Updated: 7 hours ago
The Marketing Hourglass: An Eight-Stage Retention Framework
By Trayan Hristov | Updated 13 November 2026

The Marketing Hourglass is the eight-stage model Thrivelia uses to connect acquisition with product adoption, repeat purchasing and advocacy. Its stages are Know, Like, Trust, Try, Buy, Adopt, Repeat and Advocate. The first purchase sits in the middle of the relationship, so the programme has a clear job on both sides of the sale.
Most funnel diagrams narrow towards the purchase and end there. The Hourglass makes the next part visible: what helps the customer get value from the first product, why they would return and how they might recommend the brand. It is a planning framework, not a claim that every customer moves through the same steps in a straight line.
Watch the Hourglass masterclass
Watch Trayan Hristov explain the approach at the NYC DTC Growth Summit. The 27-minute session connects list growth, personalisation, post-purchase work, retention and testing. The detailed plays and measurement definitions below turn that overview into an account plan.
Watch on YouTube · Acquisition and segmentation · Post-purchase and onboarding · Testing and iteration
The eight stages
Stage | Customer question | The programme's job |
|---|---|---|
Know | Does this brand exist and is it relevant? | Reach the right people and make the proposition understandable |
Like | Is this worth my attention? | Establish a useful first impression through product, creative and tone |
Trust | Can the brand deliver what it promises? | Provide accurate guidance, relevant proof and credible expectations |
Try | What is a sensible first step? | Help the customer take an appropriate lower-commitment action |
Buy | Am I ready to place the order? | Make the purchase decision and the handoff coherent |
Adopt | How do I get the value I bought? | Support product use, expectations, care and service needs |
Repeat | Why and when would I buy again? | Connect the product experience with a relevant next purchase |
Advocate | Would I recommend this to someone else? | Make appropriate reviews, referrals and participation easy |
Try may mean a quiz, a sample, a trial or a small first order where the business offers one. The framework does not require adding a sample programme to a brand that does not need it. The stages identify jobs; the account architecture should reflect the actual product and customer journey.
Why the shape matters
A first order is an important outcome, but it does not establish a durable customer relationship. The buyer still needs the product to meet expectations. A relevant second purchase then depends on the product, the customer's situation and the reason to return.
Customer value depends on several factors, including order frequency, order value, contribution and the length of the relationship. Acquisition brings people into that relationship. Adoption and repeat-purchase work aim to make the relationship more valuable. The appropriate budget split depends on the business and the evidence, rather than a fixed rule that retention always returns more than acquisition.
The Hourglass gives the team a shared map for those decisions. It makes a gap in product adoption or repeat purchasing as visible as a gap in welcome or checkout recovery.
The plays we use at each stage
A stage earns its place in the plan by changing what we do next. Below are the plays from our Hourglass workshops and team training, not a requirement to launch every tactic on every account. The customer state, product economics and available data decide which ones belong.
We group the work into three practical audiences: prospects across Know, Like and Trust; developing customers, or “baby whales”, across Try, Buy and Adopt; and loyalists and VIPs across Repeat and Advocate. A first purchase changes the relationship, but it does not finish the job.
Our operating playbook takes Try through the first offer and Buy through the order and early follow-up. Adopt begins as soon as the customer needs help getting value. These jobs overlap; they are not eight mandatory, isolated flows.
How to read the numbers. Targets quoted from our training are planning references, not industry benchmarks, client results or guarantees. The definitions below standardise how to measure them; the original slides did not specify every denominator or observation window.
Stage 1: Know. Capture the right interest.
The job is to turn relevant attention into a contact we can help, with enough context to make the next message useful.
Test the offer, not just the form colour. Compare an evergreen incentive with a seasonal offer, a useful free resource, store credit, or a gift. Test money off against percentage off with the margin cost visible.
Match capture to the page and traffic source. A blog reader may want a guide; a product-page visitor may need product help. Retain the landing-page context rather than sending both people into the same generic welcome.
Test the teaser, headline, CTA, delay, commitment step, number of steps and relevant gamification. Use timers only when there is a real deadline. Test a separate exit-intent form where the experience supports it.
Ask one useful question about what the visitor is trying to solve. Store the answer, source and permissions, then check that the welcome journey actually uses them.
KPIs. Form submit rate = successful submissions ÷ form views, using the same counting method for both. Also track unique new eligible contacts ÷ eligible visitors, valid-contact rate and first-purchase conversion by signup source. A high form rate cannot compensate for showing the form to almost nobody.
Training reference. 10%+ form submit rate. This is not a claim that 10% of every visitor becomes a subscriber.
Next-state signal. A valid signup with the required permission and an actionable source or preference. More fields collected is not the outcome.
Stage 2: Like. Keep the signup promise.
The job is to make the first interaction feel relevant enough that the person chooses to continue.
Personalise the welcome from the form or quiz answer. Change the product, explanation, proof or CTA, not just the first name. Use a sensible default when the answer is missing.
Mix designed emails with personal, text-led messages. Our plain-text structure uses a recognisable sender, a useful story or problem, a clear linked next step, an invitation to reply, a signature and a short P.S.
Split leads from existing customers. Someone who already bought should not keep receiving “make your first purchase” messages or an ineligible first-order offer.
Check the whole path: signup promise, first email, clicked page and next message. The creative should feel like one conversation.
KPIs. Unique clickers ÷ delivered recipients for each email; useful replies; new-lead-to-first-order conversion in an agreed window; and revenue per delivered recipient. Also compare downstream value by the answer or acquisition source.
Training references. The workshop uses 5–10%+ flow click rates and 10%+ lead conversion. The shorter playbook gives a 15%+ welcome conversion target. Those conversion targets did not specify a window, so choose one before using them. Do not report a per-email attributed conversion rate as whole-cohort welcome conversion.
Next-state signal. A meaningful product interaction, preference response or first order. An open alone is not sufficient.
Stage 3: Trust. Answer the reason they have not bought.
The job is to remove uncertainty with useful evidence, not to repeat the discount more loudly.
Use genuine reviews, customer demonstrations, accurate comparisons, product explanations, FAQs and guarantees where those address the actual objection.
Connect welcome, site or browse abandonment, cart and checkout recovery. Add back-in-stock, low-stock or price-drop messages only when the event and audience genuinely qualify.
Build follow-up around engagement recency and interests. Keep new leads visible as a distinct group; remove purchasers from first-order recovery and prevent overlapping journeys.
Give the campaign calendar a purpose beyond promotions. Our workshop starting mix is 40% educate, 25% inspire, 15% entertain and 20% convince. Examples include one benefit, how it is made, how to use it, customer proof, myths versus facts and a founder note.
Test format, sender, subject, timing, proof and creative. Frequency should follow engagement, customer state and negative feedback, not become a compulsory number of sends.
KPIs. Unique click rate, first-order conversion among eligible non-buyers, revenue per delivered recipient, and unsubscribe or complaint rates. Judge whether objection-handling changes behaviour, rather than just increasing the amount of proof shown.
Training references. 1%+ campaign click rate in the workshop and over 5% pre-purchase flow click-through in the short guide. Campaigns and flows have different intent levels, so keep those comparisons separate.
Next-state signal. A qualified trial, product decision or purchase. Route genuine questions to a human rather than another offer.
Stage 4: Try. Design a first offer worth continuing from.
The job is to make the first commitment easier without acquiring customers who only buy when the economics do not work.
Choose a hero entry product using the later customer journey, not only its first-order conversion. Look at what its buyers purchase next, when they return and their value over equal observation windows.
Test the front-end offer and the next offer together. A sample, discovery set, starter bundle, gift or small first order needs a relevant path onwards.
Separate new and returning customers in the offer and recovery logic. Record the first-order date, entry product and offer so the resulting cohorts remain measurable.
For subscription products, test whether an initial one-time purchase is the right entry point instead of forcing commitment before the customer experiences the product.
KPIs. First-purchase conversion; first-order contribution after the incentive; customer acquisition cost; and cohort contribution payback. Track the entry offer’s 30-, 60- and 90-day value alongside the initial conversion.
Training reference. The short playbook uses acquisition-cost payback below two weeks as an ambitious offer-design target. It is not a default expectation for every category. Payback means cumulative contribution covers acquisition cost, not that gross revenue equals ad spend.
Next-state signal. A completed first order or trial event. A trial that never activates is not a completed customer relationship.
Stage 5: Buy. Improve the order and the early follow-up.
The job is to help the buyer feel confident, receive the right next message and see a useful reason to purchase again.
Use relevant bundles, upsells and cross-sells to improve the order. Test the added contribution, not just a larger basket.
Start the post-purchase experience with reassurance, delivery expectations, FAQs and a brand welcome. Separate transactional information from optional promotional messages.
Use first-product and product-sequence data to choose the next recommendation. Start with a few important entry products and a fallback rather than hundreds of thinly populated branches.
Test the timing of an early next-order or bounce-back offer against product arrival and observed purchase behaviour. Introduce community, loyalty or another opted-in channel only when it adds value.
Stop the first-purchase recovery messages as soon as the order is placed. Avoid treating an active subscriber’s renewal as a brand-new first purchase.
KPIs. Average order value and contribution per order; revenue or contribution per new customer at 30, 60 and 90 days; first-to-second purchase rate; and time to second order. Report order-value gains alongside refunds and discount cost.
Training references. The short guide uses a 15–20% first-to-second purchase target within 60 days. The workshop separately says “50%+ of repeat orders <30 days”. That is a repeat-order timing target, not a 50% repeat rate across all first-time buyers. Define which repeat orders and time interval are being counted before applying it.
Next-state signal. A correctly handed-off first purchase, followed by successful use or a relevant next order. Do not manufacture a second order by splitting one checkout into two.
Stage 6: Adopt. Help them get what they bought.
The job is to turn ownership into use, confidence and habit. A product cannot earn a lasting relationship if it stays unused.
Send product-specific setup, care, recipes, usage guidance or onboarding. Match the sequence to delivery or activation rather than assuming that purchase means the product is in use.
Build useful milestones. The subscription training explores day 1, 3, 7 and 14 guidance, with content adapted to the product and term. Those are example intervals, not promises of a health or product result.
Ask about the experience and route problems to support. Refund, return or dissatisfaction signals should change the next action; do not send an upsell over an unresolved issue.
Keep one-time buyers, active subscribers and cancelled subscribers on different paths. Teach active subscribers how to manage their plan and offer appropriate add-ons, pauses or cadence changes without obstructing cancellation.
Use observed usage, replenishment needs or a meaningful activity recap to prompt the next useful action. A lack of email clicks is not proof that someone has not used a physical product.
KPIs. A defined activation or successful-use event where observable; first-renewal retention; time to first value; return/refund rate; and support or experience feedback. For physical products, separate confirmed behaviour from survey-based proxies.
Training references. The developing-customer workshop includes returns below 10% and public rating above 4.5. Product type, return policy, sample size and review platform matter. Collect honest feedback; never make a reward conditional on a positive public review.
Next-state signal. Evidence of useful adoption, or an issue that needs resolving. A subscription signup is not the same as a successful first renewal.
Stage 7: Repeat. Build the next purchase around the customer.
The job is to grow the number and value of useful purchases over the relationship, not simply send another campaign.
Analyse the sequence from first to second to third product, and the intervals between orders. Our coaching extends product-specific follow-up into 60-, 90- and 120-day windows when customer behaviour supports it.
Use recency, frequency and monetary value together. High-spend frequent buyers need recognition and access; high-spend first buyers need a reason to return; frequent lower-spend buyers may benefit from a relevant bundle or threshold gift.
Time replenishment, expected-next-order and win-back journeys around observed patterns. A lapsed valuable customer needs a different approach from a new subscriber who never purchased.
For subscriptions, distinguish plan retention from additional purchases. Offer relevant add-ons, a more suitable quantity or term, and milestone benefits where the economics work. Keep subscriber value clear instead of undercutting it with constant one-time discounts.
Continue collecting preferences after purchase, through a short survey or a click inside an email. Combine those answers with actual orders before changing the recommendation.
KPIs. Equal-window repeat purchase rate; second-to-third order progression; orders per customer; time between purchases; returning-order revenue; and realised revenue or contribution per customer at a fixed horizon. For subscriptions, also track successful renewal rate, recurring order value and reactivation.
Training references. The longer workshop lists 35–50% repeat-purchase revenue and 20%+ repeat-purchase customers. These are account-planning references with incomplete original definitions, not measured lift. Choose a period and denominator first. A revenue share, a share of buyers and a cohort repeat rate are three different measures.
Next-state signal. Another eligible order or successful renewal. Do not count a scheduled bill, failed payment or replacement shipment as a successful commercial repeat.
Stage 8: Advocate. Make recommending the brand worthwhile.
The job is to help satisfied customers share their experience and bring the right people into the top of the Hourglass.
Use VIP tiers, early access, thoughtful surprises, anniversary or birthday recognition and personal thank-you messages to acknowledge the relationship.
Create a clear referral path with an understandable benefit and an easy way to share. Track the new customer through to an order rather than stopping at referral-link clicks.
Invite appropriate customers into community, ambassador, affiliate or product-testing programmes. Those are distinct programmes, not interchangeable names for a discount code.
Ask for honest reviews and permission to reuse customer content. Put any incentive disclosures in place and route service issues for resolution without screening out negative reviewers.
KPIs. Referral participation, referred visitors, referred first orders, referral conversion and contribution after rewards. Also track review-request response and usable, permissioned customer content.
Training references. The longer workshop gives 5%+ referral traffic; the short playbook gives 2.5–5% referral conversion. The first is a traffic share, the second a conversion rate. They cannot be substituted for each other.
Next-state signal. A review, a permitted piece of customer content, a qualified introduction or a referred order. Advocacy should feed acquisition without erasing its cost.
How we use it on an account
An account review maps existing flows, campaigns and service handoffs onto the stages. We inspect both coverage and coherence. A programme can have several welcome flows while leaving the first buyer with little product guidance. It can also have a large number of flows whose entry and exit rules compete for the same person.
Before purchase, the work may involve a clearer proposition, useful product advice, preference capture or a better welcome experience. After purchase, it may involve education by product, support routing and a next-order journey based on observed customer behaviour.
The map does not prescribe a flow for every box. It helps identify which customer transition needs work and what evidence should decide the priority. Use the Klaviyo audit checklist to review the data, delivery and logic that make the map usable.
A worked example: the first apparel purchase
This example is hypothetical. It shows how to turn the framework into decisions, rather than claiming a measured result for a specific client.
A new subscriber is interested in a core top. The pre-purchase programme explains fabric, fit and styling, then helps the shopper choose. When an order occurs, first-purchase recovery stops and the relationship changes.
After arrival, the programme supplies useful care and styling advice. The team analyses mature first-product cohorts to find a plausible companion purchase and an appropriate observation window. It then tests whether a relevant follow-up helps more eligible customers place a second order.
Transition | Action to investigate | Evidence to review |
|---|---|---|
Trust to Buy | Improve accurate fit and product guidance | The chosen first-purchase outcome and a fair comparison |
Buy to Adopt | Provide useful arrival-stage education | Timing, product context and the defined customer outcome |
Adopt to Repeat | Present a relevant companion product | Equal-age first-order cohorts and second-purchase behaviour |
Repeat to Advocate | Offer an appropriate review or referral opportunity | Eligibility, customer context and the actual response |
The practical difference is the focus on a transition. A sequence exists to help the customer do something useful, rather than simply to fill a missing automation name.
Define the next decision before building
For each proposed piece of work, record the stage, customer need, required information, owner and outcome. Check whether the underlying events, product information and permissions support the plan. Define a sensible fallback when the information is missing.
Then distinguish repairs from experiments. Incorrect purchase exclusions are a defect to fix and retest. A new product-education sequence is a hypothesis whose outcome needs evaluating. Give each a clear place in the action register.
Order the work around the account's condition. Reliable sending, data and journey logic support the later creative and personalisation work. More messages are useful only when they serve the agreed priorities and remain coherent with the rest of the customer's experience.
Measure the transition, not just the channel
Use attributed revenue to understand which messages receive credit under the configured reporting model. Use customer measures to investigate whether new buyers return, when they return and which products appear next. These questions are related, but their answers are not interchangeable.
The perfectwhitetee case reports a 36.3% blended second-purchase figure and 61.5% among customers with at least twelve months of tenure. Those are different populations. The finding explains why customer age matters in diagnosis; it is not an email-caused improvement from one percentage to the other.
Our first-to-second-purchase guide provides an equal-window calculation. The email revenue share guide explains how attribution and total-store performance should be reviewed together.
Use a scorecard with explicit denominators
Before setting a target, write down who enters the measure, what counts as success, the observation window and the reporting source. The workshop numbers above are starting points for a discussion, not permission to mix unlike metrics.
Lead activation. For a first-order objective: eligible new non-buyer leads who place their first order within the chosen window ÷ all eligible new non-buyer leads with that much follow-up time. Choose, for example, 30 days and keep it fixed. Separate this customer outcome from platform-attributed email conversions.
First-to-second purchase. First-time buyers who place a second qualifying order within 60 days ÷ all first-time buyers in the cohort with at least 60 days of observation. Use the same rule for 30, 90 or 365 days. Define how cancelled, fully refunded and replacement orders are handled before comparing cohorts.
Revenue per new customer. Net order revenue accumulated by a first-order cohort through day 30 ÷ the number of customers in that cohort. Repeat at 60, 90 and 365 days. For contribution per customer, deduct the agreed product, fulfilment, payment and intervention costs. Do not label a 30-day value as lifetime value.
Returning-order revenue share. For this scorecard: net revenue from second-and-later qualifying orders placed in the reporting period ÷ total net order revenue in that period. Keep the raw amounts next to the percentage. A rising share can reflect weaker new-customer sales rather than healthier retention.
Repeat-customer share versus cohort repeat. A period-based share of purchasing customers with two or more qualifying orders is not the same as the probability that a first-time buyer repurchases within 60 or 365 days. Specify whether prior orders outside the reporting period count.
Referral traffic versus conversion. Referral traffic share = qualified referred sessions ÷ all sessions in the same period. Referral conversion = referred visitors who make a qualifying purchase ÷ referred visitors with the agreed conversion window. Keep referral definitions, deduplication and tracking consistent.
Every test also needs guardrails: unsubscribe and complaint rates, refunds, incentive cost and contribution. Use a comparable cohort or a randomised holdout where feasible before calling an observed improvement incremental. A correlation between a product sequence and high value is a hypothesis to test, not proof that forcing the sequence creates the value.
The subscription scorecard
The subscription training adds a separate view: active subscribers; net additions after new starts, reactivations and cancellations; first-renewal retention; retention at later renewal numbers; voluntary and payment-failure churn; average recurring order value; customer lifespan; and realised revenue or contribution per subscriber.
Compare customers with the same opportunity to renew. Define first-renewal retention as initial subscribers with a successful first renewal by the agreed cut-off, including a defined grace period, divided by initial subscribers whose first renewal was due. Report skips and pauses separately. A cohort that has not reached its next billing date cannot be treated as churned or retained at that renewal.
Segment recovery by the actual reason: too much product, cost, difficulty using it, failed payment or another stated issue. Keep required billing, renewal and cancellation information intact. Test optional support and merchandising around it, not whether to hide important account information.
Connect the map with progressive personalisation
The Hourglass identifies the relationship stage and the job to be done. Progressive personalisation changes the next interaction using what the brand learns about the customer.
Purchased product, order count, timing and a reliable stated preference can change the advice or next product. Keep the decisions understandable and the information current. A more elaborate customer profile is useful when it produces a better action.
How progressive personalisation changes the play
The team-coaching rule is simple: asking for data is only the start. First, use the stated need in the welcome. Second, join it to the first product, next products and purchase intervals. Third, learn which combinations are associated with stronger long-term value. Fourth, ask what the customer needs next and adjust the journey.
Do not automatically optimise for the biggest quiz segment, the product featured in the email or the largest first basket. Compare actual purchases and equal-age cohort value. A customer may click a new collection but buy an established bestseller. Start with a few meaningful product paths, then expand when volume and evidence justify the extra complexity.
Turn the Hourglass into a cross-team decision brief
A useful Hourglass review ends with one customer problem the business can act on. Bring the CRM owner, ecommerce lead and customer-service lead into the same discussion. Start with what people are trying to do, then identify the messages, store changes and service responses needed to help them. A customer does not experience those as separate departments.
Consider this hypothetical homeware example. New buyers regularly ask whether their bedding will shrink and how to wash it. The care instructions exist, but are difficult to find. The gap sits between Buy and Adopt. Sending an earlier cross-sell would add another message while leaving that uncertainty unresolved.
The shared brief could be: help first-time bedding buyers find accurate care guidance when their order arrives. Ecommerce owns the product-care page. Support checks that the advice answers the questions customers actually ask. CRM owns the eligible audience, timing and message. Operations confirms whether delivery data is reliable enough to use; if it is not, the team documents an estimated-arrival fallback and checks the wording against that uncertainty.
Write down the decision fields before producing creative: the customer group; observed problem and evidence; responsible owner; necessary data; proposed change; primary outcome; guardrails; and review date. Keep a verified issue separate from a proposed explanation. Ten support conversations can reveal a useful theme, but they do not establish how common that problem is across all buyers.
For this example, a click on the care page measures use of the guidance, not successful product adoption. Review relevant support reasons and return reasons alongside it, with consistent definitions. Then assess second-purchase behaviour over an equal observation window. If a commercial test is feasible, compare the proposed education with the existing experience while keeping unrelated communication comparable.
Prioritise the work by the consequence of leaving the gap open, the number of affected customers and the effort required to change it. Repair inaccurate information or broken routing first. Test a plausible new education approach next. A new loyalty tier can wait if the first product experience is still confusing. The framework earns its place when it changes that order of work.
The monthly testing rhythm
Our workshop sets a simple operating cadence: one activation test, one repeat-purchase test and one customer-value test each month. That is 36 planned tests over a year, not a promise of 36 statistically decisive winners. An account without sufficient volume may need a longer read window.
Activation: test whether using a stated need in the welcome improves first-order conversion against the current version. Check lead quality, margin and opt-outs as well as clicks.
Repeat: test a relevant second-product recommendation and timing against the current follow-up. Read the same 60-day customer window for each group.
Customer value: test a VIP benefit, bundle, subscriber add-on or renewal-support sequence. Choose contribution per customer or another defined commercial outcome before launch.
For each test, record the stage, eligible audience, hypothesis, owner, comparison, primary KPI, guardrails and decision date. Fix broken tracking and contradictory entry or exit rules first. Start with one valuable transition and ship the improvement, rather than rebuilding the entire account at once.
Watch, read and put it into practice
The public starting points are the NYC DTC Summit masterclass and Retention. Perfected. Hourglass playbook. This expanded guide also draws on our detailed Hourglass workshop and internal team training. Training targets remain separate from measured client results.
Frequently asked questions
Does every stage need a separate Klaviyo flow?
No. One journey may serve several stages, and some stages involve the website or service team rather than marketing automation. Build the architecture around customer needs and reliable events.
Is the Hourglass only for repeat-purchase products?
The reasons and intervals for returning differ by product. A durable-goods brand may emphasise product adoption, accessories, another use case or advocacy. Adapt the jobs and measures to the actual business.
How should a subscription brand use it?
Review product adoption and progression through billing cycles alongside acquisition. Define what counts as a successful rebill, and compare customers with equal time to reach it. Do not treat a first subscription order as proof of a lasting relationship.
Where should we start?
Map the current experience, identify the transition that needs attention and check the evidence supporting the diagnosis. Then assign the work and a review point. The framework is useful when it changes a practical decision.
Acquisition fills the top; the relationship continues
Choose the transition in the Hourglass where you lose the most customers you could realistically retain. Put the person responsible for product experience in the room with the marketer. Decide what needs to change for one customer and how the team will recognise it in the data.

.png)



Comments