Channel performance
Shapley attributionRevenue by channel
Attribution model comparison
5 models in parallelROAS under each model. Shapley is the default for every recommendation. Agreement shows how closely the models align.
AI and LLM discovery
Assisted channelReferral revenue and assisted conversions from generative AI platforms
AI platforms often influence a journey that later converts through Organic, Paid Search or Direct. Assisted revenue is reported separately and is not added to paid ROAS.
ⓘ
Estimated impact, not deterministic attribution. Offline effect is measured as digital uplift during each flight against a matched baseline period, with a lag window applied.
Offline campaign library
TV · OOH · RadioDigital uplift during flights
Weekly index, 100 = baseline · last 12 weeks · flights in the selected range are highlighted
What it changes downstream
Offline acts as a demand layer on digital decisions
Share of voice trend
SEMrush weeklyVisibility by channel
Current visibility score vs named rivals · index 100 = market average
Recommended responses
Feeds the Decision EngineWhere closing the gap pays back, and where it does not
Sentiment vs conversion rate
NLP overlayTopics driving sentiment
Entity extractionShare of positive vs negative mentions by theme
Flagged posts
Needs a responseNegative posts linked to a conversion dip in the following 72 hours
Marginal ROAS is the return on the next pound, not the average pound. It is the slope of each channel's response curve at today's spend, and it falls as spend rises.
Marginal ROAS by channel
Curve position shows how much of each channel's achievable revenue current spend already captures
Response curves · spend vs revenue
Reallocation simulator
Move budget between channels and see the modelled effect, after diminishing returns
Start from a scenario, then adjust any channel. Revenue follows each channel's response curve.
Channel budgets
Monthly spend · marginal ROAS updates live
Current vs this plan
Monthly revenue by channel
Scenario comparison
CMO viewAll scenarios on the same model. Click a row to load it.
Monthly revenue projection
Bayesian + gradient boostTrained on 24 months · seasonal adjustment applied · plan uplift reaches full effect in month two
Platform changes
Approval requiredBudgets publish directly to each platform. Every change is reversible.
Audit ledger
Tamper-evidentEvery approval, push and rollback is hashed and time-stamped
Ask the analyst
Live on your dataPick a question, type one or use the microphone
Insight
Impact
Recommendation