Next Best Action (NBA)
Determine the optimal action for every customer in real time. EkamFlow's next best action prediction selects the right campaign, channel, offer, and timing — personalized at the individual level.
Next best action (NBA) is an AI decisioning approach that predicts the single most valuable action to take with each customer at each moment — a retention save, an upsell offer, a cross-sell prompt, a lifecycle email, or holding fire entirely. Instead of running parallel campaigns and hoping the right one lands, NBA evaluates each customer individually against every possible action and returns the one most likely to advance a defined business outcome. Modern NBA systems replace rule-based decision trees with per-customer probabilities that plug directly into campaign orchestration platforms, CDPs, and CRMs.
What is Next Best Action (NBA)?
Next best action (NBA) is an AI decisioning approach that determines the optimal action to take with each customer at any given moment. Rather than batch-segmenting customers into campaigns, NBA evaluates every customer individually and selects the action most likely to achieve your business objective — whether that's retention, upsell, engagement, or conversion.
Traditional next best action systems require complex decisioning engines, separate models for each action type, and extensive rule configuration. Marketing teams spend weeks setting up campaign logic that still can't personalize at the individual level.
EkamFlow's private AI model handles next best action as part of a unified prediction. One API call returns the recommended action, channel, timing, and offer for each customer — alongside churn risk, LTV, and every other signal. No separate decisioning engine, no rule configuration, no ML team.
How it works, step by step
- 1
Connect your warehouse
Snowflake, BigQuery, Databricks, Redshift, or Postgres. Read-only access to the tables you already have — no ETL to build.
- 2
Auto-detect your data
EkamFlow inspects your customer, event, and campaign tables and maps them without a schema config file. Sparse or messy data is handled the same way.
- 3
Private model trains on your outcomes
The model learns from your specific customer actions, responses, and revenue events — not generic industry patterns. Nothing leaves your cloud environment.
- 4
One API call per customer decision
Every request returns the recommended action, next best offer, channel, and time in a single response — plus churn risk and LTV. Sub-60ms latency.
- 5
Continuous retraining
As customer behavior and campaign performance shift, the model retrains automatically. No MLOps, no monitoring rota, no drift alarms.
Signals the model uses
- Recent engagement across email, push, in-app, and web
- Lifecycle stage and tenure
- RFM (recency, frequency, monetary value)
- Past campaign responses per channel
- Product and feature adoption patterns
- Support-ticket sentiment and volume
- Price sensitivity and discount response
- Cohort behavior relative to peers
- Cross-channel activity fatigue signals
- Revenue exposure and LTV band
Where it fits in your stack
- CDPs: Segment, Rudderstack, mParticle, Hightouch
- Marketing automation: Braze, Iterable, Customer.io, Klaviyo, MailChimp
- CRM: Salesforce, HubSpot, Zoho
- Data warehouse: Snowflake, BigQuery, Databricks, Redshift
- Reverse ETL: Census, Hightouch, Polytomic
- Personalization: Dynamic Yield, Optimizely
- Ad platforms: Meta CAPI, Google Enhanced Conversions, TikTok Events API
Build in-house vs. EkamFlow
| Dimension | Build in-house | EkamFlow |
|---|---|---|
| Time to first decision | 3–6 months (rule authoring, testing) | Days |
| Team required | Marketing ops + data engineer + ML consultant | None |
| Rules to maintain | Dozens to hundreds | Zero |
| Personalization depth | Segment-level | 1:1 per customer |
| Cross-channel awareness | Manual coordination | Native |
| Latency | Depends on decisioning platform | <60ms |
| Model refresh | Quarterly retrain or manual rule updates | Continuous |
What it looks like in different industries
A DTC brand routes every browsing session through NBA. For a returning customer with high LTV but declining basket size, the model recommends a loyalty-tier reveal via email; for a first-time cart abandoner, it triggers an SMS with free-shipping code. Same infrastructure, a different next best action per customer.
A subscription analytics platform uses NBA to pick which of six lifecycle motions — feature-onboarding email, in-app tooltip, CS outreach, expansion offer, renewal reminder, or 'do nothing' — should fire for each account each week. Marketing ops went from maintaining 400+ segment rules to a single API integration.
A neobank sequences customer touches — card-usage nudges, savings-goal reminders, credit-line offers — based on NBA outputs. What used to be four disconnected teams running four campaign calendars became one decisioning layer feeding four channels, with all conflicts resolved at the API.
A streaming service replaces its retention-team playbook (churn-risk email → discount email → win-back email) with NBA. Some at-risk viewers get a content recommendation instead of a discount; others get a family-plan reveal; some get nothing (already re-engaged). Save rate improves; discount cost drops sharply.
A regional carrier's marketing team uses NBA to decide which of a dozen cross-sell motions (device upgrade, added line, home-internet bundle, streaming perk) is right for each account in each week's outbound wave. Blended cross-sell attach rate lifted 21% in the first quarter — with no new campaign creative built.
An insurance carrier uses NBA to sequence lifecycle touches — annual policy review nudges, add-on coverage prompts, referral asks, digital-tool activation — across email and app push. Regulatory guardrails stay hard-coded; NBA optimizes within the compliant action set.
How EkamFlow does it
Unified decisioning
Next best action, next best offer, next best channel, and next best time — all returned in one API call. No separate models or decisioning layers to integrate.
Context-aware recommendations
NBA predictions factor in the customer's churn risk, lifetime value, recent behavior, and engagement history — not just campaign eligibility rules.
Real-time personalization
Every customer gets their own optimal action in real time. Feed NBA predictions into your marketing automation, CRM, or customer-facing app for true 1:1 personalization.
Related predictions
Frequently asked about next best action (nba)
Next best action decides what to do — send a message, wait, hand off to a human, present a UI element. Next best offer is the specific promotion or incentive to use inside that action. EkamFlow returns both in the same API call, along with next best channel and next best time, so you don't need separate systems.
No. NBA output is a per-customer prediction returned via REST API. You can feed it directly into any campaign orchestration tool, CDP, CRM, or homegrown workflow. Most customers use their existing marketing automation platform and let EkamFlow replace the decisioning layer entirely — no separate NBA product needed.
New customers get a cold-start-aware score. The model uses cohort behavior — customers with similar early-stage patterns — and industry-appropriate defaults until enough personal signal accumulates, usually within 5–10 interactions. There's no separate onboarding model to maintain.
Yes, and often it should. EkamFlow explicitly models suppression outcomes, so a customer who's likely to churn from message fatigue will get an empty action rather than another email. This is measurable directly against revenue-per-customer, not open rate.
EkamFlow supports holdout groups by default — a small share of customers continues to receive the previous rule-based decision so you can measure NBA lift against a real control. Typical customers see 15–30% lift on the primary conversion metric within 60–90 days, with the largest gains coming from suppression (avoiding fatigue) rather than aggressive targeting.
No — it works alongside. Journey builders are still useful for defining the mechanics of a campaign: templates, timing, exclusions, unsubscribe logic. What NBA replaces is the branch logic — the 'if customer meets criteria X, put them in journey Y' rules. Instead of maintaining that logic, journeys become listeners for NBA outputs.
Ready to add next best action (nba) to your stack?
Connect your warehouse. Get live predictions in weeks, not months.
Book a demo