Customer Retention Strategies That Actually Work

1Improve the customer experience

What it does. Removes the friction, failures and small indignities that make customers stop choosing you — across onboarding, purchase, delivery, product use, support and offboarding.

Why it works. Because most churn is not a marketing failure. It is a delivery failure that marketing was then asked to compensate for. Customers who leave because the checkout broke, the delivery was late, the support ticket went unanswered or the product did not do what they expected are not persuadable by a discount; they are telling you something specific about your operation. Fixing the cause removes a permanent drag on every other retention investment you make.

Where the friction usually hides. In practice, six places account for most of it:

• Onboarding — the gap between buying and getting the first result.
• Checkout and payment — forced account creation, surprise costs at the final step, failed cards with no retry path.
• Delivery and fulfilment — silence between order and arrival, which is the single most common trigger for a support contact.
• Support — response time, and whether the customer has to explain their situation more than once.
• Product experience — the feature that is technically present but nobody can find.
• Cancellation and offboarding — a deliberately difficult exit converts a neutral departure into an angry one, and in some jurisdictions it is also a regulatory risk (see the note below).

Why loyalty cannot compensate. This is the most important sentence in this guide: a rewards program layered on top of a poor experience is a subsidy you pay forever to slow down an outcome you have not changed. It raises your cost per retained customer, it attracts precisely the price-sensitive segment least likely to stay when the subsidy stops, and it hides the underlying signal by converting "customers are leaving" into "redemption is up." If your exit survey says people left because something did not work, no amount of points will fix it.

Best for

Every business. Mandatory where exit interviews name a product or service failure.

Difficulty

High — it crosses departments and usually needs operational, not marketing, authority.

Cost

Variable; often engineering and staffing time rather than budget line-items.

Time to first result

One to two quarters for measurable movement.

Potential impact

Very high — it raises the ceiling on every other strategy.

Primary KPI

Churn rate segmented by stated reason; secondary: support contact rate, first-week failure rate.

Common mistake

Measuring satisfaction instead of friction. A satisfaction score tells you how people feel; a friction audit tells you what to fix on Monday.

Example

Suppose a subscription box business examined its first-month cancellations and found a large share clustered in the 48 hours after a delivery exception. The fix would not be a retention offer — it would be a proactive notification and a re-ship policy triggered by the carrier event.

When not to use it

Never "not" — but do not start here if you have no idea where the friction is. Run Strategy 7 first for two weeks so the expensive work is aimed at something.

2Build a loyalty program

What it does. Gives customers a structured, visible reason to consolidate their spending with you rather than spreading it across substitutes.

Why it works. Three mechanisms, in descending order of reliability. First, switching cost: accumulated progress is a real asset the customer would forfeit by going elsewhere. Second, goal pull: an incomplete progress bar is uncomfortable in a way that motivates completion. Third, recognition: being known and treated differently is genuinely valuable to some customers and costs nothing to deliver.

The second mechanism has direct experimental support. In "The Endowed Progress Effect: How Artificial Advancement Increases Effort" (Journal of Consumer Research, March 2006), Joseph Nunes and Xavier Drèze documented that "by converting a task requiring eight steps into a task requiring 10 steps but with two steps already complete, the task is reframed as one that has been undertaken and incomplete rather than not yet begun. This increases the likelihood of task completion and decreases completion time." That is a real, published finding about card design, and it is the single most actionable piece of loyalty psychology available: a ten-stamp card issued with two stamps already applied is not the same product as an eight-stamp card, even though the customer must buy the same number of coffees.

The Loyalty Mechanic Selection Matrix™

Choosing the mechanic is most of the work. The wrong one produces a program that is either too slow to motivate anyone or so generous it pays for purchases that would have happened anyway.

Best for

Repeat-purchase businesses with cycles measured in days or weeks rather than years.

Difficulty

Medium — the design decisions are harder than the software.

Cost

Software plus the margin cost of rewards. The reward cost is the part that surprises people.

Time to first result

Four to eight weeks to launch; one full purchase cycle before the numbers mean anything.

Potential impact

Medium–high, concentrated in the middle of your customer distribution.

Primary KPI

Repeat visit rate among enrolled customers vs a matched unenrolled group.

Common mistake

Rewarding customers who would have purchased anyway. If your top decile enrols first and your rewards go mostly to them, the program is a margin transfer, not a retention program. Measure incrementality, not enrolment.

Example

A lunch spot with an average visit gap of nine days sets a six-stamp card. Six visits at nine days is roughly a two-month goal — close enough to feel achievable, far enough to change behaviour. A twelve-stamp card would have been abandoned.

When not to use it

When purchases are genuinely rare (a mattress, a wedding venue, a boiler), when your churn is dissatisfaction-driven, or when your margins cannot absorb the reward without a price rise customers will notice.

Go deeper: the complete guide to customer loyalty cards · how digital stamp cards work · punch cards, paper versus digital · what free loyalty software can and cannot do

3Personalise the customer journey

What it does. Changes what a customer sees, receives and is offered based on what you already know about them, so that fewer of your communications are irrelevant.

Why it works. Not because personalisation is charming, but because it raises the hit rate. Retention communication competes for a scarce resource — the customer's willingness to keep opening things you send. Every irrelevant message spends a little of that willingness. Personalisation is best understood as relevance rationing: it lets you send fewer, better-aimed messages and preserve the channel.

The five inputs worth using, roughly in order of how much signal they carry:

1. Purchase history — what they bought, when, how often, in what category.
2. Lifecycle stage — new, activated, repeat, at-risk, lapsed. This alone justifies most segmentation.
3. Frequency and recency — the two variables that predict near-term behaviour better than almost anything demographic.
4. Stated preferences — what they told you, which is underused because it requires asking.
5. Behavioural triggers — an action or an absence that implies a need right now.

Useful personalisation versus creepy personalisation

The line is not about how much you know. It is about whether the customer can see how you know it, and whether the use serves them.

Best for

Businesses with enough transaction history that segments are real rather than invented.

Difficulty

Medium — the data plumbing is usually harder than the content.

Cost

Low to medium; mostly platform capability and analyst time.

Time to first result

Four to twelve weeks depending on data readiness.

Potential impact

Medium — reliable but rarely transformative on its own.

Primary KPI

Response rate by segment versus the unsegmented baseline; unsubscribe rate as the guardrail.

Common mistake

Confusing merge tags with personalisation. Putting a first name in a subject line changes nothing about relevance. Personalisation means the offer and the timing differ, not the greeting.

Example

A pet supplies retailer segments purely on reorder interval per product and sends nothing else. Fewer campaigns, higher revenue per send, and no additional data collected.

When not to use it

When you have fewer than a few hundred customers with history — you will build machinery to segment noise. Talk to them individually instead; it is faster and more accurate.

4Create a strong onboarding experience

What it does. Gets a new customer from purchase to first real value as quickly and reliably as possible, and establishes the habit that makes the second purchase likely.

Why it works. Because the largest single drop in almost every cohort retention curve happens in the first period. A customer who never reached the outcome they bought has no basis for a second purchase, and no later intervention can retroactively supply one. Onboarding is where retention is won or lost before any retention marketing has begun.

Three jobs, in order. Activation — the customer does the thing that makes the product work. Education — they know enough to succeed without support. Habit — the behaviour has a trigger and a rhythm attached to it.

A sample onboarding timeline

Best for

SaaS, subscriptions, memberships, gyms, courses, and any product with a learning curve.

Difficulty

Medium — needs product and marketing to agree on what "activated" means.

Cost

Low to medium; mostly design and instrumentation.

Time to first result

Six to twelve weeks — you must wait a full cohort to see the effect.

Potential impact

Very high. Improvements here move the whole curve, not just the tail.

Primary KPI

Activation rate, then 30-day cohort retention.

Common mistake

Onboarding that teaches the product instead of delivering a result. A feature tour is not activation. Define activation as the first moment the customer gets what they paid for, then remove everything between purchase and that moment.

Example

A gym analyses its own member data, finds that early attendance separates members who stay from members who lapse, and defines activation as "three visits in the first fourteen days." Everything in the first two weeks — the induction, the check-in, the class invitation — is redesigned to serve that one number.

When not to use it

When there is no activation step at all — a one-off transactional purchase with no learning curve. There, Strategy 5 is the equivalent play.

Related reading: digital membership cards and how they carry status through onboarding

5Use post-purchase engagement

What it does. Uses the window immediately after a purchase — when attention is highest and goodwill is fresh — to reduce regret, increase successful use, and make the second purchase a natural next step rather than a new decision.

Why it works. The post-purchase window is the only moment when a customer is guaranteed to be paying attention to you. It is also when the emotional risk of the purchase is highest. Communication that reduces that risk — where is my order, how do I use this, what if it is wrong — converts anxiety into confidence, and confident customers buy again.

The sequence, roughly in order of value:

1. Order and delivery status, proactively. Silence in this window is the largest generator of avoidable support contacts.
2. Usage education, targeted to the specific product bought.
3. A check-in at the point where problems typically surface — early enough to fix, late enough to be real.
4. A review request, after the customer has had time to form an opinion.
5. Loyalty enrolment, when goodwill is at its peak.
6. A relevant next step — the consumable, the companion item, the refill — timed to the product's actual cycle.

Why immediate discounting is usually the wrong move

The reflex after a first purchase is to send a discount for the second. It works, in the narrow sense that it produces orders. It also does four things people do not account for:

• It teaches the customer that your list price is negotiable, permanently reducing what they will pay.
• It gives margin to customers who were going to buy again regardless.
• It selects for the price-sensitive segment, which is the segment least likely to stay.
• It substitutes for the harder work of making the product experience good enough to earn the second purchase.

Better first moves, in order of preference: make the product work; make the next purchase easier rather than cheaper; give recognition or access rather than money; and if you must discount, discount something adjacent rather than the thing they just bought at full price.

Best for

Ecommerce, DTC and anything with a delivery gap or a usage learning curve.

Difficulty

Low — mostly automated messages against events you already have.

Cost

Low. This is the cheapest meaningful retention work available to most businesses.

Time to first result

Two to four weeks.

Potential impact

Medium, and unusually reliable.

Primary KPI

Second-purchase rate within one purchase cycle.

Common mistake

Making the entire post-purchase flow about selling the next thing. If every message asks for something, the sequence stops being opened, and you lose the channel for the messages that would have mattered.

Example

A skincare brand replaces its "10% off your next order" day-3 email with a "how to use this for the first two weeks" message and moves the offer to day 45, timed to when the product runs out. Fewer discounts given, and the reorder lands when the customer actually needs it.

When not to use it

Rarely — but if your delivery or product quality is unreliable, fix that first. Post-purchase messaging into a broken fulfilment process just draws attention to the problem.

6Build a win-back program

What it does. Systematically identifies customers who have left, works out why, and makes a specific attempt to restart the relationship.

Why it works. A former customer has already crossed the hardest barrier: they know who you are, they have transacted, and you hold their contact details and history. That makes them cheaper to reach and easier to convert than a stranger — but only if the offer addresses why they left. Win-back that ignores the cause is just a discount to people who have already told you the product was not right.

The Customer Win-Back Framework™

Defining inactivity. Derive it, do not borrow it. Plot the distribution of gaps between consecutive purchases across your customer base and pick a percentile — commonly somewhere between the 80th and 90th — as the point where a gap has become unusual for your business. A wine merchant and a phone repair shop should not use the same number, and neither should use ninety days because an article said so.

Offer selection. Run a ladder, not a single blast. Message one carries no offer at all: it exists to find the people who simply forgot, and it is the cheapest conversion you will ever get. Message two adds a small, specific incentive. Message three, if you send one, is your ceiling — and it should go only to segments whose historic value justifies it. A large offer sent first tells everyone, including the people who needed nothing, what you are willing to pay.

Best for

Any business that can identify who left and roughly when.

Difficulty

Low once identification exists; impossible without it.

Cost

Low, plus the margin cost of whatever offer converts.

Time to first result

Two to four weeks.

Potential impact

Medium — a reliable recurring source of incremental revenue.

Primary KPI

Win-back rate versus a randomly held-out control, and incremental revenue net of offer cost.

Common mistake

No control group. Some lapsed customers return anyway; without a holdout you will attribute their return to your campaign and scale a program that is mostly measuring the calendar.

Example

A garden centre notices that customers who bought in spring and not again by August are unlikely to return unprompted. Its first August message is a seasonal planting guide with no offer. Roughly the customers who respond to that message never needed a discount at all.

When not to use it

When you know why they left and it is unfixed. Winning back a customer into the same broken experience produces a second, angrier departure and a worse review.

7Use customer feedback

What it does. Turns the reasons customers leave from a guess into a list, so that retention investment can be aimed rather than sprayed.

Why it works. Because every other strategy in this guide is a bet on a hypothesis about why customers leave, and feedback is the only thing that tests the hypothesis. It is the cheapest strategy here and the one with the highest indirect impact, because it determines whether the other eleven are aimed at anything.

Where the signal actually is, ranked by usefulness rather than by how often it is collected:

An honest word about NPS

Net Promoter Score is useful as a cheap, repeatable prompt that gets you to the free-text box where the actual information lives. It is not the growth predictor it is often presented as. When Keiningham, Cooil, Andreassen and Aksoy tested Frederick Reichheld's claim in "A Longitudinal Examination of Net Promoter and Firm Revenue Growth" (Journal of Marketing, July 2007) — using longitudinal data from 21 firms and more than 15,500 interviews drawn from the Norwegian Customer Satisfaction Barometer, compared against the American Customer Satisfaction Index — they reported that "the research fails to replicate his assertions regarding the 'clear superiority' of Net Promoter compared with other measures in those industries." The scope limit matters: this is Norwegian data in specific industries, not a universal refutation.

The practical reading: run NPS if it gets your organisation asking customers questions regularly. Do not set targets on the number, do not compensate anyone against it, and never present it to a board as evidence that retention is improving. The comment field is the deliverable; the score is the excuse to collect it.

How feedback becomes retention action

Most feedback programs fail at the conversion step, not the collection step. The discipline that fixes it is unglamorous: every month, count the reasons, rank them by how much revenue each accounts for, and assign the top three to a named owner with a date. A reason that is not counted cannot be prioritised; a reason without an owner does not get fixed. Then close the loop back to the customers who raised it — the single most underused retention move available, because it converts a complainer into someone who has seen you respond.

Best for

Everyone. This is the diagnostic that directs the other eleven strategies.

Difficulty

Low to collect; medium to act on, which is where programs die.

Cost

Very low.

Time to first result

Days to first signal; a quarter to a fixed root cause.

Potential impact

Indirect but decisive — it is the difference between aimed and unaimed spending.

Primary KPI

Number of distinct churn reasons captured, and number closed per quarter.

Common mistake

Collecting feedback with no mechanism to act on it. This is worse than not collecting: you have now asked customers to spend effort telling you something and demonstrated that it changes nothing.

Example

A software company replaces its cancellation dropdown with one free-text box and reads every answer for a month. The top reason turns out to be a specific onboarding step nobody had considered a problem, and it does not resemble any option the dropdown had offered.

When not to use it

Never skip it — but do not launch a large survey program if you have not yet read your own support tickets. The answer is usually already in the building.

8Build a referral loop

What it does. Turns existing customers into an acquisition channel, and — the part that matters for this guide — deepens their own commitment in the process.

The distinction that has to be made first. Referral is an acquisition mechanism. Including it in a retention guide is only defensible because of a second-order effect: the act of recommending something publicly is generally held to strengthen the recommender's own attachment to it, and a referred customer arrives with a personal endorsement attached rather than cold. Those are plausible second-order effects, and we state them as the reasoning for including referral here rather than as findings we can cite. Neither makes referral a retention strategy in the primary sense, and a business with a retention problem that launches a referral program has usually chosen the more enjoyable project over the necessary one.

Why it works when it works. Recall the Reinartz and Kumar finding from Section 2: word-of-mouth behaviour was markedly stronger when customers scored high on both behavioural and attitudinal loyalty than when they scored high on repeat purchasing alone. Grocery-chain customers high on both measures were "54% more likely to be active word-of-mouth marketers and 33% more likely to be passive word-of-mouth marketers than those who scored high on behavioral loyalty alone." The implication for program design is direct: referral programs do not create advocates, they harvest them. If you do not have people who genuinely like you, an incentive will produce a trickle of low-quality referrals and nothing else.

Design decisions that matter:

• Timing. Ask at the moment of demonstrated satisfaction — after a good support resolution, after a milestone, after the reward is claimed — not at a fixed number of days.
• Reward structure. Our editorial preference is two-sided over one-sided rewards, on the reasoning that they give the referrer something to offer rather than something to gain, which is socially much easier to do.
• Reward type. Account credit keeps the value inside your business, and its cost to you is the margin on the redemption rather than the face value — which is why it is worth modelling against cash before choosing.
• Friction. Every step between intent and share loses a proportion of participants. One tap, pre-written, no login.
• Quality control. Measure retention of referred customers separately. A program that acquires people who churn immediately is buying you a worse cohort at a premium.

Best for

Businesses with genuine advocates, a natural share moment, and something worth recommending.

Difficulty

Medium — tracking and attribution are the hard parts.

Cost

Medium; the reward is a real acquisition cost and should be compared to your paid CAC.

Time to first result

Four to eight weeks.

Potential impact

Medium for acquisition; secondary for retention.

Primary KPI

Referrals per active customer; guardrail metric is retention of referred customers versus other channels.

Common mistake

Launching a referral program as a response to a retention problem. If customers are leaving, they are not going to recommend you, and the program will simply document that fact expensively.

Example

A meal-kit service moves its referral ask from "day 14 after signup" to "immediately after a customer rates a box five stars," and gives the referrer three free meals to hand out rather than cash back.

When not to use it

When satisfaction is unproven, when the product is private or sensitive enough that recommending it is socially awkward, or when your churn is high — fix that first or you are paying to acquire into a leaking bucket.

9Use lifecycle messaging

What it does. Replaces the marketing calendar with the customer's own timeline, so messages arrive because something is true about that customer rather than because it is Thursday.

Why it works. A large share of churn in non-contractual businesses is not rejection — it is drift. The customer did not decide to leave; they simply stopped thinking about you, and nothing arrived at a moment when the thought would have been useful. Lifecycle messaging is the intervention for drift specifically. It is close to worthless against dissatisfaction, and applying it there makes things worse, because a reminder to a dissatisfied customer reads as tone-deafness.

The Customer Attention Ladder™

Before choosing a channel, work out which rung a customer is on. Each rung unlocks a different set of options, and most businesses try to run rung-4 tactics on rung-1 customers.

The Retention Channel Selection Matrix™

Best for

Businesses whose customers drift away rather than reject them.

Difficulty

Medium — the triggers require reliable event data.

Cost

Low to medium; mostly platform and setup.

Time to first result

Four to eight weeks.

Potential impact

High and durable, because triggered programs keep working without new campaigns.

Primary KPI

Revenue per message sent, with unsubscribe and opt-out rate as the guardrail.

Common mistake

Adding messages until the numbers stop improving. Every channel has a frequency ceiling, and crossing it does not produce a gradual decline — it produces opt-outs, which are permanent. Budget your sends like a finite resource, because that is what they are.

Example

An optician sends nothing on a calendar. It sends one message twelve months after each customer's last eye test, one when a contact-lens supply is due to run out, and one when a prescription is about to expire. Three messages a year, each tied to a fact about that customer.

When not to use it

When churn is driven by dissatisfaction. More messages to unhappy customers accelerate opt-outs and turn a quiet departure into a public one.

Compare the channels in depth: wallet marketing versus email marketing · wallet marketing versus SMS marketing · how push notification platforms compare

10Use wallet-based loyalty

What it does. Puts a loyalty card, membership card, coupon or offer into Apple Wallet or Google Wallet — the pass app already installed on the customer's phone — so that the customer's status with you lives on their device, updates remotely, and does not require them to install anything.

Why it works. Two structural properties, both of which are about persistence rather than reach:

• No install barrier. Apple Wallet and Google Wallet ship on the devices people already own. Saving a pass is a single tap, which is a categorically lower bar than downloading, installing and registering an app — and it is the reason wallet loyalty is viable for businesses that could never justify an app.
• The pass stays and stays current. A pass is a durable object on the device that the business can update after issuance. Apple documents the update flow explicitly: your server sends a push notification to registered devices — one that carries "an empty JSON dictionary for the payload" — and the device then fetches the updated pass. Google's model works differently but achieves the same thing: changes made to a Google Wallet pass class will propagate to every pass object that references it, and Google's documentation notes that users see those changes reflected in their Google Wallet app the next time the pass syncs.

What it is not. Wallet is not a broadcast channel, and any vendor implying otherwise is misrepresenting how the platforms work. Both platforms deliberately ration notifications. Google documents that "you may send a maximum of 3 messages that trigger a push notification in a 24 hour period," with the same cap applying to update-triggered notifications, and restricts update notifications to a short allow-list of fields — on a loyalty pass, rewardsTier, secondaryRewardsTier and programName on the class, and loyaltyPoints.balance and secondaryLoyaltyPoints.balance on the object. Apple's lock-screen relevance is a display feature, not a notification feature: its documentation describes showing a pass on the lock screen at a relevant time or place, allows "only 10 relevant locations" per pass, and separately notes that "a push notification for a pass update works only in the production environment." If you want a channel with unlimited sends, wallet is the wrong choice — and that constraint is a design feature that keeps the channel worth having.

The Wallet Retention Loop™

Best for

In-person and local businesses with repeat visits, and any audience that resists installing apps.

Difficulty

Low to launch; the hard part is the identification step at the counter.

Cost

Low. Published PushNotice plans start at $0 and run to $29 and $79 per month, with custom agency pricing.

Time to first result

One to three weeks to launch; one purchase cycle to read the numbers.

Potential impact

Medium–high for the right business shape; negligible for the wrong one.

Primary KPI

Member share of revenue, and repeat visit rate among pass holders versus non-holders.

Common mistake

Measuring passes issued. Issuance is a layer-4 signal, not a result. The number that matters is whether pass holders return more often than comparable non-holders — and if you cannot construct that comparison, you cannot claim the program worked.

Example

A three-site bakery replaces paper stamp cards with wallet passes, mostly to stop reprinting them. The unexpected benefit is that it can now see, for the first time, how often individual customers visit — which is the input every other strategy in this guide requires.

When not to use it

Purely online businesses with no in-person moment; businesses whose customers buy once a year or less; anyone who needs long-form communication as the primary channel; and any business whose churn is caused by product or service problems. In that last case a wallet card just makes the departure tidier.

Go deeper: what wallet marketing is · how mobile wallet apps work · connecting the store floor to the wallet · digital membership cards

11Reactivate lapsed customers

What it does. Recovers customers who stopped buying without ever deciding to leave.

Win-back versus reactivation — and why the distinction is operational, not semantic.

Inactivity windows must be derived per category. The same customer can be four weeks lapsed at a coffee shop, perfectly normal at a hairdresser, and early at a car dealership. Segment your own purchase-interval data by category before setting any threshold. Where a business sells across categories with different natural rhythms, it needs more than one window.

Segmentation examples

These are structural examples of how to cut a lapsed list, not recommended values — the thresholds must come from your own data.

Best for

Non-contractual businesses with long or irregular purchase cycles.

Difficulty

Low once you can define lapsed defensibly.

Cost

Low.

Time to first result

Two to six weeks.

Potential impact

Medium, recurring.

Primary KPI

Reactivation rate versus a held-out control, and net incremental revenue after offer cost.

Common mistake

Using one inactivity window across every product category and customer type, which simultaneously chases people who never left and ignores people who left months ago.

Example

A bike shop finds that servicing customers return at roughly annual intervals while accessory buyers return within weeks. Running one 90-day rule across both produced a campaign that annoyed one group and missed the other entirely.

When not to use it

When you cannot identify individual customers at all. Reactivation requires knowing who has gone quiet — which is a rung-2 capability on the Attention Ladder, not a rung-1 one.

12Build a customer retention system

What it does. Connects the previous eleven strategies so that each one feeds the next, and so that no customer falls through a gap between two programs that do not know about each other.

Why disconnected tactics underperform. Three specific failure modes, all common:

• Contradiction. The loyalty program sends a reward reminder on the same day the win-back program sends a "we miss you" message, because neither knows the customer is active.
• Gaps. Onboarding ends at day 30, lifecycle messaging begins at day 60, and the customers who lapse in between are never touched by either.
• No feedback loop. The exit survey collects reasons that never reach the team that could fix them, so the same reason produces churn quarter after quarter.

The Customer Retention Flywheel™

Best for

Businesses already running three or more retention tactics that do not talk to each other.

Difficulty

High — this is an organisational problem as much as a technical one.

Cost

Medium to high; integration and ownership rather than software licences.

Time to first result

Two quarters or more.

Potential impact

Compounding — the only strategy here whose returns increase over time.

Primary KPI

The cohort retention curve, compared cohort over cohort.

Common mistake

Buying an integrated platform and calling that a system. A system is a shared definition of the customer, a shared set of stages, one owner per stage, and one place where churn reasons are counted. Software helps; it does not substitute.

Example

A subscription business assigns a named owner to each of the nine stages and requires each owner to report one number monthly. Nothing else changes for a quarter, and the arguments about which team owns churn stop.

When not to use it

When you are running zero or one retention tactic. Build something worth connecting first; a system with one component is a diagram.