Customer Service Platforms: Which Parts a Small Store Needs

The phrase names a bundle, not a capability. Strip the positioning away and almost every product in the category sells the same four jobs, priced as one thing.

1. A shared inbox. One place where messages from every channel land, so two people answering never answer the same person twice.
2. Routing and workload. Assignment, queues, statuses, SLA timers — the machinery that decides who owns which conversation and when it is late.
3. Self-service. A help center, an FAQ, an order-status lookup: the answers a customer reaches without writing to you at all.
4. Automation. Rules that reply on their own, or that send the message before the question arrives.

Every product in the category covers some of the four well and the rest adequately, and which is which is a question for a trial rather than a feature list. The useful move is to decide what you are buying before you start comparing. Our own overview of what Keaz covers is deliberately narrow about which of the four it does: the inbox and the automation, on WhatsApp, for Shopify stores. There is no routing engine and no help-center builder in it, and pretending otherwise would waste your trial week.

Two questions settle it, and neither is about revenue.

1. Does more than one person answer messages as a main part of their job? If the answer is no, routing is overhead. You are the queue. Assignment rules, statuses and SLA timers exist to coordinate people who would otherwise collide, and one person does not collide with themselves. This is the part most often bought first and used least.
2. Do your top three questions make up most of your volume? Count it from the last 100 conversations rather than from memory. If three buckets carry the bulk, self-service and automation pay for themselves quickly. If your volume is a long tail of one-offs, a help center is a writing project that will deflect very little.

The shared inbox is the one part nobody outgrows and nobody grows into — you need it on day one and you still need it at fifty orders a day. Automation is the part with the best return at small scale, because at small scale the repetition is concentrated: order status, shipping times, the returns window. Where exactly that boundary sits is a judgement your own transcript makes for you, and we set out which questions are worth automating and which are not in a separate piece.

Often, yes — and this is the decision the category quietly skips. A platform that unifies five channels is worth little if your customers are not on the five. Across Keaz stores the audience splits 24% reachable by email only, 38% on both, 34% on WhatsApp only and 3% on neither. About a third of the people who already bought from you cannot be reached by email at all, whatever your inbox does with the ones who can.

Reach is also why response feels different per channel. Published benchmark bands put WhatsApp read rates above 90% against email open rates of 30–40%; we treat those as a labelled band rather than a promise, and how we calculate with them is written down. The practical consequence for a service queue is mundane: a message that gets read is a message that does not become a second message.

It is worth asking where the questions come from before buying a better way to answer them. Among shoppers who abandon a cart for a reason other than browsing, 40% cite extra costs that appeared late, 18% a forced account and 12% not being able to see the total up front (Baymard Institute, 50 Cart Abandonment Rate Statistics, 2026, retrieved 2026-09-25). Those are information gaps, and each one arrives in your inbox as a pre-purchase question. A platform answers them faster. A clearer product page stops them. If you want the mechanics of one number and several people, how small teams share one WhatsApp number covers it, and the Inbox overview in our help center shows what that looks like in practice.

Every ranked list of the best customer service platforms was written by somebody with a stake in the answer — this page included. So the five questions below are about your store rather than about anyone's product, which also makes them the only ones you can verify yourself.

1. Which of the four parts will be switched on in month one? Write the answer down before the demo. Anything not on that list is a future you are paying for now.
2. Does it read your order data? An agent who has to open a second tab to find an order number is the slow part of every reply, and no amount of routing fixes it.
3. What happens to a message that arrives at 22:00 on a Saturday? Ask to see the actual out-of-hours behaviour, not the setting that controls it.
4. Can it start a conversation, or only continue one? Sending a status update on a real event — a purchase, an abandoned checkout — removes the question instead of answering it. A pure reply tool cannot do this, and it is the difference between a service desk and a conversation layer.
5. What does it cost when volume doubles? Per-seat, per-conversation and per-message models diverge sharply in a good December. Model the December, not the March.

Question four is usually the one that reorders the shortlist, because it is the only one the category does not advertise. If you want to see what the automation side of it looks like before committing to anything, the Automations overview in our help center lists exactly what exists, and the Keaz Forecast puts a number on the channel using your own store data rather than an industry average.

• Buying the bundle for the part you will not switch on. Routing and a help center are the two that usually sit idle in a small store, and they are the two that carry the price.
• Adding a channel without a documented opt-in. Messaging channels only work on explicit, documented, revocable consent. Answering an inbound message is not the same thing as being allowed to start one.
• Migrating every channel in one weekend. Move the channel with the most volume first and leave the rest where they are for a month. A half-migrated inbox is worse than two separate ones.
• Measuring tickets closed. It rewards volume. Measure questions that stopped arriving and how often the same customer writes twice about the same thing in a week.
• Treating the platform as the fix for a page problem. If the same question arrives fifty times a month, the cheapest version of the answer belongs on the product page, not in a macro.