Ecommerce CRM for Customer Retention: 10 Ways to Keep Buyers Coming Back

Some links on The Justifiable are affiliate links, meaning we may earn a small commission at no extra cost to you. Read full disclaimer.

An ecommerce CRM for customer retention helps you turn scattered purchase, browsing, support, and engagement data into repeat-purchase opportunities. The challenge is not simply collecting more customer information; it is knowing what to do with it after the first order.

A useful CRM should help you recognize customer intent, personalize follow-up, coordinate channels, spot churn risk, and measure whether retention is actually improving.

This guide explains 10 practical ways to build that system, from data preparation and lifecycle segmentation to loyalty, service recovery, win-back campaigns, and retention measurement, without burying customers in unnecessary messages.

How Ecommerce CRM Turns Customer Data Into Retention

A retention-focused CRM connects customer history with the next action your team should take. Before automating, define what it should know and trigger.

What Makes an Ecommerce CRM Different From a Basic Contact List?

A contact list mainly tells you who a customer is and how to reach them. An ecommerce CRM should also tell you what they bought, when they bought it, how often they return, which messages they engage with, and whether there are signs that the relationship is strengthening or weakening.

That difference matters because retention decisions depend on behavior. A customer who bought once yesterday should not receive the same message as a customer who has placed six orders and has not returned for four months. The CRM gives you the context needed to separate those situations.

At minimum, useful customer records should connect identity, order history, product history, marketing permissions, campaign engagement, and important service events. More mature setups may also include browsing behavior, loyalty status, subscription information, or predicted customer value.

The goal is not to collect every possible data point. Collect the information that changes what you do. If a field does not influence segmentation, personalization, service, measurement, or a business decision, it may not deserve priority.

Where Retention Actually Happens In The Customer Lifecycle

Retention begins immediately after the first purchase, not when a customer is already close to churning. The period between the first and second order is especially important because you are trying to convert a transaction into a relationship.

A practical lifecycle usually includes first-time buyers, active repeat buyers, high-value or loyal customers, customers showing declining engagement, and lapsed customers. Your exact definitions should reflect how frequently people naturally buy in your category. A coffee subscription business and a furniture store should not use the same inactivity window.

The CRM’s role is to detect movement between these stages and trigger an appropriate response. A first-time buyer may need product education. A repeat customer may respond better to complementary recommendations. A loyal customer may value recognition or early access. A lapsed customer may need a reason to reconsider the brand.

When you design retention around lifecycle movement, campaigns become less random. You stop asking, “What should we send this week?” and start asking, “What does this customer need next?”

When A Retention CRM Is Worth The Complexity

You do not need an elaborate CRM stack on day one. A small store with a limited catalog and a manageable customer base may get most of the value from reliable order data, a few well-built segments, and simple email automation.

Complexity becomes more useful when customer behavior becomes harder to track manually. Warning signs include multiple marketing channels, a growing number of repeat buyers, inconsistent customer service context, several product categories with different repurchase cycles, or uncertainty about which customers deserve different treatment.

I recommend adding tools only when a clear retention problem justifies them. If you cannot explain the decision a new tool will improve, the tool is likely to create more data than value.

A good CRM setup should reduce manual work and sharpen decisions. It should not force your team to maintain overlapping customer databases, duplicate segments, and competing automation rules. Start with the smallest system that can reliably recognize customer state and act on it.

Prepare Your Retention Foundation Before Automating

Automation amplifies its foundation. Clean data, realistic lifecycle definitions, and clear retention goals prevent technically correct but strategically poor messages.

Choose The Customer Data That Should Stay In Sync

Start by mapping the systems that already hold important customer information. Your commerce platform may own orders and products, your marketing platform may own consent and engagement, your helpdesk may own service history, and a loyalty platform may own points or tier status.

Decide which system is the source of truth for each type of data. Then make sure customer identifiers are consistent enough to connect records across tools. Email address is common, but phone number, customer ID, or platform-specific IDs may also matter depending on your stack.

Prioritize data that supports retention decisions: first and most recent order dates, number of orders, products purchased, average order value, refunds, subscription status, support events, and communication consent. Browsing or predictive data can be useful later, but incomplete order and identity data will undermine more advanced personalization.

Test integrations with real scenarios before launching broad automations. Place a test order, update a subscription preference, open a support ticket, and confirm that the expected CRM fields change. Retention logic is only as trustworthy as the data feeding it.

Define Your Natural Repurchase Cycle And Retention Goal

A generic 30-, 60-, or 90-day retention window can be misleading. Instead, estimate when customers normally need or want to buy again. Consumables may have a predictable replenishment rhythm, while fashion, electronics, or home goods often require more varied triggers.

Review historical order gaps if you have enough data. Look at the time between first and second purchase, then compare it with later purchases. You are not looking for one perfect number; you are looking for a useful range that helps separate “not ready yet” from “starting to lapse.”

Next, choose the retention outcome you want the CRM to influence. That might be second-purchase rate, repeat purchase rate, purchase frequency, subscription retention, or revenue from returning customers. Avoid making “send more automated messages” the goal.

A hypothetical example: if most second purchases occur between days 25 and 45, a win-back offer on day 14 is probably premature. Product education around day 7 and a relevant recommendation around day 30 may better match customer readiness.

Build Segments Around Behavior Rather Than Demographics Alone

Demographic information can be useful, but behavioral segmentation usually gives you stronger retention signals. What customers buy, how recently they bought, how often they return, and what they engage with are directly connected to future purchase opportunities.

Start with a small set of segments your team can actually use. Useful examples include first-time buyers, second-time buyers, repeat customers, high-value customers, recent buyers who have not purchased again within the expected cycle, discount-heavy buyers, and customers with recent service issues.

Platforms such as Klaviyo can build dynamic segments from profile and event data and use those segments in automated flows. Omnisend is another ecommerce-focused option for teams that want segmentation, lifecycle targeting, and email/SMS automation in one marketing platform.

Do not create dozens of microsegments before you know how they will be used. A segment earns its place when it changes the message, timing, offer, channel, or service experience.

Build Personalized Lifecycle Journeys

Once the foundation is stable, the CRM can continuously recognize customer state and deliver the next appropriate experience. These first three tactics create that system.

Way 1: Centralize Customer Profiles Around Meaningful Events

A unified customer profile is useful because retention rarely depends on one event. The best next action often comes from the sequence: first order, product delivered, email clicked, second product viewed, support question opened, and no repeat purchase yet.

Centralization does not necessarily mean moving every piece of data into one database. It means your key systems can share enough context that automation and service decisions are based on the same customer history.

Define a small event vocabulary first. Examples might include placed order, ordered product, refunded order, viewed product, started checkout, subscription paused, loyalty reward earned, or support case resolved. Use consistent naming and avoid creating multiple versions of the same event from different integrations.

Then decide which events should trigger action. A refund might suppress an upsell sequence. A second purchase might remove someone from a first-buyer nurture flow. A support escalation might temporarily pause promotional messages.

This is the first way a CRM improves retention: it replaces isolated messages with decisions informed by customer history.

Way 2: Segment Customers By Lifecycle, Value, And Intent

Segmentation becomes powerful when you combine three questions: Where is this customer in the lifecycle? How valuable is the relationship so far? What behavior suggests they may want next?

For example, two customers may both have placed three orders, but one ordered last week while the other has been inactive for six months. They should not receive the same retention message. Add product category, engagement, or average order value and the differences become even more useful.

Use simple rules at first. A recent repeat buyer may enter a loyalty-oriented segment. A first-time buyer who browses a related category may enter a cross-sell segment. A previously frequent buyer who has missed their normal purchase window may enter an at-risk segment.

Review segment sizes and logic regularly. If a supposedly exclusive VIP segment includes half your list, the criteria may be too loose. If an at-risk segment never changes, the timing may not reflect real purchase behavior.

The purpose of segmentation is not precision for its own sake. It is to make the next action more relevant.

Way 3: Automate The Post-Purchase Path To A Second Order

The first purchase creates permission to be helpful, but the most effective post-purchase journey is not an immediate stream of discounts. It should first help the customer get value from what they already bought.

Build a sequence around delivery and expected product use. Early messages can explain setup, care, sizing, usage, or common questions. Later messages can request feedback, introduce complementary products, or remind the customer when a logical repurchase window approaches.

Use order data to branch the journey. Someone who bought a starter kit may need education before an accessory recommendation. Someone who bought a replenishable product may need a reminder. Someone who already placed a second order should exit the first-purchase sequence.

Klaviyo flows can trigger from customer actions and profile or segment conditions, which makes it well suited to this kind of branching lifecycle automation. The important part, however, is the logic rather than the tool.

A strong post-purchase flow reduces buyer uncertainty first and sells second. That order supports trust and makes the next recommendation easier to justify.

Create More Reasons For Customers To Return

Retention becomes stronger when returning is genuinely useful, not merely cheaper. These tactics make future purchases more relevant, timely, and rewarding.

Way 4: Personalize Recommendations From Purchase Context

Product recommendations should answer a customer need created by the previous purchase. “You bought this, so here is the next logical product” is usually more useful than “Here are our best sellers.”

Start with product relationships you can explain. Accessories can support a main product, refills can replace consumables, complementary items can complete a routine, and upgraded versions may make sense after a customer has demonstrated repeated use. Avoid recommending the exact item a customer just bought unless replenishment is plausible.

Your CRM should use order history and, when appropriate, browsing behavior to narrow the options. You can also suppress recommendations that conflict with returns, service issues, out-of-stock items, or recent purchases.

A hypothetical skincare store might recommend a moisturizer after a customer has had time to use a cleanser, rather than pitching an unrelated product immediately after checkout. A pet-supply store might time food or litter recommendations around expected depletion.

Good personalization does not require hundreds of rules. A few well-designed product relationships often outperform broad “recommended for you” blocks that do not reflect real context.

Way 5: Use Replenishment And Subscription Signals At The Right Time

If your products are consumed, replaced, renewed, or reordered, the CRM can reduce the effort required to come back. The most important variable is timing. A reminder sent long before the customer needs the product feels promotional; one sent after the customer has already run out feels late.

Estimate replenishment windows by product, size, quantity, or customer order history when possible. Start with a conservative default, then adjust as you learn from repeat-purchase behavior. Give customers a simple path back to the relevant product instead of making them search the store again.

Subscription businesses should treat pauses, skips, failed payments, and cancellations as different signals. A customer who pauses may still be satisfied but temporarily oversupplied. The response should reflect that context rather than dropping everyone into the same win-back flow.

You do not need to force subscriptions on every category. They work best when recurring need is real and the customer benefits from convenience. Retention improves when the program removes friction, not when it creates commitment that customers later regret.

Way 6: Reward Loyalty Without Training Customers To Wait For Discounts

Loyalty programs can create a reason to consolidate purchases with one brand, but the design matters. If every reward is simply another percentage discount, you can reduce margin and teach customers to delay full-price purchases.

Use the CRM to recognize behaviors you actually value. Repeat purchases, referrals, reviews, membership milestones, and high-value engagement may all deserve different treatment. Rewards can include points, early access, member-only products, free gifts, priority service, or special experiences where appropriate.

LoyaltyLion is a relevant option for ecommerce brands that want structured points, rewards, and loyalty tiers that can feed into the wider retention experience. It is more useful when you have enough repeat purchase activity to justify a program; a very early store may get more value from improving post-purchase communication first.

Connect loyalty status to the CRM rather than treating the program as a separate island. A high-tier customer should be recognized in marketing and, ideally, in service interactions too.

A loyalty program is strongest when it recognizes an existing relationship, not when it tries to manufacture loyalty with endless discounts.

Coordinate Email, SMS, And Customer Service

Customers experience one brand even when your team uses several tools. Coordination prevents convenience from turning into channel overload or contradictory communication.

Way 7: Coordinate Email And SMS Around Customer Preference

Email and SMS can reinforce one another, but sending the same promotion through both channels is not coordination. Use the CRM to decide which channel fits the message, the customer’s consent, and the urgency of the situation.

Email works well for richer education, product guidance, stories, and detailed recommendations. SMS can be useful for concise, time-sensitive communication when the customer has explicitly opted in and the message deserves that level of immediacy. Your exact legal and consent obligations depend on where you operate and where recipients are located, so your setup should reflect applicable rules.

A platform such as Omnisend can manage email and SMS within ecommerce automation, while Klaviyo can also coordinate behavior-driven messaging across customer data and flows. Choose based on your store, integrations, team complexity, and reporting needs rather than feature count alone.

Set channel suppression rules. If a customer already converted from an email, they should not receive a redundant SMS reminder. Coordination should reduce noise while preserving useful touchpoints.

Way 8: Turn Customer Service Into A Retention Signal

Support interactions often contain the clearest explanation of why a customer may or may not return. A delivery problem, confusing product setup, wrong size, damaged item, or slow resolution can be more predictive of churn than an unopened marketing email.

Connect customer service context to the CRM so marketing does not ignore unresolved problems. At minimum, consider pausing aggressive promotions while a serious issue is open. After resolution, the customer may need reassurance, product education, or a service-recovery follow-up rather than an immediate sales pitch.

Gorgias is designed as an ecommerce helpdesk that brings customer conversations and commerce context together, which can help support teams see order information while handling issues. It becomes especially useful when ticket volume is high enough that agents can no longer reconstruct customer history manually.

Track recurring ticket themes as retention problems, not just support metrics. If customers repeatedly ask the same question after purchase, improve the product page, onboarding message, packaging, or knowledge base. The best service automation prevents avoidable friction upstream.

Protect Frequency, Consent, And Customer Attention

A CRM makes it easy to send more messages, which is exactly why you need rules that prevent overcommunication. A customer can qualify for a post-purchase flow, a loyalty campaign, a product launch, and a win-back segment at the same time if your logic is not coordinated.

Create a simple priority system. Transactional and service communication should usually outrank promotional messaging. High-intent behavior may justify a relevant message, while low-priority campaigns can wait. Use exclusions so customers do not receive conflicting offers or repeated reminders across overlapping flows.

Frequency caps can help, but context matters more than a universal number. A customer actively shopping may tolerate more relevant communication than someone who has ignored several campaigns. Engagement should influence how aggressively you continue.

Consent should be stored clearly by channel, and unsubscribe or opt-out actions must flow through your stack reliably. Do not treat a communication preference as a technical checkbox.

The retention goal is sustained permission to communicate. Preserving attention is more valuable than maximizing the number of sends this week.

Recover At-Risk And Lapsed Customers

Even strong retention systems lose customers. Detect meaningful decline early and use win-back campaigns selectively rather than discounting every inactive buyer.

Way 9: Detect Churn Risk Before A Customer Fully Disappears

Churn risk should be based on expected behavior, not arbitrary inactivity. If a customer normally orders every 30 days and reaches day 50 without purchasing, that matters more than a 50-day gap for someone who buys twice per year.

Build at-risk logic from recency, previous frequency, customer value, product type, engagement, and subscription or service signals where available. A previously frequent customer who suddenly stops is often more interesting than a one-time buyer who never showed repeat intent.

RetentionX can help ecommerce teams analyze cohorts and customer segments, making it useful when you need to see whether repeat behavior differs by acquisition period or customer group. This kind of analysis can reveal whether the issue is broad or concentrated in a particular cohort.

Do not assume every at-risk customer needs a coupon. First determine whether the likely barrier is timing, product fit, service friction, price sensitivity, or simple lack of need. Better churn detection improves both who you contact and what you say.

Way 10: Build Win-Back Sequences Around A Reason To Return

A win-back campaign works best when it gives the customer a relevant reason to reconsider the brand. “We miss you” may be friendly, but it does not answer why the customer should buy again now.

Start with context. Remind the customer of products they previously liked, show what has changed, introduce a new use case, highlight replenishment need, or offer help if the category requires guidance. If the customer had a service problem, do not pretend that history does not exist.

Use a short sequence rather than one generic blast. The first message can re-establish relevance. A second can address likely hesitation or introduce a specific recommendation. An incentive, if you use one, can come later for customers who have not responded.

Define an exit condition. Once the customer purchases, they should leave the win-back sequence and enter the appropriate post-purchase or repeat-buyer journey.

Also define when to stop. Continuing to contact chronically disengaged customers can waste budget and hurt channel performance. Retention includes knowing when not to pursue.

Use Incentives As A Diagnostic Tool, Not A Default Habit

Discounts can recover some customers, but they can also hide a weak value proposition. If a segment only returns when given increasingly large offers, the CRM may be measuring promotion sensitivity rather than genuine loyalty.

Test incentives against non-discount alternatives. A free gift, loyalty credit, shipping benefit, bundle, early access, personalized recommendation, or helpful service may create a stronger reason to return without reducing price directly.

Use customer history to protect margin. A high-value customer with strong full-price behavior may not need the same offer as a highly price-sensitive buyer. Likewise, a customer who recently purchased should not accidentally receive a stronger win-back discount than a customer who is actually lapsed.

Create holdout groups when possible so you can see whether the incentive caused incremental purchases or simply rewarded customers who would have returned anyway.

The practical rule is simple: use an offer when it solves a specific retention barrier. Do not make discounting the automatic response to silence.

Avoid CRM Retention Mistakes And Troubleshoot Weak Results

When retention automation underperforms, investigate data quality, timing, overlapping logic, and customer value before assuming the message copy is the problem.

Fix Over-Automation Before Adding More Campaigns

A large automation library can look sophisticated while creating a poor customer experience. If every behavior triggers a message, customers receive too many nudges and your team loses track of which flow is responsible for performance.

Audit all active journeys from the customer’s perspective. List each trigger, delay, channel, exclusion, and exit condition. Then test common scenarios: first purchase, repeat purchase, refund, subscription pause, support issue, loyalty upgrade, and inactivity. Look for points where multiple flows collide.

Remove automations that do not have a clear job. A useful flow should educate, reduce friction, recommend, recover, recognize, or collect information that improves the next decision. If its purpose is simply “stay top of mind,” consider whether a regular campaign can do the job more cleanly.

More automation is not the same as more personalization. The best systems often have fewer, stronger journeys with better data and exclusions.

When results weaken, simplify first. You need a system you can explain before you can optimize it.

Repair Duplicate Profiles, Missing Events, And Broken Integrations

Bad CRM data creates silent retention failures. A duplicate customer may receive the same campaign twice. A missing purchase event can leave a buyer inside an abandoned-cart flow. An integration delay can trigger replenishment or win-back messages at the wrong time.

Create routine data checks. Compare order counts between your store and CRM, inspect recent event timestamps, check consent status, and sample customer profiles across different journeys. Pay special attention after changing ecommerce themes, checkout systems, apps, tracking scripts, or integrations.

Define how duplicates are handled and which identifier wins when records conflict. Make sure refunds, cancellations, and subscription changes sync as reliably as successful orders; otherwise your CRM sees an unrealistically positive customer history.

If a retention metric suddenly changes, investigate instrumentation before assuming customer behavior changed. A broken tracking event can make a campaign appear to lose performance overnight.

Treat CRM data maintenance as operational work, not a one-time setup task. Reliable retention automation depends on ongoing trust in the underlying record.

Correct Poor Timing And Segment Overlap

Weak performance often comes from using the right idea at the wrong time. Replenishment reminders before customers have used the product, win-back campaigns before the normal purchase cycle ends, and loyalty pushes immediately after a complaint all create friction.

Compare message timing with actual behavior. Look at days between purchases, average delivery time, product usage cycle, and the time customers typically take before buying complementary items. Adjust delays by category where differences are meaningful.

Then inspect segment overlap. A high-value customer can also be at risk. A first-time buyer can also qualify as discount-sensitive. Decide which status should take priority and which messages should be suppressed.

You can solve many conflicts with a simple hierarchy: active service issues first, transactional needs second, high-intent lifecycle actions next, and broad promotional campaigns last. The exact order can vary, but it should be deliberate.

If a campaign underperforms, do not immediately rewrite the subject line. First confirm that the right customer received the right message at a plausible moment.

Measure Retention And Scale What Works

Measurement should reveal which journeys create value, which segments are deteriorating, and where additional CRM complexity is actually justified.

Track Metrics That Reflect Repeat Behavior

Open rates and clicks can help diagnose messaging, but they are not retention outcomes. Start with measures that reflect whether customers continue buying.

Useful metrics include second-purchase rate, repeat purchase rate, time to second order, purchase frequency, returning-customer revenue, retention by cohort, and customer lifetime value where your data quality supports it. Subscription businesses may also track renewal, pause, and churn behavior.

Avoid reading these metrics only as storewide averages. A healthy overall repeat rate can hide a weak new cohort, and a strong cohort can mask deterioration among customers acquired from a specific campaign or product.

Choose one or two primary metrics for each retention initiative. A post-purchase education flow might be evaluated against second-purchase rate and time to second order. A loyalty initiative may be assessed through repeat frequency or value among members, while accounting for the fact that loyal customers may be more likely to join in the first place.

Measurement should clarify decisions, not create a dashboard nobody uses.

Use Cohorts, Holdouts, And Controlled Tests

Retention changes slowly, which makes casual before-and-after comparisons risky. Seasonality, promotions, acquisition mix, product launches, and inventory can all change at the same time as your CRM program.

Cohort analysis helps by grouping customers according to when or how they were acquired, then comparing their repeat behavior over time. It can show whether newer customers are becoming more or less valuable than earlier groups.

Where your tools and traffic allow it, use holdout groups for important automations or offers. A holdout group does not receive the intervention, giving you a better estimate of incremental impact. This is especially useful for discounts, because some customers would have purchased without the incentive.

Test one meaningful variable at a time when possible: timing, recommendation logic, offer type, channel, or sequence length. Small cosmetic changes may improve engagement without changing retention.

Scaling should follow evidence. A tactic that produces more clicks but no improvement in repeat purchasing does not deserve more complexity simply because the campaign looks active.

Choose Tools According To The Retention Bottleneck

Your stack should reflect the problem you are trying to solve. If lifecycle marketing is weak, a platform such as Klaviyo or Omnisend may be the most direct improvement. If support context is fragmented, Gorgias may solve a different bottleneck. LoyaltyLion is relevant when a structured rewards program supports real repeat behavior, while RetentionX becomes more useful when analysis and cohort visibility are limiting decisions.

Do not buy all of these at once. Each tool adds cost, integration work, reporting complexity, and another place where customer data can diverge.

A useful decision sequence is:

• Messaging problem: Improve segmentation, flows, and channel coordination.
• Service problem: Connect order context with support and use ticket themes to reduce friction.
• Loyalty problem: Add structured recognition only after the repeat-purchase experience is sound.
• Insight problem: Add cohort and customer intelligence when spreadsheets no longer answer retention questions efficiently.

The best ecommerce CRM for customer retention is often a connected stack, not a single product. But every component should have a clear owner and measurable job.

Scale Proven Journeys Before Expanding Complexity

Once a journey works, scale it carefully rather than cloning it across every segment. First confirm that the underlying behavior is repeatable across cohorts and that the economics still make sense when volume grows.

Expand proven logic by product category, customer value, geography, or channel only when differences justify the added branches. A replenishment model may need product-specific timing, while a loyalty message may only need a few value tiers.

Document the rules behind successful automations. Record the trigger, audience, exclusions, purpose, primary metric, and owner. This makes it easier to troubleshoot later and prevents new team members from creating overlapping flows.

Also review automation quarterly or after major business changes. New products, acquisition channels, shipping policies, pricing, and customer mix can make an old retention rule less relevant.

Scaling retention does not mean making the CRM increasingly complicated. It means using better evidence to automate more of what reliably helps customers return while removing rules that no longer earn their place.

Turn Customer Data Into A Repeat-Purchase System

An ecommerce CRM for customer retention works when it helps you make better decisions after the first order. Start with clean data, a realistic repurchase cycle, and a small set of lifecycle segments.

hen build the 10 retention tactics in order of need: unify customer context, personalize journeys, improve post-purchase education, make recommendations timely, support replenishment, recognize loyalty, coordinate channels, use service data, detect churn, and run relevant win-back campaigns.

Do not measure success by the number of automations you launch. Measure whether customers return sooner, more often, and with less friction.

Your next step should be simple: identify the single biggest retention leak in your current journey, connect the data needed to see it clearly, and build one CRM workflow that addresses that problem before adding the next layer.

Juxhin

I’m Juxhin, the voice behind The Justifiable.
I’ve spent 6+ years building blogs, managing affiliate campaigns, and testing the messy world of online business. Here, I cut the fluff and share the strategies that actually move the needle — so you can build income that’s sustainable, not speculative.

Related posts:

How Ecommerce CRM Increases Sales: 9 Practical Ways Stores Earn More Ecommerce CRM for Higher Repeat Customer Rates: 8 Tactics That Work Ecommerce CRM Tools That Turn Customers Into Repeat Buyers Best Ecommerce CRM Software: 11 Top Platforms Compared for Smarter Growth Is Ecommerce CRM Worth It for Online Stores? The Revenue Truth Explained Ecommerce CRM for Growing Ecommerce Brands: 11 Features That Matter Most