Choose your first AI use case differently.
Four gates: does the work repeat, is the data usable, can a human approve the output, would the team be relieved if it worked.
Start with one boring workflow people actually want fixed.
View on LinkedIn →
Start with behavior. Artwork comes later.
Shopper behavior → barrier → intervention → execution. Most activations start at step four.
If the behavior is unclear, the activation is guessing.
View on LinkedIn →
What an AI-first commercial week actually looks like.
Before: reconcile files, build the deck, debate the numbers, chase updates. After: investigate exceptions, sharpen the story, debate the decision, see customers.
Move people from assembling information to applying judgment.
View on LinkedIn →
Pricing starts before the % increase.
SKU role, shopper comparison, promo dependence, retailer margin, competitor response — then, and only then, the percentage.
The number comes last.
View on LinkedIn →
Your AI problem may be a workflow problem.
Collect → reconcile → format → forward → follow up. Buying AI without redesigning that chain preserves the old work.
The tool is often the easy part.
View on LinkedIn →
The GCC fits on one slide. The business does not.
Keep brand direction, financial discipline and KPIs common. Adapt route-to-market, retail concentration, lead times and promo economics by market.
Consistency in direction. Flexibility in execution.
View on LinkedIn →
AI pilots rarely fail at the demo. They fail when the work becomes real.
Four tells: no owner, scope too wide, no human review, wrong problem.
Clear owner + painful workflow + usable data + team demand.
View on LinkedIn →
A forecast can be right. The assumption can be wrong.
Customer orders, promo timing, distribution changes, channel stock — ask what changed before you argue about the total.
Challenge the assumption before debating the number.
View on LinkedIn →
Account planning: automate the 70%. Protect the 30%.
Data pull, reconciliation, standard pages, follow-up — automate. What to ask, what to concede, reading the buyer — keep human.
AI should prepare the room. People still win the room.
View on LinkedIn →
Sell-in can hide the real quarter.
Weeks of cover, sell-in and sell-out together, and a widening gap treated as an alarm.
Healthy growth shows up in sell-out.
View on LinkedIn →
The hidden cost is the wait.
Market changes Tuesday. Report arrives Friday. Meeting next Wednesday. By the decision, the situation has moved.
Shorten the distance between evidence and action.
View on LinkedIn →
The best hire was ranked fourth.
Three polished CVs with strong logos. One candidate who visited the stores before the interview.
The CV gets someone through the door. It rarely tells you if they will win.
View on LinkedIn →
Trade promotion evaluation is built for AI.
Data arrives late, attribution is unclear, the next promo is already booked. Let AI reconcile and flag; let people challenge the mechanic.
AI should improve the evidence. People still make the decision.
View on LinkedIn →
Who does the retailer call first?
Most suppliers bring brand numbers and a growth ask. Category leaders bring a category view, shopper insight and a shelf recommendation.
The strongest supplier brings an insight before an ask.
View on LinkedIn →
14 pages became half a page.
What changed. Likely reason. Decision needed, and who owns it.
A report should narrow the decision, not widen the reading.
View on LinkedIn →
How disciplined is your trade spend?
An eight-point scorecard: buyer-led, managed, or disciplined.
Trade spend is not about spending less. It is about knowing what the spend bought.
View on LinkedIn →
AI can be confidently wrong.
Three mistakes, three rules: data does not contain every relationship; fix definitions before automating; speed is not accuracy.
Use AI fast. Accept its answers slowly.
View on LinkedIn →No published posts in this category yet.