Truths, half truths and little white lies: What chat vendors say when marketing to retailers

Customer experience is important to every industry, but especially so to online retailers and live chat is not just a nice alternative to more traditional customer support options these days, it’s actually the preferred option. So, choosing a chat solution that offers customers a pleasant and reliable experience is of more importance than ever.

But unfortunately, many retail companies are being misled by legacy chat platform vendors who sometimes offer half-truths and false promises when selling their products. In this blog, I’d like to examine the ways in which retailers can prevent misconceptions arising by educating themselves on the top ways that chat vendors can sometimes mislead.

1) The prediction lie

Vendors often use the term “prediction” to refer to some complex, almost mystical method, innate to their platforms, about how they use vast quantities of visitor data to target just the right hot leads on the website. This is quite often a lie.

Generally, most legacy chat vendors tend to use stacks of business rules, which are basic “IF-THEN” statements applied to individual web pages, to coarsely segment web visitors and target them based on broad-brush, non-tested assumptions about visitor behaviour. This is not ‘prediction’. Very few vendors use predictive modelling for visitor targeting and, unless your business happens to be one of their largest customers, it’s rarely worth their time to create customised predictive models for you alone.

Even if your vendor uses some form of ‘big data’ or ‘prediction’, you should ask them for real comparisons of this against other vendors. Most vendors do not invest resources towards creating, consistently maintaining, and tuning a library of industry-specific models. Using true predictive models that are constantly tuned, results in a better performance over business rules (that simply go off on basic triggers such as time on page, pages visited, etc.), and is usually at a high cost to the company.

Ask your vendor: If your vendor claims to use ‘prediction’, ask if that’s based on pre-built predictive models specific to the retail industry, or if it’s just a fancy reference to ‘business rules’. This matters because predictive models are vastly better than business rules in a plethora of ways. Most importantly, they drive significantly better visitor targeting which leads to, in our experience, as much as 90% higher chat revenues in real A/B tests than rules-based systems can deliver.

Ask your vendor: Do they invest in continuously tuning the performance of their ‘prediction’, and pass on the benefits to you? Do they cross-pollinate their models with learnings about user behaviour across customers to come up with strong and superior performing reference models that can drive dramatic lift in outcomes right off the bat?

2) The big fat data lie

The chat funnel has various stages. Visitor targeting only takes care of the top of the funnel. To have effective chat, you need to focus on every stage of the chat funnel, applying data (both “big” and “small”) to maximise the throughput of each stage. Most vendors gloss over this because their platforms do not have the capability to apply data to each stage in a systematic fashion.

Here’s an example: even after you have identified the right visitors to offer chat to on your website, how do you ensure that they click on the invitation and start a chat? Techniques such as design of experiments allow for advanced testing to identify the right combination and form of user experience (UX) elements on the chat invitation to maximize the acceptance rate.

Furthermore, predictive models allow for personalisation of the chat invitation in real-time to a specific visitor’s intent and recent actions. This is just one example of a technique that can help with maximising the value of one stage of the funnel – the acceptance rate. But the funnel has many stages, and you need to optimise them all using different data techniques and tools to ensure that your overall funnel works like a well-oiled machine.

Ask your vendor: Do they have the right data leveraging tools and techniques to maximize outcomes for every stage of the chat funnel?

3) The revenue lie

Ultimately, it all boils down to measurable outcomes. In sales chat, it’s all about revenues. This is where the idea of ‘incrementality’ comes in. Let’s say you invited a visitor to chat, and they bought £100 worth of merchandise from your online store after chatting. Most chat platforms attribute this £100 as chat generated revenue.

However, very seldom do they contrast this against the revenue that the visitor would have generated if left alone to self-serve. If we had a magic mirror that  showed us what each visitor would have bought on their own without assistance, and this number was, say, £90, then the incremental revenue that the chat generated is only £10 (£100 less £90). It is this £10 value that you really care about, because this represents the true value that the chat contributed. But, you don’t really need a magic mirror to calculate this, you just need a control group.

Here’s something related: a common tweak that vendors apply to increase revenues at the bottom of the funnel is to open up the hot lead rate (the number of people you invite) at the top of the funnel. Like doubling the size of your fishing net, this means inviting higher quantity, lower quality leads to trick the metrics into showing higher revenue gains. This is, however, a cop-out, and the sad fact is that this is the most common method in which legacy chat vendors inflate revenues. Because, while this ‘tweak’ may lead to higher revenues, it also reduces revenue efficiency.

An advanced chat program would be able to target visitors for high incremental revenue. It would also be able to leverage systematic control group tests to identify the value of incremental, or “net new” revenue, and report it back to you, eliminating the value of self-service cannibalisation.

Ask your vendor: How do they measure and report the value that they create for you? 

4) The outcomes lie

One of the worst crimes that your chat vendor is committing against you is selling you chat as if it is a utility, with no direct ownership of outcomes. This translates into the very real cost of you having to either employ your own staff full-time to manage the platform’s business rules (a large US retailer used to spend up to half a million dollars a year on staff hired to manage their chat vendor’s rules), or cough up seemingly never-ending “professional services” fees charged by the platform vendor for every single change requested. Even after charging you for all of this, at the end of the day, the onus of success or failure of the program rests on you, not on the vendor. There must be a better way. (And there is.)

Chat vendors should be partners, not utility providers. They should be able to continuously drive optimisations and deliver actionable insights to improve performance and outcomes. They should bring to your program best practices from across their experience, which can help you leapfrog over the pain of having to reinvent the wheel on your own. It is even better if they commit to a pricing structure that front-loads outcomes and incentivises them to work for your success.

Ask your vendor: How will they demonstrate ownership of your outcomes and work towards continuously improving outcomes for you?

5) The omnichannel lie

Omnichannel is a word that means different things depending on who you ask, but one thing we all agree on: it means more than just having a single view of a customer. Today, omnichannel means blended interactions that straddle channels, often simultaneously, while seamlessly sharing data and context to create a common meaning (from data) across channels. A great example is automated chat with a virtual agent (VA) to handle the most common inquiries.

Several chat providers claim to “integrate” with third party virtual agent software, but this integration is often just the passing of the VA interaction with some limited history to the chat agent.

A true omnichannel experience – which would lead to true digital transformation – would mean a tightly integrated and orchestrated engagement model:

  • Integrated visitor targeting that identifies the difference between visitors who would benefit from a VA and those who would benefit from chat.
  • Seamless transition between the VA and chat with the entire VA interaction history, customer context, and integrated recommendations to the chat agent on how to continue the interaction.
  • Integrated analytics that mine chat and VA transcripts to derive meaningful data on how more chat journeys could be automated using the VA, thus improving self-service and letting chat agents focus on the customers who really need their help.

Going beyond just VA and chat, to extend the value and insight from chat into other channels such as the IVR (Interactive Voice Response), social media, or to your digital marketing investments. These are capabilities that would require a complex omnichannel architecture that most chat providers have not really bothered to build.

Ask your vendor: Do they have true channel orchestration capabilities and omnichannel analytics to help deliver digital transformation?

Protecting yourself with knowledge

When important customer experience considerations become industry buzzwords, every vendor will quickly claim that they’re able to deliver.

Unfortunately, for the average retail enterprise, it’s often difficult to discern the often subtle differences between vendors that are able to offer what retail customers truly demand, and those who claim the same but only offer a quasi-solution.

Just like with buying a used car, the best defence is educating yourself, and knowing where vendors will try to cut corners. Bring this list of questions with you the next time you meet with your chat vendor and you’ll find out very quickly if they’re the real deal or one of the many imposters you should be staying away from.