Customer Engagement Software: What to Check Before Buying

Strip the category label away and three jobs are left. Every tool on the market does some mix of them, and the mix is what you are really choosing between.

• Reach. Sending a planned message to a group you defined — a campaign, a newsletter, a launch announcement.
• React. Sending an automated message because something happened: an order, an abandoned checkout, a contact writing in for the first time.
• Respond. Handling the replies, in a shared place where more than one person can see them.

A tool that only reaches is a broadcast tool. One that only responds is an inbox. Engagement software is the combination — and the combination is what makes attribution possible at all, because you cannot credit a message with an order unless one system saw both. Keaz runs all three on WhatsApp for Shopify stores, and what that actually covers is deliberately a short list.

The react half is where tools differ most, because it depends entirely on which events the system can see. In Keaz a sequence can start from five triggers across four sources: two Shopify events (a contact's purchase, an abandoned checkout), one internal trigger called Smart activation that picks the sending moment per contact, one external webhook from your own systems, and a chat-in when the customer writes first. The Flows overview in the help center documents each one. Ask any vendor for the equivalent list. A tool with two triggers and a beautiful editor runs out of things to automate inside a month.

Feature lists are written to survive comparison. Numbers are not. Ask for these four, in this order.

1. Reachable share. What proportion of your existing customers you may message on day one. This is a consent question before it is a technical one, and it is the number most likely to be quietly awful.
2. Time to first campaign. How many days pass between signing and the first message going out. If the answer involves a migration project, price the project.
3. Attributed revenue per campaign. Not opens, not clicks — which orders came back. In Keaz that runs through a discount code created per flow, so the credit is a real order in your shop rather than a modelled estimate.
4. Repeat-purchase rate, before and after. The store-level number the whole exercise is meant to move. Baymard's review of 50 studies puts the documented average cart abandonment rate at 70.22% (Baymard Institute, retrieved 2026-09-23) — that is roughly the size of the gap engagement software is sold against, and the reason a tool that cannot measure the gap cannot prove it closed it.

If a vendor answers three of the four and talks around one, the one they talked around is the answer. How we model the revenue side against your own store data is written up separately, because a forecast built on an industry average is a brochure, not a forecast.

The two categories are sold to the same buyer and they are not the same product. The difference is direction. A support desk is built around inbound volume: tickets, queues, routing rules, SLA timers, agent performance. Engagement software is built around the lifecycle: who should hear from you next, and what it earned when they did.

Most stores under ten people need the second and believe they need the first, because "customer service software" is the phrase that comes to mind when messages pile up. What is worth automating in customer service, and what is not is the honest version of that decision.

It is worth being plain about the boundary on our own side too. Keaz has no ticket routing, no SLA timers, no deflection scoring and no agent that closes cases on its own. Conversations land in a shared inbox and a person handles whatever the rules did not. For a team of one to ten that shape is usually right; for a tiered support organisation it is not, and no amount of configuration changes that. What the channel looks like by vertical is a better guide to fit than any feature matrix.

Assume four weeks, not a quarter.

1. Connect the shop first. The purchase and checkout events are what make every automated step after this one possible.
2. Fix the opt-in path before the campaign. Consent is the reachable-share number from the section above. Under the GDPR, consent is a freely given, specific, informed and unambiguous indication of the data subject's wishes (Regulation (EU) 2016/679, Art. 4(11)); a pre-ticked box is none of those things. This is a summary rather than legal advice — have your own counsel review your setup.
3. Build one segment, not twelve. Start with customers who bought once and never came back. The Segment Builder shows the audience size live, so you find out a segment is too small before you build a campaign for it.
4. Turn on two flows. A welcome and a cart recovery. Two is enough to learn the tool and small enough to switch off again without a meeting.
5. Read the numbers at 90 days. Not at 90 messages. Which two retention numbers actually matter is the short version of what to look at.

• Buying for the feature list instead of the reachable list. A tool with every feature and no consented contacts sends nothing. Check the reachable share first and the roadmap last.
• Treating the inbox as an afterthought. Outbound creates inbound. A campaign to 8,000 people generates replies whether or not anybody planned for them, and unanswered replies cost more goodwill than the campaign earned.
• Measuring sends. Messages delivered is a cost, not a result. If a reporting screen leads with it, ask what it is standing in for.
• Buying a support desk to solve an engagement problem. Queues and SLAs organise work you already have. They do not create the next order.
• Running it for three weeks. Lifecycle programmes show their value on the second purchase, which is by definition later than a trial. Set the review date at 90 days when you sign, not when you get nervous.