A number is only useful if you know where it came from
Our approach to inventory work rests on a small set of beliefs about how forecasting should sit inside a retail business.
Back to homeOur foundation
The people who run a retail floor understand it in ways no model fully captures. We start from that premise. Everything we build is meant to sit next to that understanding, not above it — a second set of eyes on the numbers, offered plainly and left open to disagreement.
Philosophy and vision
Retail inventory decisions are made under uncertainty, and pretending otherwise does not serve anyone. We think the more useful contribution is not a single confident number but a clearer picture of the range of likely outcomes, so the person making the call can weigh it against everything else they know.
A few core beliefs
Assumptions should be visible
A forecast without its reasoning attached is hard to trust and harder to question. We write the reasoning down.
Local calendars matter
Holidays, seasonal changeovers, and regional event patterns in Japan shape demand in ways a generic model misses.
Some things resist forecasting
New launches and irregular, event-driven items are named honestly as areas where confidence should be lower.
Simplicity holds up better
A method your team can understand and adjust tends to outlast one that only makes sense from the outside.
Principles in practice
In practice, this means every forecast we deliver comes with a margin of error rather than a single figure, every reorder review documents the assumptions behind its recommendations, and every engagement has a defined scope so there's no ambiguity about what was and wasn't examined.
A human-centered approach
Two retailers with similar sales figures can have very different constraints — storage space, staffing, supplier relationships. We ask about those constraints before proposing an approach, and the output is shaped to fit the range and the team using it, not the other way around.
Innovation through intention
We adopt new methods when they demonstrably improve on what we already use, not because they are new. Established statistical methods are combined carefully with newer techniques where the latter add something the former cannot.
Integrity and transparency
We do not claim more accuracy than the data supports, and we say so directly when a category is a poor fit for forecasting. The caution bay on our home page exists specifically for this reason — to name limitations rather than gloss over them.
Community and collaboration
The review session at the end of each engagement is treated as a conversation, not a handover. Questions and pushback are expected, and often improve the output more than any additional data would.
Long-term thinking
A forecasting setup or reorder table is only as useful as its upkeep. We write our notes so they can be revisited and adjusted later, whether or not that revisiting happens with us.
What this means for you
In practical terms: you will receive figures with their reasoning attached, honest notes on where those figures are less reliable, and a defined piece of work rather than an open-ended commitment. What you do with the output remains your decision throughout.