Pourquoi adopter une logique de repli multicanale

Your customer is waiting for a delivery update. You send it via RCS: rich, interactive, beautiful. Their phone doesn't support RCS. So the message goes... nowhere. Meanwhile, a package is sitting on their doorstep getting rained on because they didn't know it arrived.

This is the multi-channel problem in miniature: the gap between where you send a message and where your customer actually receives it.

The multi-channel reality

Let's start with an uncomfortable truth: no single communication channel works for everyone. Salesforce's 2023 State of the Connected Customer report found that the majority of customers use multiple channels to interact with businesses and expect consistent experiences across all of them.^[1]Your customers are on SMS, email, WhatsApp, RCS, and push notifications, sometimes all in the same day, sometimes all for the same transaction.

The instinct, when faced with this reality, is to do one of two things: pick a single channel and hope for the best, or blast every channel simultaneously and hope the customer doesn't mind. Both approaches are wrong, but the second one is worse.

Why "just send it everywhere" is the worst strategy

Imagine walking into a building and hearing someone shout your name from every room simultaneously. That's the experience of receiving the same message via SMS, email, push notification, and WhatsApp within the span of 30 seconds. It doesn't feel comprehensive. It feels desperate.

Beyond the customer experience problem, the "send everywhere" approach has real consequences:

• Cost multiplication. If an SMS costs 3-5p and you're sending 500,000 messages a month, adding redundant SMS sends alongside email and push "just in case" turns a manageable budget into an expensive habit. One fintech company we spoke with was spending 40% more on messaging than necessary because they'd defaulted to parallel sends on every channel.
• Channel fatigue. Customers who receive duplicate messages across channels are measurably more likely to opt out. Industry consumer surveys consistently show that the vast majority of respondents have opted out of at least one brand's communications due to excessive messaging. You're not just wasting money; you're actively shrinking your reachable audience.
• Signal degradation. When every message comes through every channel, customers stop paying attention to any of them. It's the communication equivalent of the boy who cried wolf. When you actually need to reach someone urgently, your messages are lost in the noise you created.

A framework for channel selection

Choosing the right channel for a given message isn't guesswork. It's a decision that can be decomposed into four factors. We've found this framework useful both for building routing logic and for having productive conversations with product teams who want "everything sent via SMS" without thinking through the implications.

1. Urgency

How quickly does the recipient need to see this? A two-factor authentication code needs to arrive in seconds. A monthly billing summary can arrive whenever. This single axis eliminates a lot of channel candidates immediately.

• Immediate (seconds): SMS, push notification, RCS. These are interruptive channels that demand attention now.
• Timely (minutes to hours): WhatsApp, RCS, SMS. The recipient should see it soon, but it's not an emergency.
• Informational (hours to days): Email. The message has value, but there's no time pressure.

2. Content richness

What does the message need to contain? A six-digit code is fine in an SMS. A detailed invoice with line items, payment links, and your brand's visual identity? That's an email, or possibly an RCS message with a rich card.

RCS occupies an interesting middle ground here. It supports images, carousels, suggested actions, and verified sender branding while still living in the native messaging app. For messages that need to be both immediate and visually rich, it's often the best primary channel. When it works.

3. User preference

Here's a radical idea that somehow remains underutilised: ask people how they want to be contacted.

Not in a buried settings page that requires seven clicks to reach. Not in a preference centre that hasn't been updated since 2019. Actually ask, during meaningful touchpoints, and then actually honour the response. Epsilon's research found that 80% of consumers are more likely to purchase when brands offer personalised experiences,^[2] and channel preference is one of the lowest-effort, highest-impact personalisation signals available.

The catch: preference should inform default behaviour, not override safety-critical communications. If a customer prefers email but needs a fraud alert, that's still going via SMS (and probably email too, in that case). Preference is a guide, not a straitjacket.

4. Cost

Let's talk about money, because nobody else seems to want to.

Not all channels are priced equally. At volume, these differences compound fast:

• Email: Fractions of a penny per message. Essentially free at scale once your infrastructure is in place.
• Push notifications: Free (from a per-message cost perspective), though they require an installed app.
• RCS: Typically 1-3p per message, varies by market and provider. Cheaper than SMS in most cases.
• SMS: 3-5p per segment in the UK, significantly more internationally. And don't forget: messages over 160 characters split into multiple segments, each billed separately. That "short" message with a URL and a personalised greeting? It might be three segments.
• WhatsApp Business: Conversation-based pricing, roughly 3-8p depending on the conversation type and country. The 24-hour conversation window adds complexity to cost modelling.

A thoughtful routing system considers cost as a tiebreaker, not the primary decision factor. If email and SMS are both appropriate for a given message, and the customer has no stated preference, sending via email saves real money at scale without meaningfully degrading the experience.

The fallback waterfall: RCS → SMS → Email

For many use cases (appointment reminders, delivery updates, account notifications), the ideal fallback sequence starts with the richest viable channel and degrades gracefully:

1. Try RCS. Check device capability first (the RCS capability API tells you if the recipient's device supports it). If supported, send the rich version. Wait for a delivery receipt.
2. Fall back to SMS. If RCS delivery fails or the device doesn't support it, send a plain-text version via SMS. You lose the rich formatting, but the message gets through.
3. Fall back to email. If SMS delivery fails (invalid number, carrier rejection, or no mobile number on file), send an email version. It's slower, but it's a safety net.

This isn't the only valid waterfall. For non-urgent messages, you might start with email and only escalate to SMS if the email isn't opened within a defined window. For app-first businesses, push notification might be step one, with SMS as the fallback for users who've disabled notifications or uninstalled the app.

The key insight: a fallback waterfall should be configurable per message type, not hardcoded for the entire system. Your 2FA codes and your marketing messages have fundamentally different requirements, and routing them through the same logic is a category error.

Smart routing: let the data decide

Static fallback rules are a good starting point. But the real power comes from making routing decisions based on observed behaviour.

If you're tracking delivery receipts and engagement across channels (and you should be; see our article on building reliable communication at scale), you're sitting on a dataset that tells you something valuable: for each customer, which channel actually works?

• Customer A opens 90% of their emails within an hour but has never clicked an SMS link. Route informational messages to email.
• Customer B ignores email entirely but responds to SMS within minutes. For time-sensitive messages, lead with SMS.
• Customer C has an RCS-capable device and engages with rich cards at twice the rate of plain SMS. Prioritise RCS.

This isn't machine learning or AI (despite what every vendor's marketing page would have you believe). It's straightforward analytics: track what works per recipient, per message type, and use that data to inform routing decisions. A simple scoring model (channel engagement rate weighted by recency) gets you surprisingly far.

Industry research suggests that businesses using intelligent message routing can reduce their per-message costs by 25-35% while improving delivery rates. The maths is intuitive: stop sending SMS to people who ignore them and email to people who never check their inbox.

Good vs. bad multi-channel: a tale of two experiences

The bad version

You book a dentist appointment. Over the next 48 hours, you receive: an email confirmation (reasonable), an SMS confirmation (okay, a bit redundant), a WhatsApp message with the same information (starting to feel stalked), an SMS reminder the day before (fine, but you've already confirmed twice), and a push notification from their app the morning of (you forgot you had the app). Five messages across four channels for a single appointment. You now associate this dentist with notification anxiety.

The good version

You book the same appointment. You get an email confirmation with all the details, a calendar invite, and a note saying "We'll send you a reminder via text the day before." The day before, a single SMS arrives: "Reminder: dentist tomorrow at 10am. Reply Y to confirm or call us to reschedule." You reply Y. Done.

Two messages. Two channels. Each chosen for what it does best: email for detail and record-keeping, SMS for timely action. The customer feels informed, not bombarded. That's the difference between multi-channel done thoughtfully and multi-channel done indiscriminately.

When to break the cascade

A fallback waterfall implies linear escalation: try A, then B, then C. But there are cases where you should skip steps or send on multiple channels intentionally:

• Security alerts. If someone's account is compromised, you send SMS and email and push. This isn't redundancy; it's urgency. You need to reach them through whatever channel they happen to be looking at.
• Regulatory requirements. Some industries require specific channels for specific message types. Financial services often mandate that certain disclosures be sent via a durable medium (email or post), regardless of the customer's preferred channel.
• Known delivery failures. If your data shows that SMS consistently fails for a particular carrier in a particular region, skip SMS entirely and go straight to email. Don't waste time (and money) on a channel you know won't work.

Building this into your system

If you're implementing multi-channel fallback logic, the architecture looks roughly like this:

1. A routing engine that accepts a message request with metadata (type, urgency, recipient) and returns an ordered list of channels to attempt.
2. Channel capability checks that verify, before sending, whether the recipient can receive on the chosen channel (RCS capability, valid phone number, email verification status).
3. Delivery confirmation with timeout. If no delivery receipt arrives within a channel-appropriate window (30 seconds for SMS, 5 minutes for email), trigger the next channel in the cascade.
4. A feedback loop that records which channel succeeded and updates the recipient's profile for future routing decisions.

This is, more or less, what we built into Conductor's routing layer. Not because the concept is complicated, but because getting the details right (timeout tuning, deduplication across channels, cost tracking, preference management) requires a system that's purpose-built for it rather than stitched together from independent channel integrations.

The goal isn't to use every channel. It's to use the right channel, at the right time, for the right message, and to have a thoughtful plan B when plan A doesn't work. That's it. Everything else is implementation detail.

Multi-channel fallback logic isn't glamorous. It doesn't make for impressive conference talks or eye-catching demos. But it's the difference between a communication system that works and one that merely sends. And at the end of the day, your customers don't care about your channel strategy. They just want their messages to arrive.