SMS Marketing ROI: What Harris Farm’s 60x Result Really Teaches Retailers

SMS marketing ROI can look almost unreal when it is presented as a single headline number. Harris Farm Markets, the Sydney-based independent grocer, says it generated a 60x ROI with Klaviyo in the first half of 2026 and grew SMS subscribers 58% in Q2—but the more useful story for retailers is what must happen behind the scenes for results like that to be sustainable.

In a Klaviyo customer video, Harris Farm representatives Alexander, Head of Performance Marketing, and Will, Marketing and Ecommerce Coordinator, describe SMS as a high-adoption, personal channel that can scale alongside the business. They also point to a next-stage ambition that matters just as much as campaign revenue: using AI agents to understand and respond to high volumes of customer messages in real time. (youtube.com)

The case study is compelling, but it should not be treated as a universal benchmark. A reported 60x return is a company-specific, vendor-published outcome. It depends on how revenue and marketing costs are attributed, which customer journeys are included, whether margin is considered, and how the retailer distinguishes marketing texts from service conversations. The transferable lesson is not “copy the number.” It is to build the operating system that makes each relevant message useful, measurable, compliant, and deliverable.

The Harris Farm result: 60x ROI and 58% SMS list growth

Harris Farm Markets describes itself as a fresh-food retailer focused on products that are good for farmers, the environment, and customers. In its video with Klaviyo, the company says it achieved a 60x ROI in the first half of 2026. It also says its SMS subscriber base grew 58% during Q2 2026, which the team links to consumers’ strong adoption of the channel and the more personal character of text messaging. (youtube.com)

Those are two related, but very different, metrics:

• SMS subscriber growth measures whether a brand is earning permission to contact more people.
• SMS marketing ROI measures whether revenue attributed to the program exceeded its total program cost.

A fast-growing list does not automatically create a profitable program. It can create a cost problem if new subscribers receive the same broad messages as everyone else, if they never make a second purchase, or if they rapidly opt out. Conversely, a smaller list of customers who explicitly opted in for delivery updates, weekly specials, back-in-stock alerts, and local-store offers may produce much better economics.

That distinction is particularly important for grocery and fresh-food retail. Purchase frequency can be high, but baskets are often price-sensitive and margin can be thin. A text that simply gives a customer a timely reason to complete a planned order can be more valuable than a generic “20% off” blast. The operational trick is using behavior and context—not just a phone number—to decide whether that message deserves to be sent.

Why SMS can work especially well for fresh-food retail

Fresh-food retail has characteristics that make SMS attractive, provided the program respects customer attention. Inventory moves quickly, promotions can be time-sensitive, local availability matters, and customers may need a fast answer about pickup, substitutions, delivery windows, or product quality.

Email is still excellent for richer content: weekly catalogues, recipes, seasonal storytelling, loyalty updates, and product discovery. SMS is more naturally suited to short, immediate actions. The best programs do not force customers into an email-versus-text choice; they give each channel a defined job.

The high-intent moments are already there

A grocery brand rarely needs to manufacture urgency. It can serve moments customers already care about, such as:

1. A customer leaves a basket containing dinner ingredients.
2. A delivery cutoff is approaching.
3. A frequently purchased item is back in stock.
4. A store-specific special starts today.
5. A substitution needs approval before an order can be packed.
6. A loyalty customer has a reward or personalized offer that is about to expire.

Not every one of these should be an SMS. The channel should be reserved for the moments where speed, visibility, and relevance justify an interruption. A basket reminder for someone who is already actively shopping may be helpful. A daily promotion sent to an inactive subscriber who has not purchased in six months is more likely to create fatigue and unsubscribes.

Personal does not mean inserting a first name

Harris Farm’s team characterizes SMS as a more personal touch. That is true only when personalization changes the decision behind the message. “Hi Sam” is cosmetic. “Your usual organic bananas are available for tomorrow’s delivery to your postcode” is operationally personal.

Useful personalization for a grocer may include store location, delivery zone, dietary preference, typical basket size, prior product category, purchase cadence, loyalty status, weather-sensitive product demand, and whether the customer uses pickup or delivery. It can also include what the brand should not send: a vegetarian shopper should not receive a meat-box promotion, and a customer who placed an order ten minutes ago should not receive a conversion reminder.

Klaviyo’s own SMS guidance emphasizes segmentation by engagement, preference, geography, interests, value, and shopping behavior. That is a sensible baseline: segmentation protects the customer experience while reducing wasted send volume. (help.klaviyo.com)

How to interpret a 60x SMS marketing ROI claim

The most common mistake with ROI case studies is assuming everyone uses the same formula. They do not. Before using Harris Farm’s reported 60x outcome to set a target, a marketing, finance, and engineering team should agree on definitions.

A basic ROI calculation is:

ROI = (attributed incremental profit or revenue − program cost) / program cost

The apparent simplicity hides major choices. Is the numerator revenue or gross profit? Does “program cost” include the marketing platform, SMS carrier fees, creative, agency time, customer-support labor, discounts, and engineering work? Is the revenue direct last-click revenue, modeled attribution, or a holdout-tested estimate of incrementality?

Revenue attribution is not incrementality

A customer can click an SMS link and buy without the text being the reason they bought. They may already have planned to shop, opened an email earlier that day, or seen an offer in an app. Last-click attribution gives the conversion to the final touchpoint. That can be useful for operating a campaign, but it can exaggerate a channel’s causal impact.

A stronger measurement approach uses several views at once:

• Attributed revenue: Useful for daily optimization and campaign reporting.
• Revenue per delivered message: Helps compare audiences and campaign types.
• Contribution margin after discounts: Prevents high-revenue, low-margin offers from looking better than they are.
• Repeat purchase and retention: Shows whether the program creates healthy customer habits.
• Incrementality tests: Compare a randomized holdout group against recipients to estimate causal lift.
• Unsubscribe and complaint rates: Reveal the long-term cost of aggressive messaging.

For example, imagine a retailer sends a $10-off-$80 offer to 20,000 subscribers. The campaign produces $200,000 in attributed sales. That sounds excellent. But if $140,000 would have happened anyway, the discount cost is $30,000, gross margin is 30%, message and platform costs total $8,000, and support costs rise because customers misunderstand exclusions, the contribution may be far smaller than a dashboard’s headline revenue suggests.

The conclusion is not that attribution is useless. It is that marketers should report attribution honestly and validate major changes with controlled tests. A 60x ROI figure can be a valuable directional signal, especially when a company uses consistent methodology over time. It should not be treated as a promise that every retailer can generate $60 for every $1 spent.

The SMS program architecture behind sustainable returns

Durable SMS marketing ROI comes from system design. A brand needs a reliable chain from consent capture through profile creation, segmentation, sending, delivery reporting, purchases, replies, and suppression. Any weak link reduces both performance and trust.

Start with consent as product data

A phone number is not permission. A checkout form, popup, QR code, loyalty signup, or keyword campaign must record what the customer agreed to receive, when they agreed, and through which collection point. This data should be available to the messaging platform and retained for auditability.

For Australian commercial SMS, the Australian Communications and Media Authority says senders must have consent, accurately identify themselves, and provide a functional unsubscribe method. The rules apply to commercial electronic messages, including promotional SMS and email. (acma.gov.au)

That means a healthy implementation stores more than a boolean such as sms_subscribed = true. It should retain fields such as:

• consent status and consent timestamp;
• collection source, such as checkout, form, QR code, or in-store signup;
• disclosure or terms version shown at opt-in;
• country and phone number normalization status;
• preferred store, language, and contact preference where supplied;
• unsubscribe event, timestamp, and the system that processed it.

This structure reduces compliance risk, but it also improves relevance. If the marketing team knows whether a subscriber joined for recipe ideas, delivery alerts, loyalty benefits, or a competition, it can tailor the welcome experience rather than treating every opt-in as identical.

Treat suppression as a first-class event

A suppression list is not an inconvenient back-office file. It is a core part of customer identity and a safety mechanism. When a person opts out, that status must reliably reach every marketing workflow, integration, agency tool, and export.

ACMA guidance warns that consent and unsubscribe failures are frequent sources of non-compliance, including in automated welcome journeys. In other words, automation does not excuse a poor opt-out process; it can amplify the error at scale. (acma.gov.au)

Marketing systems should also distinguish between marketing suppression and transactional communication preferences. A customer who opts out of promotional texts may still need an order-delay notification if they chose SMS as the channel for operational updates. The classification has to be defensible, clear to the customer, and implemented consistently. Smuggling promotions into a service message is a short-term tactic with long-term legal and trust costs.

Connect behavioral events cleanly

The data model should capture the events that actually matter to a grocery customer journey: product viewed, item added to cart, checkout started, order placed, order fulfilled, delivery updated, item substituted, order refunded, loyalty reward earned, and customer-service case opened or resolved.

Each event should contain only the properties needed for segmentation and analysis. For example, an order event can include store, postcode, fulfillment type, basket value, category mix, discount, and order ID. It should not indiscriminately push sensitive customer data into every tool. Sound data governance improves personalization while lowering the chance of accidental exposure or incorrect targeting.

For teams building this layer themselves, reliable event ingestion, idempotency, and webhook handling matter more than a polished campaign editor. Marketing automation becomes brittle if the same purchase event arrives twice, if cancellations never arrive, or if customer profiles are split across several systems. Engineering teams should use documented API contracts and validate event behavior in staging before launching flows; a well-designed email API reference and setup guide is useful when the same operational rigor must extend to transactional email.

A practical SMS flow map for retailers

Campaigns create bursts of attention. Automated flows create the repeatable, context-driven interactions that are more likely to support strong ROI without increasing manual workload. The following flow map is a better starting point than sending several promotional broadcasts per week.

1. Consent-aware welcome flow

A welcome sequence should confirm what the customer signed up for, set realistic frequency expectations, capture preferences, and provide a reason to engage. It should not immediately train subscribers to wait for discounts.

A grocer could ask whether the customer wants local specials, recipe inspiration, delivery updates, or rewards. Responses create durable segmentation data. If a welcome offer is used, measure not just first-order conversion but second purchase rate and unsubscribe behavior after redemption.

2. Browse and basket recovery

These messages work best when they are selective. Suppress customers who purchased, who are already in an active service issue, who have received too many recent messages, or whose basket contains items with uncertain availability.

A useful message provides an actual next step: a basket link, a reminder of a delivery cutoff, or a store-pickup option. It should not overstate scarcity or send an identical reminder across email, SMS, push, and paid retargeting at the same time.

3. Replenishment and recurring-shop reminders

Fresh-food purchase cycles are not as predictable as subscriptions, but many categories have recurring rhythms. A shopper who buys pet food every four weeks or coffee every two weeks may benefit from a reminder near their normal reorder window.

This is where first-party data outperforms generic calendar-based promotions. A replenishment flow should be based on observed behavior, use a broad date range to avoid creepy precision, and stop as soon as the customer buys.

4. Back-in-stock and local availability alerts

These can be highly valuable because the customer explicitly signaled interest. But inventory feeds must be accurate. Sending a back-in-stock message for an item that is unavailable at the shopper’s selected store creates support tickets and damages trust.

For local retail, include location context only when it is reliable. “Available at your Bondi store today” is useful if the inventory system supports it. “Now back in stock” without a qualified store or fulfillment promise can create confusion.

5. Post-purchase care and service recovery

Post-purchase messaging is often more important than another promotion. Order status, substitutions, delivery delays, refunds, and quality issues can determine whether a customer comes back.

This is also the boundary where marketing and support need shared data but different rules. Promotional messages require appropriate marketing consent. Operational messages should be limited to genuinely necessary information and should not become disguised cross-sell sends.

AI customer agents: the next opportunity and the biggest governance test

Harris Farm says it receives thousands of customer messages a day and hopes to use a customer agent that can hear, understand, and react in real time. That ambition fits a broader shift in commerce: marketers and support teams want one customer context across web chat, email, SMS, and messaging channels. (youtube.com)

Klaviyo’s current Customer Agent documentation describes an AI support tool that can handle inquiries across web chat, email, SMS, and WhatsApp, using Shopify information and additional business-provided knowledge. Its platform updates also describe retail-oriented skills such as order tracking, returns and exchanges, subscription editing, and loyalty lookup. (help.klaviyo.com)

The potential is real: a good agent can answer a delivery question instantly, explain a return process at midnight, surface the correct policy, and pass complicated exceptions to a human with context. The danger is treating an agent as a generic chatbot rather than a controlled operational system.

What an AI agent should do first

The best initial use cases are high-volume, low-risk, rules-based questions with clear source data:

• “Where is my order?”
• “What are your delivery windows?”
• “Can I change my pickup time?”
• “Is this item available at my local store?”
• “How do I return or report a damaged item?”
• “How do I update my loyalty details?”

These requests can often be answered from trusted data sources. They can also be measured clearly: containment rate, resolution time, escalation rate, repeat-contact rate, customer satisfaction, and error rate.

Where humans must remain in the loop

The agent should escalate when it cannot verify a fact, when a customer is distressed or angry, when a request involves a material refund or exception, when identity verification is uncertain, or when a response could create a safety, legal, or reputational issue.

For a food retailer, escalations may include allergy questions, food-safety complaints, missing or damaged orders, pricing disputes, and privacy requests. A fast but incorrect answer is not a successful resolution. Teams need approved knowledge sources, permissions boundaries, audit logs, escalation routes, and periodic review of conversations where the agent failed or hallucinated.

The more deeply an AI agent can access customer data and operational tools, the more critical identity verification becomes. Before an agent changes an address, edits an order, reveals loyalty information, or initiates a refund, the business must define what qualifies as authenticated, what actions are allowed, and what needs human approval.

Deliverability still matters when the message is short

SMS does not have inbox placement in the same sense as email, but delivery quality still determines whether customers receive a message, recognize the sender, and trust the interaction. Carrier filtering, invalid numbers, unrecognized sender identity, country-specific registration requirements, and poor engagement can all undermine a program.

For email, the equivalent foundation is authentication and reputation: properly configured SPF, DKIM, and DMARC; a consistent sending identity; complaint control; bounce handling; and careful segmentation. For SMS, the operational equivalents are valid consent, number hygiene, sender identification, messaging cadence, appropriate content, and opt-out reliability.

The shared principle is simple: relevance is a deliverability strategy. When brands send only to people who asked to hear from them and tailor messages to demonstrated intent, they reduce complaints, disengagement, and avoidable spend. Klaviyo’s industry benchmark material likewise frames SMS performance around metrics such as click rate, order rate, revenue per recipient, and unsubscribe rate—not revenue alone. (klaviyo.com)

Use a channel decision policy

An omnichannel program should make deliberate channel choices. A basic policy might look like this:

A policy like this prevents channel collision. It also makes frequency caps more meaningful. The goal is not a universal number of messages per week; it is avoiding unnecessary overlap across the channels a customer uses.

What marketers should copy from Harris Farm—and what they should not

The right takeaway from the Harris Farm story is operational, not cosmetic. The reported outcome suggests that the company found a productive combination of channel fit, customer adoption, platform usability, and scalable workflows. It does not prove that every retailer should immediately increase SMS volume or buy an AI agent.

Copy these principles

• Build permission before promotion. Subscriber growth is valuable only when opt-in is clear and people understand what they will receive.
• Use customer data to narrow sends. Better audience selection can beat bigger audiences.
• Make SMS a useful channel. Delivery deadlines, stock alerts, and service updates often create more value than generic discount blasts.
• Measure the full economics. Include discounts, message costs, labor, gross margin, and unsubscribe effects.
• Connect marketing and service context. Customers do not care which internal team owns the conversation.
• Scale support carefully. Start AI agents with constrained, measurable tasks and a clear human handoff path.

Avoid these shortcuts

• Reporting only last-click revenue as if it were incremental profit.
• Treating every customer number collected at checkout as a marketing opt-in.
• Using service notifications as a place to insert promotional offers.
• Creating broad, identical flows for every subscriber.
• Automating order changes, refunds, or sensitive disclosures without guardrails.
• Ignoring opt-outs, number validity, and suppression synchronization because the campaign dashboard looks strong.

These shortcuts may inflate short-term attributed revenue. They are also the quickest path to declining engagement, compliance problems, customer complaints, and a support team that has to clean up the consequences.

A 90-day plan to improve SMS marketing ROI

Retailers do not need a complete replatforming project to make progress. A focused 90-day operating plan can establish the data, measurement, and controls needed to scale.

Days 1–30: Audit permission, data, and measurement

Inventory every place phone numbers enter the business: ecommerce checkout, in-store loyalty, customer service, competitions, lead forms, and third-party integrations. Confirm what consent language is shown, where proof is stored, and whether opt-outs propagate immediately to all marketing tools.

Then map the event flow from store or ecommerce platform to the customer-data and messaging systems. Check for duplicated purchase events, missing refund events, delayed inventory data, and profiles with conflicting email or SMS consent states. Define a standard ROI calculation and a dashboard containing delivered messages, clicks, purchases, revenue, discount cost, contribution margin, opt-outs, and complaints.

Days 31–60: Launch high-value, low-risk flows

Prioritize a consent-aware welcome flow, a purchase-suppressed basket recovery flow, and a post-purchase service flow. Use small audiences at first. Create frequency caps and make sure email and SMS do not fire redundant messages within the same customer journey.

At this stage, establish a holdout group for at least one major flow or campaign. Even a modest randomized test can reveal whether a flow is producing genuinely incremental orders or merely capturing credit for purchases that would have occurred anyway.

Days 61–90: Add service intelligence and optimize

Review message performance by acquisition source, customer value, store location, purchase recency, and message purpose. A list source with lower top-line growth may produce customers who stay subscribed and buy repeatedly; that source may be more valuable than a giveaway-led signup funnel.

Choose one narrow AI service use case, such as order tracking or delivery-window questions. Give the agent a controlled knowledge base, a list of permitted actions, an escalation threshold, and owners in both support and engineering. Compare its results with human-handled cases before expanding its scope.

The broader lesson: customer experience is the ROI engine

Harris Farm’s reported 60x result is attention-grabbing. Its longer-term strategy is more revealing: the company is not framing SMS only as a promotional broadcasting tool. It is treating messaging as part of a customer experience system that includes acquisition, personalization, operational updates, support, and eventually real-time AI assistance.

That is where email infrastructure, transactional messaging, marketing automation, and customer service converge. A customer does not separate a receipt from a delivery update, an abandoned-cart reminder, and a support reply into different software categories. They experience all of them as one relationship with the retailer.

For companies trying to improve SMS marketing ROI, the most durable path is therefore not greater message volume. It is a trustworthy identity layer, explicit consent, clean event data, disciplined segmentation, credible measurement, reliable suppression, and a support strategy that knows when automation should yield to a person. Harris Farm’s reported results offer a useful north star—but only the underlying operating discipline can make that kind of result repeatable.

FAQ

What does a 60x SMS marketing ROI mean?

It generally means the company reports roughly $60 in attributed return for every $1 it counted as SMS program cost. The exact meaning depends on the company’s attribution window, whether it uses revenue or profit, and which costs it includes. Treat it as a case-specific metric, not an industry guarantee.

Is SMS better than email for retail marketing?

Neither channel is universally better. SMS is often stronger for urgent, concise, high-intent actions such as delivery cutoffs, stock alerts, and cart reminders. Email is better for detailed merchandising, receipts, editorial content, and longer product discovery. Strong programs coordinate both rather than duplicating every message.

How can retailers grow an SMS list without damaging trust?

Use clear opt-in language, state what messages customers will receive, collect consent proof, offer a real benefit such as useful alerts or relevant offers, and make unsubscribing easy. Avoid pre-checked boxes, vague consent disclosures, and incentive schemes that attract people who do not actually want ongoing messages.

Can an AI customer service agent answer SMS messages?

Yes. Modern customer-service agents can be configured to work across channels including SMS, email, web chat, and WhatsApp. They should begin with factual, well-defined tasks such as order tracking and delivery questions, then escalate complex, sensitive, or uncertain requests to human agents. (help.klaviyo.com)

Which metrics matter besides SMS revenue?

Track delivery rate, clicks, conversion rate, revenue per delivered message, gross margin after discount, repeat purchase rate, opt-out rate, complaint rate, frequency, support contacts caused by messages, and incremental lift from holdout testing. These metrics show whether the program is creating durable value rather than only short-term attributed sales.