Step 01 · Analytics Engine
Available todayNordkade Fulfilment is a fictional composite company, not a Theovya client. The scenario is illustrative; the industry figures are sourced.
The contract was as good as signed. Nordkade Fulfilment, a Rotterdam 3PL moving about 120,000 parcels a month for sixty direct-to-consumer brands, was renewing its largest client - Brand A, nearly a quarter of revenue - and the commercial director had already conceded a six percent volume discount to get it done. Everyone in the Monday meeting understood the logic. You do not let your biggest account walk.
The CFO had a different feeling about Brand A, and it was only a feeling. The account shipped small, frequent orders that kept splitting across parcels. Its fashion assortment came back through the returns dock in volumes nobody had ever tallied against the account. Its customers seemed to generate a disproportionate share of support work. None of this appeared anywhere, because revenue lived in the order database, freight and returns handling in the warehouse system, and support minutes in the ticketing tool, and no report had ever joined the three.
Consider what answering the CFO’s question used to require. She writes a request. It goes to the analyst, who translates it for the data engineer, who actually knows the WMS schema, who has a backlog. Three weeks later a spreadsheet arrives, built on assumptions two hops removed from the person who asked. Except the renewal deadline was ten days out - so in the old world, the question simply doesn’t get asked. That is the real cost of the management-to-analyst-to-database-and-back loop: not the slow answers, but the questions that die unasked because everyone already knows the answer won’t arrive in time to matter. The discount gets signed on instinct, and instinct in this case was expensively wrong.
Ask it in the room
Nordkade runs the Analytics Engine over its three databases - read-only connections, nothing migrated, every generated query logged. So the CFO asked her question in the meeting, in plain language: contribution per order by brand over the trailing twelve months, net of returns handling, split-shipment freight, and support time at loaded cost.
The answer took less time than the argument about whether to request it would have.
Brand A - first by revenue - was the only account in the portfolio losing money on every order. Minus 1.8 percent contribution, dragged under by a split-shipment rate triple the portfolio average and a returns rate the standard rate card had never been built to absorb. Brands B, C and E, the ones nobody discussed in meetings, were quietly carrying it.
Headline revenue ranked the accounts one way. Cost to serve ranked them another. The renewal price changed.
Composite scenario; margins net of returns handling, split-shipment freight and per-order support cost across OMS, WMS and ticket data.
The renewal still happened. But it happened as a negotiation instead of a surrender: a split-shipment surcharge, a returns-handling fee above a volume threshold, and a discount that reflected what the account actually contributed. Brand A’s procurement team pushed back for a week and then signed, because Nordkade was the first fulfilment partner to show up with the account’s own cost structure on one page. The commercial director now opens every renewal by running the same query for that client. It has become the house style: nobody prices blind anymore.
The habit spread upward from there. When the board debated leasing a second warehouse - a seven-figure commitment justified, in the deck, by “capacity pressure” - the managing director asked the question that had never survived the old three-week loop: pick-station utilisation by hour, by weekday, over the past year, against each brand’s submitted Q4 forecast. The answer showed the pressure was real but pointed: two brands’ promotion calendars colliding in the same three weeks of November. The second warehouse became a negotiated shift in one brand’s launch window and a temporary night shift, at roughly a twentieth of the cost. Nobody had been wrong about the pressure. They had been about to solve it with real estate because real estate was the only lever visible without a query.
That is the layer the Analytics Engine adds for the people at the top of the org chart - the uncommon, unrepeatable questions that decide contracts, capacity investments and client mix, asked directly against the company’s own data, answered while the decision is still open. Transparency isn’t a dashboard nobody opens; it is the ability to interrogate the business the moment a hunch appears.
The standing layer
One floor down, the same engine does quieter work on a schedule.
Roughly one parcel in nine hits an exception somewhere in transit - DigitalGenius puts shipping exceptions at about 11 percent of packages - and the damage is rarely the exception itself. It is the silence that follows, while the customer watches a stalled tracking page and the brand fields the fallout. Nordkade’s ops manager typed one sentence: alert the on-duty lead by SMS whenever a shipment has no carrier scan for 24 hours and its promised delivery falls within 48, grouped by carrier and depot. When DHL’s Utrecht sortation hub jammed on a Tuesday night last winter, the rule fired before midnight. Affected customers had a revised delivery window in their inbox by breakfast; the sixty brands got an FYI before their own support queues filled; the carrier conversation happened Wednesday with a depot-level exception list attached, not at the quarterly review with a grievance and no evidence.
Reactive (status quo)≈36h
- Carrier scan missed
- Customer notices stalled tracking
- WISMO ticket opened
- Agent digs through WMS + carrier portals
- Reply sent
Proactive (Analytics Engine)<1h - ticket never created
- Scheduled no-scan rule fires
- SMS + email to customer; ops alerted
Proactive shipment alerts are reported to cut WISMO contacts by 40–75% (LateShipment; ShippyPro).
The same pattern replaced Nordkade’s most persistent commercial irritant. Every brand onboarded had wanted “just one custom report,” and the part-time BI developer’s backlog had become a line item in sales objections. Now each brand receives a Monday digest - open shipments, exceptions by carrier, on-time performance against SLA - generated from a standing instruction anyone can edit in plain language. Sixty bespoke reports collapsed into one sentence, and the backlog stopped being a topic.
What actually changed
It is tempting to score this in deflected tickets and saved hours, and those are real: proactive exception alerts are the highest-leverage support intervention in the industry, and the digest work alone returned most of a headcount. But the honest accounting is about decisions.
Before, Nordkade’s management made its most consequential calls - pricing, client mix, carrier allocation - on revenue reports, because revenue was the only number that crossed system boundaries without a project. The renewal story would have ended with a six percent discount on a money-losing account, and no one would ever have known, because the loss was distributed across three databases in amounts too small to see individually. The alerting layer has a parallel logic: a jammed depot was always going to jam; the difference is whether the company learns it from its own data at 23:40 or from angry customers at 10:00.
And the BI developer did not become redundant. He stopped being a human pipeline between databases and management and started doing the work he was hired for - this quarter, a carrier-allocation model that the Analytics Engine’s ad-hoc layer made cheap to validate, projected to cut linehaul cost by a low single-digit percentage that dwarfs his salary. The pattern the leaders proved holds at every scale: Otto, the German e-commerce group whose autonomous data-to-action loop is documented by Harvard Business School’s Digital Initiative, spent a decade and an in-house data science division wiring decisions directly to operational data. The wiring is the advantage. The Analytics Engine is the wiring, without the decade.
What Nordkade bought, in the end, was not analytics. It was the collapse of the distance between a question and its answer - for the CFO with a hunch in a Monday meeting, for the ops lead who wants to know before the customer does, and for the analyst who finally gets to analyze. The renewal that almost signed itself is the version of the story with a happy ending. The uncomfortable question for any operator reading this is how many Brand A’s are sitting in the current book right now, profitable on the revenue report and nowhere else - and whether anyone in the building can afford to find out before the next renewal can’t wait.
Sources
- DigitalGenius (via ReadyCloud, 2026): ~11% of parcels experience a shipping exception.
- Harvard Business School Digital Initiative (RCTOM): Autonomous Stock Replenishment at Online Retailer OTTO.