Three: Free ($0), Pro ($19/mo), Plus ($39/mo). Free is your at-a-glance store health check: 1 recommendation per week in your dashboard, no email. Pro turns Monday morning into a ritual: 5 ranked recommendations land in your inbox every Monday at 9am local time, plus full action history with one-click reactivation. Plus is for merchants who run their store daily: 10 fresh recommendations every morning at 9am, ready-to-send draft emails, and 4 funnel-aware rules powered by the opt-in visitor analytics pixel.

Yes. The rules engine that decides what to recommend is identical on every plan, Free, Pro, and Plus all get the same math. What changes is the model we use to polish the write-up: Free runs on fast, cost-efficient models to keep the plan free to offer; Pro and Plus route through the latest Claude, ChatGPT, and Gemini models for sharper, more natural copy. Either way, numbers and product names always come from your store data, never the model, see "Does the AI ever invent numbers or product names?" above.

Yes, 7 days on Pro and Plus. We pass trialDays to Shopify Billing explicitly, so the trial period is enforced by Shopify itself. You won't be charged until day 8.

Through Shopify Billing. Your DropifyXL subscription appears on your existing Shopify invoice, no separate Stripe account, no second credit card form. Cancel from your Shopify admin in two clicks.

Any time. The pricing page in-app routes through Shopify's standard plan-change flow. Plus → Pro downgrade also automatically removes the visitor pixel from your storefront.

Yes, and it costs you nothing to find out. Four cash-on-delivery rules switch on automatically once we detect COD in your order history, and stay completely silent if you don't take it. They cover: what dispatched-and-refused orders actually cost you each month, which unfulfilled COD orders to check before you ship, which buyers repeatedly refuse deliveries, and how much of your risk a prepay incentive would move. All four are available on every plan, including Free, because a merchant losing money to failed deliveries deserves the diagnosis rather than a paywall in front of it.

Shopify has no cash-on-delivery flag, so we match against the payment gateway names on your orders, both Shopify's manual payment methods and the regional COD apps. We match confidently or not at all: 'manual' is deliberately excluded because plenty of stores use it for bank transfer or in-person pickup, and telling you your prepaid orders are failing would be worse than staying quiet. The COD rules also stay silent until COD is at least 5% of your volume and you have at least 20 COD orders, so a one-off never triggers them.

A ranked list of your unfulfilled COD orders, scored on the patterns that precede a refused delivery on your store: order value against your average, the buyer's own COD history, repeat orders inside 48 hours, unusual quantities, and time of day. Each order shows why it was flagged and deep-links into your Shopify admin. It's read-only by design, we don't cancel, tag, or hold fulfilment, which is why it ships without asking for write access to your orders. Available on Pro and Plus; Free sees the counts.

The strongest cash-on-delivery signals are phone validity and delivery-address quality, and both are Shopify protected customer data that we don't currently hold approval for. Rather than scoring those as clean, which would silently mark risky orders safe, those signals abstain and the page tells you what percentage of the picture you're seeing. The moment Shopify grants those fields, the same page sharpens on its own with no change on your side.

Five standard Shopify events: page_viewed, product_viewed, product_added_to_cart, checkout_started, checkout_completed. For each event we capture timestamp, anonymous session ID (cookieless, daily-rotating), first-touch source (UTM or referrer), product ID where applicable, page type, and device class. No PII, no full URLs, no IP storage.

Raw events live in our database for less than 36 hours. A nightly job rolls them up into anonymized daily summaries (per source / per product / per page type) and DELETES the raw events. We never accumulate visitor-level history.

Yes. The pixel checks Shopify's Customer Privacy API on load. If the visitor hasn't consented to analytics, we never send anything. There's also a per-shop kill switch in Settings (defaults on for Plus, off for everyone else).

Seven scopes, five of them read-only: read_products and read_inventory (catalog and stock levels for the restock rule), read_orders and read_all_orders (order history), and read_customers (for the win-back rule). The two remaining are write_pixels, which only lets us install our own analytics surface on your storefront, and read_customer_events, which lets us receive the events that pixel emits. Both stay dormant until you enable visitor analytics on Plus. We request no write access to your products, orders, customers, or theme, every recommendation deep-links you into the Shopify admin to make the change yourself.

Shopify's standard read_orders scope only exposes the last 60 days. Two of our rules need more than that: the win-back rule looks at a 60–180 day window to find customers who have genuinely gone quiet (60 days alone can't tell you that), and the pricing rule needs a longer baseline to distinguish a real margin trend from a slow week. read_all_orders is the only way to read past that 60-day wall. It's read-only, the same as read_orders, just without the time limit.

No. This is enforced in the architecture, not by policy. The rules engine produces two separate things: structured numbers, which go to the model to be written up, and references to the specific customers involved, which never leave our database and are joined back in only when your dashboard renders. There is no code path that puts a customer name, email, or phone number into a model prompt. Contact lists are also masked by default in the dashboard, so a screen-share doesn't leak them, and unmask one row at a time only when you click.

Shopify approves protected customer fields individually. We hold approval for name and email; we do not read phone numbers or addresses. Where a feature would be sharper with a field we can't read, it abstains and says so rather than guessing, which is why some cash-on-delivery risk signals show as unavailable.

Postgres on Supabase. Your data is isolated per shop. The pixel ingest endpoint uses per-shop bearer tokens and rate limits (2000 events/min/shop hard cap).

Under 5 minutes for the install + OAuth handshake. Your first scan completes within ~30 minutes for most stores; very large catalogs may take longer and queue automatically.

Right after the initial scan completes, you receive a 'first action plan ready' email. Then weekly on Mondays at 9am store-local (Pro) or daily on Plus.