Shipments SANOMI
Every purchase order from Alibaba and the forwarders, from deposit to warehouse. Several products per shipment; freight and customs are spread over them by value so every product gets a real landed cost.
| ID | Supplier / products | Status | Mode | Pcs | Goods | Landed | Shipped | Expected | Tracking |
|---|
CRM
B2B pipeline: kindergartens around Znojmo, electronics retailers and the hot leads. The buyer in a school is the ředitelka; in a chain it is the purchasing desk. Status is logged on the same screen you call from.
| Ref | Company | Who to call | Contact | Stage | Next step | Value |
|---|
Paid
What went into Meta, Google, Sklik and Heureka, what the platforms claim came back, what the shop actually booked, and profit per product after purchase cost and ads. The shop is the truth; platform numbers are shown so you can see how far they drift.
By channel spend ex-VAT · conversions and value are what each platform claims
| Channel | Spend | Share | Clicks | CPC | Claimed sales | Claimed value | Claimed ROAS |
|---|
Verdict lines from Settings · break-even moves with the product mix
Day by day orders and revenue from Shoptet · MER = revenue ÷ all-channel spend · break-even = 1 ÷ that day's gross margin
| Day | Orders | Revenue | Margin | Meta | Sklik | Heureka | Spend | MER | Break-even | Cost / sale | Contribution |
|---|
Tick days to total them together · Shift+click a box takes a range · the header box ticks every day.
Tick the boxes to pick products, then edit them all at once. Shift+click a box takes the whole range; the header box ticks everything in the list.
| SKU | Product | Qty | Revenue | Margin | Ad spend | Clicks | CPC | Claimed sales | ROAS | 7D | 14D | 30D | Net | POAS | Tags |
|---|
ROAS here is shop revenue ÷ product ad spend, not the platform's number. Product ad spend is what Google Shopping / Performance Max, Meta catalog ads and Heureka attach to a product id; Search, Flexible Meta ads and Sklik cannot be attributed to a SKU and sit only in the account totals above. Net = revenue − purchase cost − product ad spend. POAS = net ÷ ad spend.
Platforms claim vs. the shop booked each platform counts the same order for itself
How much of each channel's spend reaches a product the rest is real spend that no per-SKU table can show
| Channel | Spend | Attached to a product id | Coverage | Why not 100 % |
|---|
Product ids that did not match a Shoptet code spend counted in the channel total, missing from the product table — map them and the next sync fixes it
| Channel | Platform id | Spend | Clicks | Map to SKU |
|---|
Tick the ids to map several of them to one product at once.
Purchases SANOMI
Where we bought each product, and who quoted it. Suppliers are ranked by landed price per piece in EUR — goods plus the supplier's shipping plus freight and duty to CZ once those are in Shipments. Lead time runs from payment to arrival at the agent. Ordered rows read their numbers from Shipments.
Returns
Every OLENOVA return in one list: the customer, the order (opens in Shopify), the return tracking number (opens the carrier or 17TRACK), where the package is, the refund, and the Consumer Care case. Type the order number and the case, name and e-mail fill in by themselves. The arrow button moves a return one step and stamps the date.
Suppliers
Every company we buy from and every agent we ship through, in one list. It fills itself: a new supplier name on a shipment or a quote creates the row, and an agent on a shipment or consolidation is marked as an agent. Open a row to add contacts, links, terms and notes, or jump to its shipments, consolidations and purchases. Orders, pieces and goods value come from Shipments; quotes and quality come from Purchases.
Labels
Every product and variant from Shoptet with its photo, SKU, EAN, weight and size. Press Label to download a print-ready PDF with the barcode. Shoptet stays the master: fix a missing weight, size or EAN there and it shows up here after the next sync. Where Shoptet has no weight or size, the box values from the portfolio Excel are used and marked Excel.
Bills
Every OLENOVA order as an invoice of MEGALU RWS, s.r.o., and every cancelled order as a credit note. Pick what goes to the accountant, export it as Pohoda XML, and find every past export with its files in the history.
Campaigns
The product timing framework: what to push in each quarter for fashion, home & decor and general products, with dates for the northern hemisphere (US / UK / EU) and the southern hemisphere (AU / NZ / BR / ZA), where the seasons are six months apart. Calendar dates like Valentine's, Black Friday and Christmas stay the same everywhere. Colours follow today's date: green is live now, yellow is next up, grey has just ended.
AMZ Finder
Products worth selling on Amazon.nl. Demand, prices and the long tail come from Amazon's own Product Opportunity Explorer, read by the PC job. The profit check uses our import rules: freight per kg from China, Amazon's fee, VAT and the parcel.
ALB Finder
Products MEGALU could import from Alibaba and sell on megalu.cz. Demand is Google Keyword Planner for Czechia, the shelf is Zboží.cz read live, the factory price is Alibaba's public search, and the money is our import rule: freight per kg from China, duty, 21% VAT and gross profit per sale against what an order costs in ads.
Creative performance
Every MEGALU creative: when it was posted, how it did organically, when it ran in paid ads and how it did. Verdicts follow the vault bar, nothing is called a winner below it. Click a card for posts, ads and rights.
Creative types
The funnel grid: brands that win run TOFU, MOFU and BOFU ads at once. Five levers change from stage to stage: format, brand mention, product visibility, message and CTA (the call to action). One real example per cell.
Other brands' ads, study only, never reuse.
Video effects
Every effect we can put on a video, one row each, so a brief can name it and a builder can make it: camera moves, framing, transitions, speed, graphics, captions, looks, AI transforms and presets, compositing, sound sync, 3D, lenses and UGC editing idioms. Each row says what it does for the viewer, how we make it and with which tool, what it costs in credits, which funnel stage it fits, where our winning ads already use it, and plays an example (a preset preview, a tool sample, a public clip at the right second, or one of our own renders). An effect marked "available now" that fails on first use is flipped to "needs test" here and the row is kept.
Business validator
Our MiroFish simulations: AI agents play shoppers, competitors and media around one seed document, then a report says what spread and what broke. Each run shows every step (seed, knowledge graph, agents, the simulated feed, report, interviews, transcript) and ends with a two-point verdict quoted from the report. Speculation, not evidence: use it as a list of questions to test on real data.
Studio
Order a video from menus, Claude builds it, finish it together on a timeline. A card moves Brief to Edit to Review to Published. The brief reads the ad type from Creative types, the reference reels and competitor videos from Swipe analysis, every effect example as a clip cut from those reels, and our own footage database with the licence on every clip.
Agents
Agents that watch OLENOVA and prepare work for your approval. Nora reads the product feed twice a week and proposes draft, tag or exclude actions. Nothing changes on the store until you approve it and press apply.
Feed signal thresholds
Configure the spend and ROAS boundaries for each signal. Editing a value in any row updates all rows that share that boundary.
Latest run
Proposals
Run history
Swipe analysis
Competitor video ads that have run for months, torn down beat by beat: why each one works, what we cannot copy, and the principle to reuse. Click a card for the full teardown.
P&L
OLENOVA · daily ledger, pulled from Shopify and Google Ads
Summary
OLENOVA at a glance. Click a figure or a chart to open the tab behind it.
Refunds
Every refund since March with its reason, and refunds by month, reason, product and size. Same figures as the Refunds sheet in the Excel P&L.
Refund drivers
Seven questions about what causes refunds and what they really cost.
Paid
Google Shopping spend against Shopify sales, per product. Rule = feed review (FA 7.3): judged once it spent the target cost per sale ($30), target ROAS 2.5.
How the rule decides · live products only · read top to bottom, the first line that matches wins
Nothing is drafted automatically. The rule only proposes. Every draft, optimise and watch tag waits for Ole's OK, then you press Execute. At most 7 drafts a day. The review runs every 14 days while the store sells up to $3,000 in 30 days, every 5 days above that. A draft is never a delete: the product goes to Draft with a date tag and leaves the Google feed.
- 1Under the spend line or paying back, still selling, and its season is about to turn (peak in the last 3 months, none this month or the next two)Watch tag
- 2Spent less than the target cost per sale in 30 days: $30 (Ole, 5 Nov 2026)Under spend line, not judged yet
- 3Pays back: ROAS 2.5 or more on Google or on Shopify sales. At twice break-even with 2+ orders it is a best sellerKeep
- 4From here on the product spent $30 or more in 30 days and is under target. Optimised less than 14 days ago, or repriced recently (hold list)Waiting
- 5Searches falling: down 20%+ in the last 3 months against the 3 before (a peak month leaving the window does not count)Draft now
- 6Evergreen: steady all year, peak month at most 1.5× the average monthOptimise, never draft
- 6bEvergreen, already optimised, and the price is at the 30%-cost floorCheck: Ole decides
- 7Season turning (or tagged Watch) and now under targetDraft
- 8Out of season: no peak this month or the next two, and not risingDraft
- 9Optimised 14+ days ago and still under targetDraft
- 10Anything else: in season or rising, never optimisedOptimise
Optimise = lower the price $1 to $5 (never under the 30%-cost floor) and add lifestyle photos in slots 2 to 10. Photo 1 never changes. Tag OPTIMIZE<date> starts a 14-day clock. Season comes from the product's own search phrase in Keyword Planner. When only the category tag is known, the answer is Check, never an automatic draft. Source: Freedom Achievers lesson 7.3 + Ole's rules of 2 and 4 Nov 2026 (.claude/rules/olenova-feed-drafting.md).
How the numbers are built
Spend, clicks and impressions come from Google Ads account 101-040-3041, split per product by the Shopping feed id. Revenue, orders and refunds come from Shopify and are already net of refunds. Cost per product is estimated from the store-wide COGS share of revenue until you type the real one into the Cost column - a typed cost always wins. Break-even ROAS is calculated from the same ledger the P&L tab uses, never typed. Google Trends compares the last 4 weeks against the 8 before, US, on the garment keyword. AOC is ad spend divided by the product’s selling price — 3.26x means we spent three and a quarter times the ticket price on ads for it. Hover it for the same figure against the price of the units actually sold. It rests on the price, which we know, rather than the estimated cost. It is a read, not one of the rules. Changing a Decision records it - together with the recommendation that stood against it and the numbers behind both - in the decision log.
Competitors
Every store we watch, refreshed once a day by the PC job without anyone touching it. Monthly visits come from hypestat.com (the public SimilarWeb estimate). Product counts come from each store's own product feed, and the yellow number is how many new products appeared since the last run. The arrow says whether traffic is up more than 10 % against last month (scaling), down more than 10 % (slowing), or in between (the same). Open a scaling store to see its top 20 best sellers, each with a Google Trends line (US, 12 months): rising, falling or steady, and the quarter the search peaks in.
Product timing
What to list and push in each quarter, per hemisphere. The AU / NZ set is the framework as written; the US / UK / EU set is its six-month mirror (their winter is our list's summer). Green is the quarter we are in now, yellow is the one coming next, so "list now" is the yellow column and "push now" is the green one.
Lister
Products picked from competitors, waiting to go live on OLENOVA. Press + on a competitor's product and it lands in the open batch. The price starts at 0.9 × the competitor's price and compare-at at the catalogue ratio; change them here, then press "List on OLENOVA". The PC job picks the batch up within 30 minutes, runs the listing chain in OLENOVA's brand template (scrape, size chart, house copy, create, publish) and reports each product back here as "photos pending". The house-model reshoot is the last step and runs from a Claude session with the Magnific connector ("reshoot batch B-001"); the product is done when its photos are on the brand cast. No size chart = draft, never invented sizes.
Shoots
Step 2 of the Lister. A batch that has been listed sits here on the supplier's photos, as a draft nobody can see. Press "Generate shoots" and the batch is queued for the house-model reshoot on the OLENOVA cast: five frames per product plus one per colourway, then the frames go onto the product in Shopify and the title gets the "(DONE)" marker. That marker is the only thing the 04:00 job looks for, so a product is in the queue to go live only once its photos are ours. The reshoot needs the Magnific connector, so it runs from a Claude session ("shoot batch B-001"); this page shows each product move from pending to shot to marked.
Quotations
Every product on sanomi.co with its photo, SKU and vendor. The vendor comes from the Portfolio spreadsheet, else from the Portfolio folder name; NA means nobody has written one down yet. Type the quantity you can order next to a product, it saves as you type. Download PDF on a vendor gives you an order sheet with pictures, titles, SKUs, supplier codes and quantities to send to that vendor. Download Excel gives the same sheet as an .xlsx, with an empty unit-price column the vendor can fill in. Shipping and consolidation tabs like MEGALU's come next.
Labels
Every product on sanomi.co with its SANOMI SKU, the supplier’s SKU and its EAN. Press Label for a print-ready PDF at 600 dpi. Two shapes: the full label is MEGALU’s, with the SANOMI logo, and the 30 × 20 mm jewellery tag carries only the SANOMI SKU, the supplier SKU and the barcode. SANOMI had no barcodes, so every EAN here was taken from MEGALU RWS’s own GS1 pool — the same Portfolio sheet the MEGALU labels draw on, with the SANOMI SKU written back into it so no code is ever handed out twice.
Store check
Google Merchant Center readiness. Crawls the live store the way a reviewer would: links, contact details, policies, products, collections. Every finding comes with the page, the evidence and the fix, sorted into what blocks approval, what weakens the review and what is housekeeping. Runs on the server, about 2-4 minutes.
Projekt
AP · A/B testy
Testy prvků na www.aretacni-pripravky.cz — detail produktu, homepage, tlačítka, cenový řádek. Test se založí tady, e-shop si ho do 10 minut načte, data (zobrazení, košík, objednávky, scroll, čas, kliky) tečou zpátky sem. Vítěz se nevyhlašuje před 14. dnem a bez dostatečného vzorku.
Paid · Feed
MEGALU ad spend per product across Google, Meta, Seznam and Heureka against what megalu.cz actually sold, for up to 24 months. Where ads earn, what is falling and where spend brings no sale, with a recommendation and a note on every product.
AP · Paid
Google Ads spend vs. tržby e-shopu artacni-pripravky.cz po produktech, za 24 měsíců. Kde reklama vydělává, co klesá a kde spend nepřináší prodej — a ke každému produktu doporučení s poznámkou.
AI
Ask anything. It runs on this PC when our router is on, otherwise on Cloudflare. Questions and answers are never saved.
Router on this PC
Checking
When the router runs on this PC, questions go through it and provider keys never leave the PC. The model named private is only ever answered here.
The key is the ROUTER_API_KEY line in the .local-ai-router folder of your Windows user. It is kept in this browser only and is never sent to Cloudflare.
Settings
Who can sign in (global), then per project: where the numbers come from, sync, and the lines each page judges against.
| Role | Modules | Sees prices | Added |
|---|
After adding someone: send them https://is.megalu.eu. They click Create account, use exactly this e-mail, choose a password and confirm the e-mail Supabase sends. Until then the row simply waits. Sees prices off hides purchase prices and the Paid tab — for warehouse hands. Modules decide which tabs a person sees and which tables the database lets them read or write; the owner always has all. Consultancy clients are one module each — for an AP team member tick only Consultancy · AP; they will not see other clients.