Field-Signal Demand Sensing
Build a recovery hypothesis, then instrument three independent signals and act before the data is official
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 75%
The method begins with a hypothesis rather than a dashboard. Brooks had seen running grow through the previous recession, so the team built an explicit frame for how running would recover from the pandemic shutdowns and then instrumented it. Three independent signals fed the frame: third-party activity data from Strava that grew every day after the shutdowns, forty-five field marketing people physically counting runners in high-traffic parks at four o'clock every afternoon, and visibility over eighty-five percent of the company's own retail sell-through. The channel mix shift, from thirty percent of product moving through websites to eighty percent by the end of April, was watched daily. Because the signals agreed, Brooks turned its supply chain back on six to twelve weeks before broad-based retailers who had no clarity on when their customer would return.
Origin
Jim Weber and the Brooks team ran this in March 2020 when European and North American retail shut down in a matter of weeks and ninety percent of Brooks product moved through retailers. Extracted from Acquired.
Core principles
- 01In a shock, the official data arrives too late to act on
- 02A hypothesis about how demand recovers is what turns noise into signal
- 03Serving one narrow customer is what makes a clean demand signal possible at all
- 04Counting people in the real world beats waiting for a report
- 05The advantage is realised in the supply decision, not in the insight
How to run it
- 1
State the recovery hypothesis first
Write down how you believe demand will move through phases, drawing on the last comparable shock. Running grew double digits after the great recession because it is cheap and convenient, which framed what to look for.
Pro tip Use the previous downturn as the prior; behaviour under stress tends to rhyme.
Watch out Collecting data without a hypothesis produces a dashboard nobody can act on.
- 2
Pick three independent signals
Choose signals that fail differently: a third-party dataset, a physical observation, and your own sell-through. Agreement across all three is what justifies acting ahead of the market.
Pro tip Third-party activity data is often published free during a crisis; take it.
Watch out Two signals drawn from the same source are one signal wearing two hats.
- 3
Send people to count in the real world
Put field staff in the highest-traffic locations at a fixed time every day to count actual users. A daily human count is a leading indicator no reporting system will give you in time.
Pro tip Fix the time and the location so the daily numbers are comparable.
Watch out Varying the method day to day turns a trend into anecdote.
- 4
Watch the channel mix move daily
Track where sell-through is happening as distribution shifts. Brooks watched digital go from thirty percent of product to eighty percent between March and the end of April.
Pro tip Compare a month against the same month last year to strip out plan bias.
Watch out Monthly reporting cadence will hide a shift that completes inside six weeks.
- 5
Act on supply before consensus forms
Convert the reading into a supply decision and restart production ahead of competitors who cannot tell which of their many customer types will return.
Pro tip Focus is the enabler here; a single customer type gives you a single signal to trust.
Watch out In apparel and footwear inventory is life and death; too much ruins the next cycle of inline product.
- 6
Move inventory to where demand appeared
Use multi-channel flexibility to redirect stock to the channel the signals identified, because product that is not physically where the customer is cannot be sold.
Pro tip Rehearse the inventory transfer path before you need it.
Watch out A correct demand call is worthless if the goods sit in a closed channel.
In the wild
European retail shut in one week and North America followed, closing essentially every store by the end of March while ninety percent of Brooks product went through retailers and the cash cycle froze. Instead of waiting, Brooks built a frame for how running would recover and instrumented it: Strava activity climbing daily, forty-five US field marketing staff counting runners in high-traffic parks at four each afternoon, and eighty-five percent visibility on retail sell-through. Digital went from thirty to eighty percent of product sold by the end of April, and May 2020 beat May 2019 almost entirely through digital.
→ Brooks turned its supply chain back on six to twelve weeks ahead of broad-based retailers and grew twenty-seven percent in 2020.
Common mistakes
Waiting for clean market data
Official category data in a shock arrives after the supply decision window has closed, which is exactly why broad-based retailers restarted months later than Brooks did.
Instrumenting without a hypothesis
Signals only become decisions when there is a stated frame for how recovery unfolds; otherwise a rising number is just interesting.
Reading the signal but not moving the supply chain
The advantage is realised in production and inventory placement. An accurate call that does not change the order book leaves the upside to a competitor.
Is it for you?
Best for
Operators of focused consumer businesses with long supply lead times facing a sudden shock to demand or distribution.
Not ideal for
Broad-line businesses serving many unrelated customer types, where no single field signal can represent overall demand.
From the transcript
“we created this frame on on how we thought running would recover”
“we put them in parks high traffic running parks at 4 pm every afternoon and they counted runners”
“we turned our supply chain on at least six to 12 weeks before anybody else did”
From the episode
Arena Show Part II: Brooks Running (with CEO Jim Weber)