Category
221 articles in this category · page 3 of 19
The vast majority of marketplace sales go through one box that an algorithm fills. That makes your sales data a record of ranking decisions, not customer preference - and the two can move in opposite directions. Here is how to tell them apart.
Koji Team · August 15, 2026 · 11 min read
AI-sourced traffic to retail is growing fast, but the first generation of in-chat checkout has already been pulled. Here is what the 2026 evidence actually shows, why agent-mediated buying destroys every impression-based shelf metric, and the one research move that still works.
Koji Team · August 15, 2026 · 12 min read
Market basket analysis tells you which products travel together. It never tells you why - and four completely different shopper realities produce identical affinity scores. Here is how to read co-purchase data honestly, and how to recover the missing why.
Koji Team · August 14, 2026 · 9 min read
The average US shopper visits 5.4 grocery banners a month. Every dataset you own is bounded by one of them. Cross-shop analysis is not a hard measurement problem - it is a problem about who owns the sensor, and the sensor decides what counts as the universe.
Koji Team · August 14, 2026 · 10 min read
Most segmentation sorts people. But the same person runs a stock-up, a fill-in and an immediate-need trip in the same week, behaving differently on each. When the variation within a person exceeds the variation between people, your persona is an average of incompatible occasions.
Koji Team · August 14, 2026 · 9 min read
The FTC found intermediaries setting individual prices from mouse movements and abandoned carts, working with at least 250 clients. Academic mystery shopping finds personalisation in only 6% of cases. Both matter, and together they break the assumption underneath all price research.
Koji Team · August 13, 2026 · 13 min read
A decade of NielsenIQ scanner data shows shoppers barely react to downsizing. Five preregistered experiments show they consider it more unfair than an equivalent price rise. Both are true, and the gap between them is where the real risk lives.
Koji Team · August 13, 2026 · 12 min read
Price pack architecture assumes the ladder you design is the ladder the shopper sees. Journal of Marketing evidence shows retail price setters invert it on your best-selling size 79.1% of the time. Here is how to research a ladder that survives contact with the shelf.
Koji Team · August 13, 2026 · 11 min read
In many US grocery categories a leading manufacturer advises the retailer on assortment, shelf and promotion for the whole category, including your products. Here is the one evidence class the category captain does not already own.
Koji Team · August 12, 2026 · 13 min read
Store brands hit a record 23.8% unit share in 2026, and 94% of shoppers say they will not switch back even if prices fall. Share data proves the switch happened but cannot explain it. Here is the research design that can.
Koji Team · August 12, 2026 · 12 min read
The famous claim that 74% of a promotion bump comes from brand switching is a units error. On a unit-sales basis it is roughly 33%. Most of your lift is your own buyers buying earlier and buying more.
Koji Team · August 12, 2026 · 12 min read
Retail media networks sell the ads, define the attribution window, own the transaction data and report the result. Here is what the IAB/MRC guidelines actually require, why cross-network ROAS is incomparable by construction, and how to test what the ads really did.
Koji Team · August 11, 2026 · 13 min read