Model budget shifts before you commit the spend.

Forecast the scenario: "what if we shift 20% of paid search budget to LinkedIn?" Feed in 24 months of weekly spend and lead data across six channels; the model predicts MQL impact with 84% historical accuracy. Allocation decisions are data-backed in hours, not weeks.

Run Forecast
AuthorMariete
AgentForecast
Runtime~ 4 minutes
IndustryGrowth
  1. 1

    Describe the task

    Forecast can answer the question every allocation debate stalls on, "what happens to pipeline if we move budget?", before you spend a pound. Feed it 24 months of weekly spend and leads across your channels, describe the shift, and the modelling node predicts the MQL impact at 84% historical accuracy, so a reallocation that used to take weeks of arguing becomes a data-backed call in an afternoon.

    The more precisely you frame the scenario, the more decision-ready the answer. Tell Forecast which channels to model, the exact shift to simulate, and which metric to predict, and it returns a side-by-side of the base case and the scenario with a confidence band on the difference.

    Model this scenario against channel-spend-leads-24mo.csv: "Shift 20% of paid-search budget to LinkedIn starting next quarter." - Fit each channel’s spend-to-MQL response from the last 24 months - Simulate the reallocation and predict the change in MQLs per week - Account for diminishing returns, LinkedIn shouldn’t scale linearly forever - Return base case versus scenario with a confidence band on the delta End with a recommendation: net MQL change, cost per MQL before and after, and how confident the model is.

  2. 2

    Give Forecast context

    In a Forecast workspace, load your channel spend-and-leads history into a source node and describe the scenario on the modelling node. Forecast fits a response curve per channel, so it knows paid search saturates differently from LinkedIn, then re-plays the reallocation against those curves rather than assuming a lead costs the same everywhere.

    Make sure spend and leads are attributed to the same channel taxonomy. If leads are tagged by a different scheme than spend, the response curves fit to noise and the scenario’s confidence collapses.

    Required context
    A weekly channel history with spend and leads per channel over enough time to fit a curve
    A consistent channel taxonomy shared by both the spend and the lead data
    Optional context
    An MQL-to-opportunity conversion rate, so the output extends past MQLs to pipeline value
    Per-channel saturation notes, so the model knows where a channel is already near its ceiling
    A budget constraint, so Forecast only returns scenarios that fit the total you can actually spend
    channel-spend-leads-24mo.csvchannel-taxonomy.csvconversion-rates.csv (optional)
  3. 3

    What Forecast creates

    Forecast fits a response curve per channel and replays the reallocation against them. You get a base-versus-scenario comparison of predicted MQLs, the change in cost per MQL, and a confidence band that tells you how much to trust the delta.

    From Forecast: Simulated a 20% paid-search-to-LinkedIn shift against 24 months in channel-spend-leads-24mo.csv.

    +11%Predicted MQL change
    84%Historical scenario accuracy
    −£9Cost per MQL delta
    Scenario vs base case6 channels
    ChannelBase MQLs/wkScenario MQLs/wkDelta
    LinkedIn84128+44, before diminishing returns
    Paid search210176−34, spend reduced 20%
    Net across all channels612682+70 (+11%)
    Confidence and caveats2 flags
    FactorReadNote
    LinkedIn saturationMediumReliable to ~£40K/wk, thin history above that
    Attribution windowHighBoth channels use the same 30-day model

    "The +11% holds up to about £40K/week on LinkedIn, above that the history thins and confidence drops. Want me to cap the shift at the reliable range, or model a phased ramp instead of the full 20% at once?"

  4. 4

    Follow-up prompts

    Extend the scenario to pipeline value

    With conversion rates connected, Forecast can carry the MQL change through to opportunities and revenue, so the recommendation lands in the currency the finance team cares about.

    Using conversion-rates.csv, extend the scenario past MQLs to predicted opportunities and pipeline value, and show me the payback versus the current allocation.

    Sweep for the best split

    Instead of one shift, ask Forecast to search the allocation space and return the mix that maximises MQLs within your budget, so you test the whole frontier, not a single guess.

    Hold total spend fixed and find the channel allocation that maximises predicted MQLs. Show me the top three splits and where each one starts hitting diminishing returns.

    Model a phased ramp

    Forecast can simulate easing into the shift over several weeks rather than all at once, so you see how quickly the new mix pays off and where the risk sits early.

    Re-run the scenario as a phased ramp, 5% a week for four weeks, and compare the cumulative MQLs against moving the full 20% on day one.

  5. 5

    Tips and troubleshooting

    Align spend and leads to one taxonomy

    If spend is tagged one way and leads another, the response curves fit to noise. Give Forecast a single channel taxonomy both datasets share before you trust any scenario’s confidence.

    Respect diminishing returns

    A channel that converts well at £10K/week rarely does at £50K. Forecast models saturation, but tell it where you already suspect a ceiling so it doesn’t over-promise on a channel you’re about to flood.

    Trust the delta only within the data’s range

    The 84% accuracy holds where the model has seen similar spend levels. When a scenario pushes a channel far past its historical range, treat the prediction as directional and phase in rather than committing the full shift at once.

  6. Ready to try it yourself?

    Load your channel spend and leads into Forecast, describe the shift, and get a data-backed MQL projection before you commit a pound of budget.

    Run Forecast