Prescriptive analytics
Definition
Prescriptive analytics
Prescriptive analytics is the branch of advanced analytics that recommends a specific action by combining predictive models, optimization, and business rules. It answers “what should we do?” rather than “what happened?” or “what will happen?” The output is a decision, not a chart: a ranked list of moves with the trade-offs spelled out.
Key takeaways
- Prescriptive analytics outputs a ranked decision, not a dashboard, by combining prediction with optimization and business rules.
- Fewer than 15% of large enterprises had operationalised prescriptive workloads by 2024, according to Gartner.
- The standard pipeline runs four layers: data, predictive model, optimisation engine, and decision interface.
- Companies embedding ML into prescriptive workflows reported 1.5 to 3 times the cost-reduction impact of descriptive-only AI use, per McKinsey 2023.
- Pilots can launch under $50,000 on existing cloud data infrastructure, while production-grade systems typically run $250,000 to $1 million in year one.
It sits at the top of Gartner’s four-stage analytics maturity model, above descriptive, diagnostic, and predictive analytics. According to Gartner’s 2023 analytics framework, fewer than 15% of large enterprises had operationalised prescriptive workloads by 2024 — meaning most analytics teams are still stuck producing dashboards instead of recommendations.
That gap is the opportunity. Once predictive models tell you a customer is likely to churn, prescriptive analytics tells you which retention offer to send, at what discount, and on which channel.
How it works
Prescriptive analytics works by feeding a predictive model’s output into an optimisation engine constrained by business rules, then ranking the resulting actions by expected value. The engine simulates outcomes across thousands of variable combinations and surfaces the move that maximises a chosen objective, whether revenue, margin, service level, or risk-adjusted return.
The pipeline has four moving parts:
| Stage | What it does | Common tools |
|---|---|---|
| Data layer | Pulls historical + real-time signals into a feature store | Snowflake, Databricks, BigQuery |
| Predictive model | Forecasts likely outcomes (churn, demand, fraud) | scikit-learn, XGBoost, TensorFlow |
| Optimisation engine | Searches the action space under constraints | Gurobi, CPLEX, Google OR-Tools |
| Decision interface | Surfaces the ranked recommendation to a human or system | Tableau, Power BI, custom apps |
Machine learning matters here because the optimisation engine needs accurate probability estimates to weigh actions. A 2023 McKinsey State of AI report found that companies embedding ML into prescriptive workflows reported 1.5 to 3 times the cost-reduction impact of teams using AI only for descriptive reporting.
The link to predictive analytics is direct: prescriptive analytics consumes predictive output and acts on it. Without good prediction upstream, the recommendations downstream are guesses dressed up in math.
Examples
Prescriptive analytics shows up wherever decisions repeat at scale and the cost of getting them wrong is measurable.
- UPS ORION, the package giant’s route-optimisation system, has run on prescriptive models since 2013 and saves the company roughly 10 million gallons of fuel a year, per UPS’s own 2022 sustainability disclosures. Drivers get a turn-by-turn sequence, not a heatmap.
- Netflix’s recommendation engine is prescriptive at the action layer: it doesn’t just predict which titles you’ll like, it decides which artwork, row placement, and trailer to show you. Netflix has publicly attributed over $1 billion a year in retained subscriber value to the system.
- Maersk uses prescriptive analytics to reroute container ships in real time around port congestion and weather. The shipping line reported in early 2024 that the system trimmed average voyage times by 4-8% across affected lanes during the Red Sea diversions.
- Philippine BPO operators are now applying prescriptive workforce models to agent scheduling: predicting call volume per skill queue, then prescribing exact headcount, breaks, and overtime offers to hit SLA at minimum cost. It’s a natural fit for knowledge process outsourcing teams that already own the data.
The pattern is consistent: a prediction alone is interesting, but a prescription is what gets actioned in production.
Related terms
These adjacent concepts round out the analytics stack and frequently appear in the same conversation.
- Predictive analytics: forecasts what will happen; the layer directly below prescriptive in the maturity model.
- Descriptive analytics: summarises what already happened; the dashboards and KPI reports most teams start with.
- Diagnostic analytics: explains why something happened; the bridge between descriptive and predictive.
- Business intelligence: the broader discipline of turning data into decisions, of which prescriptive analytics is the most mature rung.
- Data analytics: the umbrella term covering all four maturity stages.
- Incremental analysis: the decision technique prescriptive engines often automate, comparing the marginal cost and benefit of each candidate action.
- Knowledge process outsourcing: where prescriptive analytics work is increasingly delivered from offshore, particularly for finance and pharma clients.
FAQ
What’s the difference between predictive and prescriptive analytics?
Predictive analytics estimates the probability of a future outcome, for instance a 70% chance this customer churns next month. Prescriptive analytics takes that probability and recommends a specific intervention — say, a 15% loyalty discount over email by Tuesday — then ranks it against other options.
Do you need machine learning to do prescriptive analytics?
Not strictly. Simple prescriptive systems run on linear programming and hand-coded business rules. But ML sharpens the predictive inputs and lets the optimisation engine handle non-linear, high-dimensional problems that older rule-based tools can’t scale to.
How expensive is it to set up?
A pilot on existing cloud data infrastructure can launch for under $50,000 with off-the-shelf optimisation libraries like Google OR-Tools. Production-grade systems with custom solvers (Gurobi or CPLEX licences plus engineering) typically run $250,000 to $1 million in year one, per industry benchmarks published by Forrester in 2024.
Which industries benefit most?
Logistics, financial services, retail, healthcare, and telecom lead adoption because their core decisions repeat millions of times daily: routing, pricing, underwriting, scheduling, inventory. Any sector with a high-frequency decision and a measurable objective function is a candidate.
Can you outsource prescriptive analytics?
Yes, and many companies do. Specialist analytics providers in the Philippines, India, and Eastern Europe deliver the modelling, optimisation, and dashboard work on a managed-service basis — typically at 40-60% of onshore cost, per Outsource Accelerator’s 2024 pricing surveys.
Ready to put prescriptive analytics to work without building the team from scratch? Talk to Outsource Accelerator about matching you with a vetted analytics partner.
Independent
