Customer engagement platform: what retailers should look for

A customer engagement platform manages the ongoing relationship between a retailer and its customers across whatever channels the business runs, holding what's known about each person, deciding what should happen next, and delivering it. Where the category gets slippery is that products from four quite different origins all now describe themselves this way.

Knowing which origin a product came from tells you more about what it's good at than any feature list will.

The four origins, and what each is good at

Messaging and marketing automation products such as Klaviyo, Emarsys and Attentive grew up sending email and SMS, and they remain the best in the market at doing that. They hold a customer profile because sending requires one, which isn't the same as holding a customer record, and the distinction shows up the moment you ask about in-store purchases or product margin.

Service and conversation products including Intercom, Gorgias and Zendesk came at the problem from support. They own the conversation history, they handle inbound well, and they usually know nothing about what the customer has bought or what they're worth to the business.

Enterprise suites such as Adobe Experience Platform, Salesforce Marketing Cloud and Braze cover the broadest surface, and they cover it properly. The cost isn't only licence fees but the implementation team and the timeline that comes attached, which is why they tend to suit organisations with both already in place.

Customer data platforms approach it from the other end, building the unified customer record first and letting the other tools act on it. They don't send anything themselves, which sounds like a limitation until you notice how much of the trouble in the other three categories traces back to an incomplete record.

Most retailers end up running two or three of these at once. The question worth settling isn't which one wins, it's which one holds the customer record that everything else reads from, because that decision quietly determines the ceiling on all of them.

What the category tends to be weakest at

Three gaps show up again and again when retailers audit what they've bought, and they're related.

The first is offline behaviour. If in-store purchases aren't attached to known customers, the platform is running an online relationship and describing it as a customer relationship, which works acceptably for a pure ecommerce brand and badly for anyone with shops. It's worth asking any vendor what share of point of sale transactions they'd expect to match to a person for a retailer of your size, and treating a vague answer as an answer.

The second is product economics. Very few platforms in this category hold product cost alongside the order, so they can optimise for response rate and revenue but not for margin. That's the mechanism behind a campaign programme that reports steadily improving results while gross margin drifts the other way, and it's more common than it should be.

The third is that generic platforms arrive with no view on what anything means in your business. What counts as a VIP here, when a customer should be treated as at risk, which categories are seasonal and which are core: someone on your team will spend the first quarter writing those definitions into the tool, and they'll then exist only inside that tool.

How it has to fit the rest of the stack

Nobody buys one of these into an empty building. A retailer evaluating an engagement platform already runs a point of sale system, an ecommerce platform, probably a loyalty programme and almost certainly something that sends email, and the new product has to earn its place among them rather than pretend they aren't there.

Three integration questions decide how that goes. Whether the platform reads point of sale transactions directly or expects them delivered as a file, which sounds like a technical detail and turns into a six-week difference in practice. Whether activation runs in both directions, so campaign results come back rather than only audiences going out, since one-way integration means you can send but never learn. And whether the product can coexist with your existing sending tool or assumes it's replacing it, because a platform that quietly requires you to migrate email is a much larger project than the one described in the proposal.

It's also worth asking what happens when a source system changes. Point of sale platforms get upgraded, field names shift, and integrations built once tend to fail quietly some months later, which is usually discovered by a marketer wondering why a segment has stopped growing.

Pricing, roughly

Most vendors price on some combination of contacts, profiles or messages sent, which means the bill scales with the size of your database rather than with the value you get from it. Enterprise suites price on modules and usually involve a services engagement to go with them.

Two costs get underestimated consistently. Implementation is the obvious one. The less obvious one is that a platform your team finds awkward will quietly go unused while continuing to invoice, and that outcome doesn't show up in any comparison spreadsheet.

What to test before choosing

The useful demo is a short one with specific asks in it. Show a customer who's bought both in store and online, and show where each record came from. Show gross margin for a segment, and show where the cost figure originates. Have someone non-technical from your own team build something while everyone watches, because that single test predicts adoption better than any feature comparison. Then push the result to the tool your team already sends from, and ask to see the audit trail for everything that just happened.

Anything a vendor needs to take away and prepare for next week is a roadmap item, whatever it's called in the meeting.

Where Lexi fits

Lexi is the agentic customer data platform for omnichannel retail, which means it sits in the fourth of those four categories rather than competing with your sending tools.

It ingests transactions, inventory, POS, loyalty, reviews and signals, resolves identity across all of them, and holds product cost alongside the order so margin per customer is a number you can actually see. A semantic layer carries what your terms mean inside your business, so the difference between a full-price buyer and a discount-dependent one is something Lexi knows rather than something you configure into each new tool you buy.

You ask a question in plain language, Lexi builds the answer from your data and shows its working, then builds the segment and activates it into the platform your team already uses. Lexi makes that platform smarter rather than replacing it, and the send still happens where the team works today. Small reversible actions proceed on their own, larger ones wait for someone to approve them, and every action is logged and can be undone.

Lexi runs inside AWS Bedrock, data never leaves the platform, and no personally identifiable information enters AI processing. Lexer is SOC 2 certified.

Measuring whether it worked

Resist judging this category on its own metrics. Open rates, click rates and satisfaction scores all move for reasons that have very little to do with money, and a programme can improve all three while contributing nothing.

The numbers that tell you something are repeat purchase rate by segment, gross margin per customer, time to second purchase, and the share of in-store transactions attached to a known customer. That last one is the leading indicator for everything else, and if it isn't climbing then none of the others will move far either.

‍

Related Articles

📄 Customer Data Platform (CDP)

📄 Clienteling in Retail

📄 Customer Segmentation Matrix

📄 Retail Data Analytics

📄 Retail Data Solutions

📄 Customer Insights Platform

📄 Retail Clienteling Software

📄 Retail Data Analytics Solutions

📄 Retail Data Analytics

📄 Retail Sales Data Analysis

📄 Retail Data Collector

📄 Customer Experience in Retail

‍