For MarketingThe patterns
The patterns
your dashboards never built.
Your warehouse already holds ad spend, product events, and revenue. The cross-cuts nobody had time to pull — the ones that flip the story, the segments no dashboard renders, the friction that sits between click and revenue — are sitting there, unasked. DecisionBox runs them.
CRITICALSupply concentration
169 hosts (4.4% of the base) produce 31% of revenue — and the top 10 are all in a single country the campaigns don't target
31% of revenue · $8.18M from 169 hosts
Vacation-rental platform · top 10 UAE-based
What it answers
Five questions your dashboard doesn't
01
Which acquired cohorts quietly refund within 90 days — and what was different about the creative that brought them in?
Refund records backed out to the original acquisition event. Shows which campaigns bring in users who feel good now and refund later.
02
How much of our paid acquisition pushes buyers toward inventory that isn’t actually available most of the year?
Supply-side availability joined to demand-side spend. Exposes the share of budget matched to catalog that doesn't clear.
03
Is our top-revenue cohort concentrated in a geography or trait we aren’t deliberately marketing to?
Revenue cross-cut by dimensions your attribution report doesn't render. Catches revenue coming in from somewhere you didn't aim at.
04
Of the users who didn’t convert, how many never even saw the offer? Our funnel only counts the second group.
Impression-side cohort built alongside the conversion-side. Splits "didn't convert" into "didn't see" and "saw and didn't act."
05
Is a slow post-click step hiding a conversion cliff our funnel ends before?
Walks the time-to-event between click and first revenue. Flags friction points that happen after marketing's last dashboard.
From real runs
From a pattern to a playbook
Every finding the agent produces comes paired with a recommendation — a target segment, an expected impact, a concrete set of actions. Three real pairs below, pinned as the agent surfaced them.
HIGHDead inventory
7,068 listings (39% of supply) generate zero revenue 10 months of the year
39% of supply · dark off-peak
Vacation-rental platform
ship this one first →
P3Recommendation
Demand activation
Launch an 'Off-Season Escape' campaign: 40–60% discounted rates on the 7,068 dormant listings for Oct–May stays
Expected impact
+0.3–0.8pp occupancy · $500K–$2M annual revenue
Target: 7,068 properties6 concrete actions
HIGHAudience blindness
Of 1,637 users shown a rewarded ad, only 810 ever registered a response. Half the addressable base is ad-blind — the spend is chasing the wrong half
50.5% zero-response · 827 users never signal
Casual mobile game · 30-day window
P1 · quick win
P1Recommendation
UX cap
Hard-cap rewarded-ad and booster offers to 3 per session
Expected impact
+10–15% conversion
Target: 20,110 users4 concrete actions
HIGHHidden friction
24,445 items (22% of catalog) have freight exceeding 30% of total. For 4,124 of them, freight is more expensive than the product itself
22% of catalog · freight > 30% of total
Marketplace · 4,124 items freight > product
needs finance sign-off ~
P3Recommendation
Pricing intervention
Cap customer-visible freight at 25% of item price for sub-R$50 products; marketplace subsidizes the delta
Expected impact
+5–10% GMV in sub-R$50 segment
Target: 24,445 items5 concrete actions
Every pair above is from a real DecisionBox run. Source datasets are anonymized for the public site.
The honest list
What it does, and what it doesn't
What it does
- Finds cross-cuts of channels, campaigns, and cohorts your team never had time for.
- Runs against the attribution already in your warehouse — whatever it already is.
- Produces recommendations with target segments, effort estimate, and expected impact.
What it doesn't
- Pull from ad platforms directly — your warehouse has to hold the data.
- Run media-mix modeling from scratch. Uses the attribution that already lives in your data.
- Replace a marketing analyst. It removes the "which slice should I look at today?" question.
app.decisionbox.io/insights/marketing
Keep reading