Customer intelligence in the AI era.
The bottleneck in AI-era marketing isn't the model. It's having customer intelligence that's structured, explainable and consent-safe enough for AI to act on. Preset customer tags are that layer — a governed vocabulary of first-party labels that turns raw data into decisions your teams and your agents can trust.
For North American retail & CPG marketing, data and loyalty leaders — 29 preset tags across 12 categories.
The thesis
Tags are the semantic layer between data and autonomous action.
Every retailer now has more data and more capable AI than ever. What most lack is the layer in between: a stable, shared way of describing who a customer is right now that a person and a machine can both reason over. Raw events don't make decisions. Labels do.
A preset tag is exactly that label — a first-party, auditable statement of customer state: at risk, due to reorder, suppressed from marketing. Each one is stable enough to build programs on, explainable enough to defend, and safe enough to hand to an automation or an autonomous agent.
This is what lets intelligence become action at scale. Personalization, targeting, win-back and suppression only happen automatically when customer state is expressed as governed labels — the same labels a marketer selects in a campaign, a rule fires on, and an AI agent reasons over before it acts. Tags turn data into decisions.
The methodology
A discipline, not a data dump.
The tag system works because it follows five rules — the same rules that make its labels safe to automate on.
Built from data you already own
Every tag is derived from first-party signals a retailer already holds — purchases, receipts, loyalty activity, on-property engagement, declared preferences and consent. No third-party data, no rented audiences. The intelligence is yours, and it stays yours.
Sparse, not guessed
A member is tagged only where the signal is genuinely known. If the brand doesn't actually know a customer's age, the demographic tag simply isn't applied — it is never inferred to fill a gap. Honesty about what is unknown is what makes the labels trustworthy enough to automate on.
A clear taxonomy, delivered in packs
Tags are organized into plain-language categories — lifecycle, value, RFM, engagement, purchase-timing, preference, risk, consent, loyalty, advocacy and channel — and shipped as packs. A Universal pack every brand shares, plus Retail and CPG packs, so a business starts on day one with the right set for its model.
Confidence and governance on every label
Each tag carries a confidence level and rules for where it may be used — in campaigns, by AI agents, in recommendations. Compliance tags act as hard guardrails, and operational-risk tags are walled off from marketing entirely. Governance isn't a wrapper around the tags; it's part of each one.
Made to be activated
Tags exist to drive action: lifecycle programs, RFM plays, reorder and replenishment, win-back, cross-category growth — and safe suppression. Because every tag is a stable, shared label, the same customer state means the same thing to a campaign, an automation and an autonomous agent.
Consent is a first-class tag — not a footnote.
Three consent tags — Email Opted In, Do Not Sell / Share (CCPA / US-state) and Suppressed from Marketing— sit alongside every other label and are enforced independently at send time. Suppressed members are held out automatically; opt-out members' data is not exported or shared; marketing email reaches only opted-in members. Activation stays CAN-SPAM- and CCPA-safe by construction, not by remembering to check.
Email Opted In
The permission basis — only these members get marketing email.
Do Not Sell / Share
A hard rule: data must not be exported or shared.
Suppressed from Marketing
Filtered out of every audience, automatically.
The complete catalog
Every preset tag, in plain language.
29 out-of-the-box tags across 12categories. For each: what it means, the signal it's built from, and how a marketer or AI agent would use it.
Lifecycle
Where a customer sits in their relationship with the brand — the backbone of every lifecycle program, from onboarding to win-back.
Lifecycle Stage
UniversalOne value / memberNew · Growing · Mature · At-Risk · Dormant · Churned
A single label describing the member's overall stage of relationship — from freshly joined, through actively building a habit, to fully lapsed.
- Built from
- Derived from how recently and how consistently the member has been buying and staying active over their history with the brand.
- How it's used
- The master routing signal for lifecycle marketing: onboarding for New, habit-building for Growing, recognition for Mature, and re-engagement for At-Risk and Dormant. One stage per member.
New Member
UniversalA member who joined the program within the last 30 days.
- Built from
- Enrollment date within the last 30 days.
- How it's used
- Trigger welcome journeys, first-purchase nudges and early education while attention is highest — and measure onboarding conversion cleanly.
At Risk
UniversalA previously active customer who has gone quiet — no purchase in 90 or more days.
- Built from
- No purchase recorded for 90+ days.
- How it's used
- The early-warning flag. Prompt a check-in offer or a relevant reminder before the relationship cools further — the cheapest moment to save a customer.
Lapsed
UniversalA customer who has drifted well past normal — no purchase in 180 or more days.
- Built from
- No purchase recorded for 180+ days.
- How it's used
- The core win-back audience. Reserve stronger incentives and reactivation messaging for this group, and hand it to autonomous win-back for one-to-one recovery.
RFM
The classic value-and-behavior read — recency, frequency and monetary value combined into one segment every marketer already understands.
RFM Segment
UniversalOne value / memberChampions · Loyal · Potential · New · At-Risk · Hibernating · Lost
A single, well-known segment name that captures how recently, how often and how much a customer buys — all in one label.
- Built from
- Calculated from the combination of recency (how recently they bought), frequency (how often) and monetary value (how much they spend).
- How it's used
- Run classic RFM plays: protect and reward Champions, deepen Loyal and Potential, and stage graduated win-back for At-Risk, Hibernating and Lost. One segment per member.
Value
How much a customer is actually worth to the business today — measured from real, historical spend and points, never a speculative projection.
Value Tier
UniversalOne value / memberVIP (top ~10%) · Growth · Standard · Entry
A single value tier ranking the member by real net spend — what they have actually spent after refunds are taken out.
- Built from
- Net spend after refunds, ranked across the base. This is historical value, not a predicted lifetime-value estimate.
- How it's used
- Tier recognition, service levels and offer generosity by real contribution. VIP marks roughly the top tenth of the base. One tier per member.
High Basket Value
UniversalA big-basket shopper who spends heavily each time they buy — averaging $150 or more per order.
- Built from
- Average order value of $150+.
- How it's used
- Distinct from total spend: this finds customers who buy big when they buy. Target with bundles, premium ranges and threshold offers — even if they shop infrequently.
VIP
UniversalA high-standing loyalty member with 1,000 or more lifetime points.
- Built from
- 1,000+ lifetime points earned.
- How it's used
- Recognize your most engaged earners with status perks, early access and appreciation moments that reinforce the habit.
High Spender
UniversalA customer with $500 or more in approved receipts.
- Built from
- $500+ in verified/approved receipts.
- How it's used
- A spend-based value flag grounded in confirmed receipts — pair with tiering and premium offers for customers whose purchases are proven.
Engagement
How actively a customer engages with your owned properties — a read on attention, independent of whether they've bought recently.
Engagement Level
UniversalOne value / memberHigh · Mid · Low · Dormant
A single label for how active the member has been on your owned properties — the member portal and member pages — over the last 90 days.
- Built from
- On-property portal and member-page activity in the last 90 days. (Email opens are not yet part of this signal.)
- How it's used
- Modulate contact intensity and channel choice: lean into High engagers, and re-warm Low and Dormant members before pushing offers. One level per member.
Purchase Timing & Offer
When a customer is likely ready to buy again — turning purchase rhythm into a well-timed, relevant prompt.
Due to Reorder
UniversalA repeat customer who is coming up on their next expected purchase — someone who has ordered at least twice and is now past about 75% of their typical gap between orders.
- Built from
- An estimate from the member's own purchase cadence: 2+ prior orders, now past ~75% of their usual reorder interval.
- How it's used
- Time replenishment reminders and reorder offers to land just before the customer would naturally rebuy — the highest-intent moment for consumables and routine purchases.
Preference
What the customer has told you — declared choices and clear behavioral leanings that make messaging more relevant.
Preferred Channel
UniversalOne value / memberEmail · SMS · Push
The single channel the member explicitly chose to hear from you on.
- Built from
- A declared choice made by the member — not inferred from behavior.
- How it's used
- Respect the customer's stated preference to lift response and reduce opt-outs. Reach them where they asked to be reached. One channel per member.
Category Focused
UniversalA customer who concentrates their spending — 60% or more of their order lines sit in a single primary category.
- Built from
- 60%+ of order lines in one primary category.
- How it's used
- A cross-sell and category-deepening cue: either go deeper in their favorite category, or introduce the adjacent categories they haven't tried yet.
Demographic
A known demographic attribute, used only where the customer's record actually carries it.
Age Band
UniversalOne value / member18–24 · 25–34 · 35–44 · 45–54 · 55+
A single age band, applied only when the member's record includes a known age.
- Built from
- The age on the member record, where known.
- How it's used
- Tune creative, product emphasis and tone by life stage — applied only to members whose age is actually on file. One band per member.
Operational Risk
Internal service, logistics and policy signals — these describe operational behavior and are NOT marketing targets.
High Return Rate
UniversalInternal · not a marketing targetA customer who returns or cancels a meaningful share of what they order — 30% or more of order value.
- Built from
- Returns and cancellations at 30%+ of order value.
- How it's used
- A service and logistics signal, not a campaign audience. Use it to inform fulfillment, sizing guidance or service outreach — never to push more promotions.
Serial Returner
UniversalInternal · not a marketing targetA customer whose returns and cancellations reach 50% or more of order value.
- Built from
- Returns and cancellations at 50%+ of order value.
- How it's used
- Strictly for service, fraud and policy handling. This tag is never a marketing target — it exists to protect the business, not to sell.
Consent & Compliance
First-class permission and suppression tags. These are hard guardrails — enforced independently at send time so activation stays compliant by construction.
Email Opted In
UniversalCompliance guardrailA member who has given an explicit opt-in to email marketing.
- Built from
- A recorded, explicit email-marketing opt-in.
- How it's used
- The permission basis for email marketing. Only members carrying this can be sent marketing email — the positive side of the consent guardrail.
Do Not Sell / Share
UniversalCompliance guardrailA member who has exercised a CCPA / US-state opt-out of the sale or sharing of their data.
- Built from
- A recorded CCPA / US-state 'do not sell or share' opt-out.
- How it's used
- A hard data-governance rule: this member's data must not be exported or shared. Enforced as a guardrail, not a preference.
Suppressed from Marketing
UniversalCompliance guardrailA member who has opted out of marketing on at least one channel.
- Built from
- An opt-out recorded on one or more channels.
- How it's used
- Use it to EXCLUDE. Every audience is filtered against this tag so suppressed members are held out automatically — the safety net under every send.
Loyalty
How members engage with the loyalty program itself — points balances, redemption behavior and purchase frequency.
Points Rich
UniversalA member sitting on a healthy balance — 500 or more redeemable points.
- Built from
- 500+ redeemable points on balance.
- How it's used
- Nudge redemption to convert dormant points into a visit, and manage outstanding points liability with well-timed prompts.
Reward Redeemer
UniversalA member who has redeemed at least one reward.
- Built from
- One or more rewards redeemed.
- How it's used
- These members already 'get' the program's value. Encourage the next redemption cycle and reinforce the earn-and-burn habit.
Frequent Shopper
RetailA regular retail customer with 5 or more receipts on record.
- Built from
- 5+ receipts submitted or recorded.
- How it's used
- Identify habitual shoppers for frequency rewards and recognition. Part of the Retail pack, tuned to receipt-driven retail models.
Repeat Buyer
CPGA consumer who has bought more than once — past the one-and-done stage, with 2 or more purchases.
- Built from
- 2+ recorded purchases.
- How it's used
- The critical CPG milestone: the shift from trial to repeat. Reinforce the second-and-third purchase to build a lasting habit. Part of the CPG pack.
Advocacy
Customers who don't just buy — they bring others in and contribute content.
Referrer
UniversalA member who has driven at least one referral that converted.
- Built from
- One or more converted referrals attributed to the member.
- How it's used
- Recognize and re-activate your advocates — the customers already growing the base for you — with referral rewards and ambassador programs.
Content Contributor
UniversalA member who has submitted at least one approved piece of user-generated content.
- Built from
- One or more approved UGC / content submissions.
- How it's used
- Identify contributors for creator programs, testimonials and social proof — customers who actively invest their own voice in the brand.
Channel & Origin
How and where a customer was captured and shops — the omnichannel picture, with retail- and CPG-specific reads.
In-Store Captured
UniversalA member who registered in a physical store — via an in-store QR code or capture flow.
- Built from
- Enrollment originated from an in-store QR / capture point.
- How it's used
- Bridge the physical-to-digital gap: keep in-store-acquired customers engaged online and measure the value of store-front acquisition.
Omnichannel Shopper
RetailA retail customer who buys both in store and online.
- Built from
- Purchases recorded across both in-store and online channels.
- How it's used
- The highest-value retail pattern. Protect and grow these customers with consistent cross-channel experiences. Part of the Retail pack.
Store Visitor
RetailA customer with a confirmed, deterministic in-store visit on record.
- Built from
- A deterministic (not inferred) in-store visit recorded.
- How it's used
- Ground store-level marketing and footfall programs in real, confirmed visits. Part of the Retail pack.
Marketplace Buyer
CPGA consumer with a verified marketplace or direct-to-consumer order — for example Amazon or TikTok Shop.
- Built from
- A verified marketplace / D2C order (e.g. Amazon, TikTok Shop).
- How it's used
- Reach and reward customers who buy through marketplaces where you'd otherwise have no direct relationship. Part of the CPG pack.
From tags to action
This is what lets an AI agent act safely, at scale.
When customer state is a set of governed, explainable labels, autonomy stops being risky. An autonomous win-back agent can find the Lapsed, high-Value Tier customers, reason about their RFM Segment and Preferred Channel, and act — while the Suppressed from Marketing and Do Not Sell / Share guardrails hold it inside the lines, on every single member, without a human re-checking each one.
That's the payoff of treating tags as a discipline: the same labels power a marketer's campaign, an automation, a recommendation and an autonomous agent — one shared, first-party language for customer intelligence in the AI era.
One shared language
The same tag means the same thing everywhere
Campaigns
read the same tags
Automations
read the same tags
Recommendations
read the same tags
AI agents
read the same tags
governed & consent-safe by construction