The next best offer for every loyalty member, or none, with consent and caps checked first
One right offer per member, or none, and never one the member did not agree to receive.
See one case, screen by screen ↓Airline Next Best Offer is next best offer software for airline loyalty marketing: offer decisioning that picks one offer, or none, for each member in an approved play every night, so 1:1 loyalty marketing does not mean more messages. Seven agents read the member, check consent and frequency caps first, price each option against a holdout control and write the message with its full price and a plain reason. The loyalty marketing manager approves the play and reviews a sample of real decisions before anything is sent.
How to choose AI for airlines: a buyer's guide to agentic AI solutions →
Updated
Without a next best offer decision for each member, a loyalty team runs a dozen campaigns at once: card acquisition, paid upgrades, status challenges, lounge passes, partner hotels, win-backs. Each one is built as a list and a send date. A CRM analyst pulls the audience from the warehouse, suppresses unsubscribes and complaints, and the email service sends on the campaign date. The same member lands on five lists, gets five offers in a week, and the one that would have mattered is crowded out by whichever campaign sent first.
Consent and frequency caps are checked per campaign, not per member, so a text goes out on a transactional consent and a card offer lands the morning after a six-hour delay. Members who would have bought anyway are counted as wins because only some campaigns have a holdout. And when a member, the Head of Loyalty or a regulator asks why she got this offer, nobody can answer from the record. Loyalty personalization, in practice, is a list per campaign.
Estimated split for one targeted play at a network carrier's loyalty program, worked by hand and with the solution.
Seven specialist agents read each member, check consent and caps first, value every offer against control and pick one, or none; marketing approves the play and reviews the sample.
Corvane Airways runs its loyalty program, Corvane Waypoint, with about 21.6 million members. On Friday, October 16, the agents drafted a new play overnight — Close to Gold · Q4 2026 — for 18,420 Silver members within 3,000 status points of Gold who have a flight booked. Here is how Morgan Ellis, the loyalty marketing manager, works it with the agents, screen by screen, in the working solution.
The member reader read 1,284,600 members with a trip or a tier change in 90 days; the complaint log arrived 18 minutes late and was re-read in full at 1:52 am. The rules keeper checked consent, caps and suppressions first and held back 438,210 members — 34% — before anything was valued. The arbiter picked one offer for 846,390 members, sent 312 to a person because it was under 70% sure, and dry-ran the new play for all 18,420 members.
Held back: 151,240 with no consent on an open channel, 196,880 over a frequency cap, 41,906 after a disruption, complaint or fraud hold, 48,184 where nothing was worth sending.
Her Today screen opens on five items: the Close to Gold play waiting for her signature (16,578 members in test, 1,842 in control, bonus cost $186,400 of $210,000), 19 of 25 sample decisions left to review, Hannah Brooks — where the arbiter is only 64% sure — a proposal to send fewer messages to Blue members, and rule change R7 with the Privacy Officer.
Arbiter: “Start with sample 7, Elena Marsh — she is 1,800 points short and shows the play at its best.”
Elena has 33,200 of 35,000 status points — 1,800 to Gold with 76 days left — and is booked on CV 1442 ORD–SFO on Oct 28 in Main Cabin. Her trips to SFO slowed from every 4 weeks to every 7. The offer scout found 7 offers she qualifies for and kept three: a First Class upgrade at $129, the Corvane Waypoint Card with 70,000 bonus miles, and a status challenge — double status points on her next 2 flights, Gold for 2027. Each shows the chance she says yes, the value over doing nothing, what it is worth sending now and the rules it passed.
“The agents picked status challenge for Elena, 88% sure. Nothing is sent until the play is approved.”
Before anything is valued, the rules keeper holds the card back: Elena had a card offer on Oct 7, and F3 allows one every 30 days, so the next is allowed Nov 6. Email and app are open; text is not. When Morgan asks the agents “What if we text her?”, the rules keeper refuses and says why.
Rules keeper: “I can't: Elena has never given written consent for marketing texts, and R2 requires it. Email reaches her; she usually opens at 7:30–8:00 am.”
The value agent answers with its sources: the upgrade is worth $21 now — a 22% chance of $96 over doing nothing. She ignored the Sep 22 upgrade offer, and members like her buy at the gate 9% of the time anyway. The challenge is worth $405, about 19 times the upgrade, valued against the Q3 play's control results. The upgrade is planned for her Oct 27 check-in, when one more message is allowed.
Arbiter: “Silver members who reached Gold in the Q3 play flew 2.3 more trips a year.”
Morgan can run the agents again and watch the decision rebuild. The member reader reads the loyalty platform, reservations, customer data platform and preference center in 1.4 s; the offer scout reads the offer library and seat inventory in 0.9 s; the rules keeper applies rule set v14 in 0.6 s and holds back the card; the value agent reads the Q3 control results — 14,210 test, 1,580 control — in 1.1 s; then the arbiter and the message writer finish.
The last step always reads “Waiting for Morgan Ellis”: sample review, nothing is sent until the play is approved.
The message writer drafts the app inbox card and the email: “Elena, you're 1,800 points from Gold”, with her progress bar and a line explaining why she is seeing it. The rules keeper checks every claim before it can be sent: full price including taxes and fees, terms quoted from the offer library, consent for email and app, marked as an advertisement with a postal address and one-click unsubscribe, and no urgency that is not true.
“Why am I seeing this? You are 1,800 status points from Gold and have a trip booked on Oct 28.”
Morgan approves Elena's pick and moves to the next sample. The sample review holds 25 decisions from the dry run — 14 status challenges, 5 upgrades, 2 card offers and 4 held back — each with the pick, the reason and her verdict. Darnell Price is left alone on purpose: a 6 h 12 min delay two days ago and complaint CR-88412 still open. For Hannah Brooks, on a Gold status-match trial, the arbiter asks her to decide.
Every verdict — approve, change to another option or send nothing, or reject — carries a reason code the agents learn from.
The play is approved in seven guided steps: audience, offers, rules, test and control, budget, sample review, approve and sign. Each step shows the agents' proposal and its evidence. Test and control is 90 / 10 — 16,578 and 1,842 — so the smallest lift the play can see is 2.1 points, read out on Jan 8, 2027. Control members keep the standing calendar, so the lift is measured, not guessed.
“Nothing is sent while you review.”
The sign-off lists what she is signing — 16,578 members in test and 1,842 in control, a $210,000 budget cap, the sample decisions reviewed, every decision cited and every send logged — with a reason code and her name. The play goes live Saturday, Oct 17 at 7:00 am CT; if she signs before the last sample decisions are reviewed, every send waits until they are. A budget cap over $250,000 would need Grace Lindqvist, Head of Loyalty, to co-sign.
Elena's challenge goes by app inbox and email on Sat Oct 17 at 7:40 am CT, her usual open time, outside quiet hours.
The same play, seen by the four people who carry it — what the campaign calendar looked like, and what it looks like now.
Seven specialist agents read each member, check consent and caps first, value every offer against control and pick one, or none; marketing approves the play and reviews the sample.
Builds each member's moment: tier and status gap, trips booked, spend, recent offers and replies, consent per channel and open complaints, from the loyalty platform, reservations, customer data platform and preference center.
Finds every offer the member qualifies for in the approved offer library and live inventory — upgrade seats, lounge capacity — and keeps the three worth weighing.
Checks consent per channel, frequency caps across all plays, quiet hours, suppressions and disclosure rules first, and holds an option or a member back with the rule that applies.
Estimates each option's chance and the value it adds over doing nothing, using the play's held-out control group, after the offer's cost.
Picks one option, or none, writes why the others lost and plans the next step in sequence — such as the upgrade at check-in when the frequency rule allows it.
Drafts the message in the brand voice with the full price, the terms and a plain “why am I seeing this”, then has the rules keeper check every claim.
Measures each play against its control by segment, flags drift and unsubscribes over the cap, and proposes pauses and changes.
Approve plays, sign rules, co-sign budgets. The agents propose; a named person decides.
Ask in plain words or tell the agents what to do. Every answer points to the record it came from, and the agents open the screen it is about.
Why did Elena Marsh get the status challenge?
Because it is worth $405 now — about 19 times the upgrade. She is 1,800 points from Gold; Silver members who reached Gold in the Q3 play flew 2.3 more trips a year. The card waits: she had one on Oct 7 and F3 allows one per 30 days. No texts — she never gave written consent.
Who was held back last night, and why?
438,210 of 1,284,600 members — 34%: 151,240 with no consent on an open channel, 196,880 over a frequency cap, 41,906 after a disruption, complaint or fraud hold, and 48,184 where nothing was worth $5.
Make sure nobody with an open complaint gets an offer
That is already rule S2: seven days quiet after a disruption over 3 hours or an open complaint. Last night it held back 38,906 members, including Darnell Price, whose complaint CR-88412 is still open. Changing it to 14 days needs Ruth Calloway's signature.
Draft a play for Gold members with no long-haul trip this year
Drafting it now: Gold members with no long-haul flight in 2026, one reason to book before March, a 90 / 10 control and a $120,000 cap. Consent and caps come first. The dry run is ready by 5:40 am tomorrow; nothing is sent until you approve it.
13 screens from the working solution, on its sample data. Pick one to see it large.
The bars the agents work to are rules and settings, signed, versioned and on the audit trail. Rule set v14 was signed by the Privacy Officer on Oct 9.
| Setting | Default | Choose from |
|---|---|---|
| Consent per channel (R1, R2) | Email and app need opt-in marketing consent; texts only with written consent | Email · App inbox and push · Text — each switched on or off |
| Quiet hours (R3) | 9 pm to 8 am, member local time | Send time follows each member's usual open time, inside the window |
| Frequency caps (F1–F4) | One marketing message a day · three a week · one card offer per 30 days · one upgrade offer per trip | Each cap switched on or off; card offers count the card partner's own calendar |
| Suppressions (S1, S2, S4, E2) | Fraud hold · seven days quiet after a disruption over 3 hours or an open complaint · partner offers removed after an opt-out of sharing · no card offers under 18 | Changed only with the Privacy Officer's signature, after a replay |
| Value floor (T1) | $5 per message, after cost | Any dollar amount; signed by the Head of Loyalty with a reason code |
| Arbiter confidence | 70% | Below it, the arbiter asks a person |
| Sample size | 25 + 0.2% for plays over 50,000 members | Decisions reviewed before a play goes live |
| Co-signature and pause | Plays with a budget cap over $250,000 · pause proposed at unsubscribes over 0.30% | Saved with a reason: business decision · pilot result · regulatory change · other |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample member data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (FTC, CAN-SPAM Act: A Compliance Guide for Business — up to $53,088 per email; 47 U.S.C. 227 and 47 CFR 64.1200 (TCPA) — $500 per text, $1,500 if willful; IdeaWorksCompany, Global Estimate 2025 — $157B airline ancillary revenue projected for 2025, 15.7% of revenue). Corvane Airways, Corvane Waypoint and the people and members named on this page are fictional — characters and sample data in the working solution.
Airline Next Best Offer is next best offer software for airline loyalty marketing: offer decisioning that picks one offer, or none, for each member in an approved play every night, so 1:1 loyalty marketing does not mean more messages. Seven agents read the member, check consent and frequency caps first, price each option against a holdout control and write the message with its full price and a plain reason. The loyalty marketing manager approves the play and reviews a sample of real decisions before anything is sent.
Next best offer chooses among offers; next best action can be any step — an offer, a service message or nothing at all. Here the agents choose across both: offers such as a paid upgrade, the co-brand card or a status challenge, service messages such as “Your progress to Gold”, and “send nothing” when consent, a cap or a recent disruption says no. That is airline personalization by choosing less, not sending more.
Offer arbitration is the core of airline offer optimization. The offer scout finds every offer the member qualifies for and keeps the three worth weighing. The rules keeper removes any the rules stop, the value agent estimates what each adds over doing nothing after its cost, and the arbiter picks the highest — writing why the others lost and planning the next one for later, such as an upgrade at check-in.
Every play carries its own held-out control group from day one — 90 / 10 by default — and control members keep the standing calendar. The value agent prices options by what they add over doing nothing, and the results reader reports conversion, added revenue and unsubscribes against control by tier.
The rules keeper checks consent per channel before any choice: marketing texts need prior express written consent, quiet hours are 9 pm to 8 am member local time, and caps count across every play, including the card partner's own calendar. Each email is marked as an advertisement with a postal address and a one-click unsubscribe, honored within 1 hour.
Yes. Nothing is sent outside a play the loyalty marketing manager approved and signed with a reason code, and plays over $250,000 need the Head of Loyalty to co-sign. She reviews a stratified sample of 25 real decisions first, the arbiter asks a person below 70% confidence, and only the Privacy Officer can sign a change to the consent and frequency rules.
Yes. Every message carries a plain “why am I seeing this” line in the member's own terms, every decision keeps its reason and sources, and the audit trail holds every decision, approval and send for seven years, with an evidence pack per member or play.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your offer library, consent rules, frequency caps and a sample of past campaigns — and it goes live once every quality gate has passed. A first-time setup connects the loyalty platform, reservations, preference center and messaging systems and tests each connection.
Stolen miles stopped before the award is flown, and honest members never locked out.
1-min demoMissing Miles Claims ProcessingMissing miles credited in minutes, each one tied to the flown coupon and its record.
1-min demoLoyalty Partner Billing ReconciliationEvery partner statement tied out line by line, every missed mile billed or credited.
1-min demoWe'll run Airline Next Best Offer on a sample of your own members, offers and consent rules.