What to ask your post-purchase conversational commerce provider: 27 important questions

‍Full article available as a PDF here.

‍‍

So you’ve decided to open up and automate popular messaging channels for your ecommerce customers like WhatsApp and Messenger. That’s great! But what now? 

The world of conversational AI is still new and most organisations don't yet have in-house expertise. It's pretty tough to know what to ask, let alone what answers to expect.

In this four-part series we hope to make things a little bit easier for you.

You may be focused on reducing customer service costs, enhancing revenue and retention and providing a better digital experience for your customers. 

Whatever your goals, you need to make sure you’re as aware as possible of likely pitfalls before you get going, or you could have a costly mistake on your hands. 

This four-part article contains insight to help you make sense of what you need, and what you’re being offered. 

It covers some key questions to ask your candidate supplier before moving forward. We’ve also provided an example of the sort of responses you should look for (and we'd recommend making certain the answers are backed up with evidence, of course!). 

In each part, we examine questions we recommend you ask.

PART 1 – Platform & Technology

1. What does your technology actually enable us to do?
2. Do you own the Natural Language Processing technologies driving the platform or do you use 3rd parties to support this?
3. Are you channel agnostic?
4. How robust is your Natural Language Understanding? (or in other words, “How well will your system understand all the weird and wonderful ways that our customers will say what they need?”)
5. How do you test your system?
6. How will you continuously improve the CX for our customers?
7. How do you authenticate customers in-channel?
8. How do you enable feedback from customers?
9. How readily can your platform architecture scale? What impact does this have on overall compute cost/per-conversation costs?
   

PART 2 – Supplier Focus & Expertise

1. Are you solely focused on ecommerce and the post-purchase experience?
2. What expertise do you have with last mile logistics?
3. Who are your key industry partners?
4. Can you provide live stats covering the success of the platform for the post purchase needs of other ecommerce retailers?
   

PART 3 – Platform Delivery & Deployment

1. How much effort will you need from us to deploy this?
2. Couldn’t we just build this ourselves?
3. Can you help us with our channel strategy?
4. What integrations are required? What integrations already exist?
5. Who builds the conversation models and designs the flows, you or us?
6. Do you provide everything we need to get going?
7. How do you measure success?
   

PART 4 – Post Go-Live Ops and Support

1. Do we get access to data and how?
2. How do you handle customer privacy? 
3. How do CS agents resolve those queries not satisfied by the automated platform? 
4. How does your architecture handle other countries/geographies/languages? 
5. How do you handle product images? 
6. Can you both use and understand emoji?
7. What does your roadmap include for the next few years?

‍

Part 1 - PLATFORM & TECHNOLOGY

1. What does your technology actually enable us to do?

Although it sounds odd, this one is important because there are many suppliers out there who would actually fall short of providing everything you need. Directly requesting an overview of the various elements of the technology platform to determine what is actually being provided is really important. Terms like conversational engagement, automated conversations, NLP etc seem to cover everything for you, but is that the case in reality?

What to look for:

• The supplier’s platform should provide an architecture which covers everything required to deliver branded, personalised and automated conversations with your customers on any of the channels they use every day
• The existing platform architecture should be available to show you, with an overview of each existing module and a description of its functionality - e.g. security and authentication, agent handover, live analytics etc. (to ensure you’re not simply outsourcing a new platform build)
• The platform should be focused on resolving your customers’ post-purchase needs immediately and in-channel
• The platform should not be tightly coupled to the capabilities of any specific channel but should be channel agnostic
• The platform should be capable of the full delivery of the solution without any reliance on 3rd parties for any element of the experience (apart from the channel providers themselves). You do not want your supplier to include dependencies on 3rd party roadmaps.

A good response: “We can enable you to provide branded, personalised, automated conversations with your customers for them to self-serve on their favourite digital messaging channel to get their needs resolved immediately. You do not need the involvement of any third party except the digital channels themselves. We can provide a complete overview of the modules of our architecture and what part of the process they deliver.”

2. Do you own the Natural Language Processing technologies driving the platform or do you use 3rd parties to support this?

If you have a contract with a supplier to deliver your conversational AI, it is very important that this has been built by the supplier themselves specifically to tackle your challenges in post-purchase customer experience. Laying a veneer of NLP created by a third party over the top of some integrations and business logic will not provide an engaging experience. At best this will satisfy a limited number of inbound customers at a superficial level and at worst this will create an awful experience, brittle in its conversational capabilities which does more harm than good.

Also, you do not want to find out that control of the roadmap for that capability actually sits with a third party. This will mean that your requirements, sometimes urgent, may well sit way down on the prioritised backlog for the NLP 3rd party, perhaps never getting completed. This also means that the expertise for relevant NLP doesn’t sit with your supplier - an alarm bell on its own.

What to look for:

• Supplier should have designed and built the conversational capability in-house otherwise it is clear that the platform will not have been built to deliver specifically on your ecommerce requirements.
• Supplier expertise should sit across not only the full conversation orchestration but also each supporting system as outlined in the architecture doc.
• Supplier should be able to demonstrate key differences at a technical level between their focused platform and that of 3rd party tools in general use by other suppliers.
• The roadmap should be clearly articulated with detail on how client input is enabled

A good response: “We own it and we built it from the ground up. The roadmap belongs to us and we determine the prioritisation based on the needs of our clients.”

IMPORTANT NOTE ABOUT HIDDEN COSTS: Avoiding the use of 3rd party capabilities for NLP will hugely reduce your built-in costs. For instance, using the HelloDone platform instead of Lex, Dialog Flow or similar for NLP could reduce the direct NLP compute per transaction cost by as much as 95%. For more information, please do get in touch.

3. Are you channel agnostic?

This effectively means both “Can you operate across any and every digital messaging channel with B2C capabilities” and “are you future-proofed for what might come along in the next few days, months or years?”.

What to look for:

• Platform must be future proofed and channel agnostic.
• Supplier must be able to demonstrate how the platform renders conversations across the various different channels in question.
• Supplier should be able to show how rich media like product photos are delivered in-channel, especially where some of the source material may not be optimum for the channel in question.
• Supplier should have partnerships with channel providers to ensure they are developing the platform to align with any upcoming changes. 

A good response: “We can provide the same level of capability whether this is for one or multiple digital messaging channels. We have fully capable channel renderers and the platform is future-proofed to handle new channels as they become available and popular. Additionally whether a customer uses one channel one day and another channel the next, we take a customer-centric view and will identify them and the conversation against one single profile - making true omnichannel a reality.”

4. How robust is your Natural Language Understanding? (or in other words, “How well will your system understand all the weird and wonderful ways that our customers will say what they need?”)

I think we can all agree that poorly implemented automated conversational experiences can be frustrating in the extreme. Simply plugging generic solutions into open channels, supported by general NLU systems may well be a road to failure as engagement drops off a cliff and brand perception is damaged. If your candidate supplier does not demonstrate that conversation of the most robust and highest quality, designed and built specific to the use cases you have, is their technical priority... it’s time to look elsewhere.

What to look for:

• Continuous testing in place to increase the flexibility of the conversation understanding
• Supplier is able to provide positive real-time statistics for the live platform covering exactly the use cases you need
• The supplier should be able to clearly articulate the AI patterns used to deliver the experience, including an overview about which type of input is handled by which design pattern and how data sets are used to enhance the experience over time

A good response: “We have built the platform from the ground up for your precise needs. We’ve used continuous, in-depth testing, in order to resolve enquiries in the most effective way possible for ecommerce customers. We understand the use cases in this environment and have built the platform specifically for these. We can show you the live stats for the platform, in real time, to demonstrate the effectiveness of our platform in practice. We can also provide the technical background for the system build and a clear reasoning as to the AI and machine learning approach adopted. This includes a blend of deep learning using multiple neural nets in conjunction with traditional AI algorithms for improved recognition and accuracy.”

5. How do you test your system?

It is not good enough to simply “try it out”, as edge cases will be missed.  Ideally, systematic automated testing should be used at all levels, even if that requires specialist AI tools to do so.

What to look for:

• Testing at the heart of delivery for you as a client
• Scalable regression tests which can emulate inbound requests by large cohorts of users (HelloDone enable this through AI-driven test cohorts - i.e. using AI to test AI). This enables full, daily CX testing for production and high levels of confidence in terms of regression tests for each new release 
• User acceptance testing prior to go-live. You should have a direct opportunity to test out the live platform in-channel prior to go-live.

A good response: “We place testing and QA at the heart of everything we do. By creating a separate AI-driven testing engine, we have also been able to use “AI to test AI”. This allows for testing at scale with cohorts of automated customers. This is for ongoing testing and monitoring end-to-end and also regression testing any and every new release.”

6. How will you continuously improve the CX for our customers?

Conversational UIs have a unique benefit in that in order to deploy new releases, there is no requirement for updates to the front end. All deployments can be delivered behind the scenes and invisible to the end user. This means that releases can be much more frequent. Expectations therefore should be high in relation to ongoing continuous improvement of the platform at the correct cadence. However this can only take place where a focus is placed on continuous improvement and this is what you should look for from your supplier.

What to look for:

• Testing at the heart of delivery for you as a client
• •calable regression tests which can emulate inbound requests by large cohorts of users (HelloDone enable this through AI-driven test cohorts - i.e. using AI to test AI). This enables full, daily CX testing for production and high levels of confidence in terms of regression tests for each new release 
• User acceptance testing prior to go-live. You should have a direct opportunity to test out the live platform in-channel prior to go-live, not least to ensure consistency of tone, style and content with the rest of your branded comms

A good response: “We analyse the behavioural data on the platform, including the failure sets, to enable us to improve the platform. We deploy regularly, with robust regression tests in place, to ensure the experience continually improves for our clients’ customers.”

7. How do you authenticate customers in-channel?

Where customers are asking for personal information or wanting to make changes to an in-flight delivery, it’s important to ensure that they are who they say they are. Authentication in the current key messaging channels such as WhatsApp, Messenger and Instagram is not a trivial task. Doing so in a frictionless, delightful way for the customer even less so.

What to look for:

• No authentication is required in the messaging channels for generic inbound queries from your customers. This is to ensure a really quick and easy way to get standard information.
• Where a customer requests personal information or wants to make a change to their delivery, it is of course important to ensure they are authenticated. 
• There are several methods on offer, depending on client preferences.
• We focus on as friction-free a process as possible and, where necessary at all, only require authentication once per channel.
• A typical authentication process would be for the customer to receive a one-time PIN via SMS in order to continue.
• For some channels authentication can occur behind the scenes without any impact to the customer experience.

A good response: “The ability to authenticate customers in the channel they are using is a key element to the platform. We’ll show you how we do this and how we have worked hard to ensure that it is as friction-free as possible.”

8. How do you enable feedback from customers?

Data points in the analytics platforms showing enquiry resolution rates are one thing, knowing for sure that you have resolved customer needs something else entirely. There are many ways to reveal this data from customers and your supplier should be able to cover these off with you.

What to look for:

• Feedback is gathered across two categories: Passive and Active. Passive includes gathering data from customer behaviour directly and Active covers specific feedback from customers, either prompted or unprompted
• A conversational platform should have the capability to run very lightweight satisfaction polls in-channel and provide links out to more in-depth surveys
• Once the data is collated, the platform provides the information to the client in real-time 

A good response: “We enable in-channel methods for customer feedback, additionally we can work with suppliers to be able to connect customer touch points to work out exact channel impact and resolution stats.”

9. How readily can your platform architecture scale? What impact does this have on overall compute cost/per-conversation costs?

The architecture behind the scenes will have a major bearing on the scalability of the platform across your user base. You don’t want to be having difficult conversations with your supplier due to performance issues as your customers start using the conversational channel at scale. Ensure that you are able to get full details of how the architecture is put together - paying special attention to how the systems have been tested at load and how it scales to cope with both peak traffic and consistent, steady growth. 

Also ensure that you include questions about what 3rd party dependencies exist - not just in relation to how well the system can cope at scale but also what costs are involved. Dependencies on 3rd party Natural Language Processing and AI technologies (e.g. Amazon Lex, Dialog Flow etc) will inevitably incur disproportionate costs - something to avoid or your bill may be a shock after a few months.

What to look for:

• An architecture that is horizontally and vertically scalable, and demonstrably so.
• Right-sized technology stack which will flexibly scale according to need on an hour-by-hour basis
• Zero use of, or dependency on, 3rd party services - this will keep costs far lower at scale. For instance, HelloDone have been able to present customers with 50x reduced cost of compute due to zero reliance on 3rd party services such as this for end-to-end compute
• An in-house AI and NLP capability built specifically for your use-cases

A good response:

“We designed and built the technology in-house to deliver enterprise-grade deep tech capability. The platform is highly scalable and cloud-based with a robust, secure architecture designed specifically for ecommerce & logistics use-cases, not as a catch-all generic solution. Low operating and scaling costs due to no reliance on 3rd party services” 

‍

Part 2 - SUPPLIER FOCUS & EXPERTISE

1. Are you solely focused on ecommerce and the post-purchase experience?

You’re going to need a branded experience that can cope with specialised post-purchase customer requests. If vendor propositions span multiple verticals and hundreds of disparate use-cases, take that as a warning sign that your particular needs, and those of your customers, may not be well met. 

What to look for:

• Total focus on success for the post-purchase stage of customer journey
• Supplier KPIs aligned directly to your post purchase KPIs
• A platform built from the ground up and developed to tackle the challenges of the post-purchase experience
• Testing and live data focused entirely on customers in this space

A good response: “Yes we are. We are focused on the exact needs of your customers and align directly to your specific KPIs. In fact, we built the platform from the ground up for the post-purchase experience.”

2. What expertise do you have with last mile logistics?

For a supplier to deliver what you and your customers need, they cannot simply be a leader in the tech for NLP or conversational AI. Nor can they just know logistics and last mile and hope to plug in a 3rd party NLP platform from elsewhere. They must be intricately involved across all aspects of the industry in addition to bringing leading-edge AI expertise to create a successful outcome for you and your customers.

What to look for:

• Deep, specific experience in logistics and last mile
• Effective working relationships across carriers, 3PLs and other major players
• Knowledge and evidence of challenges and pitfalls
• Unique access to industry organisations
• Existing integration into relevant operational and tracking data
• Evidence of platform performance in this particular section of the customer journey

A good response: “We have deep experience in the logistics industry and established, effective working relationships in place across carriers, 3PLs and other major players. We know the challenges and pitfalls, we have seen what works and what doesn’t and we have unique access to industry organisations, data and key stakeholders.”

3. Who are your key industry partners?

You want to ensure that your supplier is able to deliver all the necessary elements to enable your conversational capabilities - a one-stop-shop. However you should also consider their partners and the relevance of these partners to your business. Could deployments be accelerated by existing partnerships and integrations? If so, then you could capitalise on these relationships and move quickly to go-live.

What to look for:

• No reliance on partners for full delivery of robust solution
• However, existing, relevant partnerships are in place and can be used to accelerate delivery and simplify integration
• Easy integration points exist should new partners become important

A good response: “We have multiple partners and could accelerate the delivery and/or remove the need for further integrations if these partners are relevant to your business right now. Equally we don’t need to rely on these relationships and we can get the platform into production without this approach, if more appropriate to your needs.”

5. Can you provide live stats covering the success of the platform for the post purchase needs of other ecommerce retailers?

In order to get a true picture of the capabilities of the platform you need not just to see it in action but also to get under the hood and dig into the data relating to adoption, accuracy, resolution rates, handovers etc etc. This data should already be representative of the use cases that you need to see with KPIs that would satisfy your objectives.

What to look for:

• Live customer data at a client level (if permission granted by client) and/or at a platform level
• A view over time of how the data changes based on either a soft-launch or a full go-live for a brand
• Insight into misrecognition and action taken to resolve unmet needs

A good response: “We can provide specific data and insight relevant to your use cases and objectives. When we go through this data in the live and periodic charts we can provide the narrative in terms of what insight was derived at each milestone, what action was taken and what the outcome was.”

‍

Part 3: Platform Delivery & Deployment

1. How much effort will you need from us to deploy this?

The answer here is rather more in-depth than first meets the eye. How much effort will be required from your technical teams for integrations? What about conversational content from customer experience people? Marketing alignment and setup? Conversational flows and configuration? Customer Service setup? The choice of supplier can mean the difference between a heavyweight, resource intensive project and a lightweight, quick to deploy solution already fit for production.

What to look for:

• Client-side effort minimised as much as possible
• Platform already includes the necessary modules to provide a robust solution for your needs and those of your customer
• Any configuration effort is minimised (and fully guided by supplier)
• Support and assistance continues post go-live to ensure ongoing success

A good response: “We can get up and running with minimal requirements on teams in any of your functions. This is because we’ve built a platform that has all the necessary modules already built for your needs. If there’s some configuration to be done, then this is kept to a minimum. As the platform moves into Business-As-Usual, we will continue to provide insight and continuous improvement, often without your teams needing to lift a finger.”

2. Couldn’t we just build this ourselves?

The age-old build/buy question is just as relevant in this case. If you opt for an in-house build, or even a bespoke professional services build, can you be sure that your outcomes will match the capability or quality that you need? And how long will it take before you start seeing a return on the large investment that it will require? Once the system reaches production you will need to rely solely on your resources and your data to continuously improve the experience and develop the roadmap - would it not make sense to capitalise on a supplier that can augment your experience based not just on your own data but on industry-wide insight? 

What to look for:

• Supplier can demonstrate that the post-purchase capabilities of their platform outweigh the likely in-house solution as they have had to meet and resolve specific challenges already, creating a full end-to-end customer journey.
• •upplier can provide statistics to demonstrate the adoption and success of the platform in post-purchase production environments already, with high volumes of customers
• Supplier can demonstrate that the costs of their platform are more efficient than the likely in-house solution as the burden of design & development has been met already and absorbed into the SaaS model
• A roadmap of increasing value can be provided as the supplier will augment the client’s capabilities with every new release of the platform
• Client’s own input into supplier’s roadmap can be evidenced to show the prioritisation of post-purchase clients needs without dilution from other client groups, industries or verticals.
• A proposal can be provided by the supplier which includes a mutual low-risk proof of value.

A good response: “We believe we are in a position right now to provide you with a capability that exceeds any solution that an in-house team would be able to put together in the medium to long term. We can show you live statistics that demonstrate not only the customer appetite but also the success of the experience. We are more than happy to walk you through what it took to get us to this point and help you in your decision-making on this point.”

3. Can you help us with our channel strategy?

Deciding when and how to open up the various messaging channels (and conversations on existing domains) impacts how customers adopt and engage with you on an ongoing basis. It’s important to get this right and you should expect this assistance from your supplier.

What to look for:

• Channel strategy is a key part of the customer experience and ongoing engagement
• Supplier can provide details of previous channel approaches across messaging apps such as WhatsApp, Messenger etc.
• The customer journey is clear in relation to your existing digital channels
• Supplier is aware and able to describe the specific onboarding process for each channel in question
• Supplier evidences full understanding of the advantages and disadvantages of each channel, the associated costs based on communication types
• Supplier is aware of upcoming changes to channels and cost models, based on partnerships with key channel owners such as Meta (Facebook).

A good response: “Yes we can help. We’ve seen the impact that channel choice has on our clients’ success and we will bring this knowledge to bear for you, as appropriate. This will take into account current digital domains and any processes you have in place for digital deployments.”

4. What integrations are required? What integrations already exist?

Before even considering the quality of the front-end experience, there are three important integrations that need to be available in order to resolve customer needs in-channel. At a minimum these are: Carrier Data, Order Data and Customer Service Integration. 

Prior to deployment you will not want the vendor to need to hook up all the various parties in order to piece together a coherent conversation with your customers. If they are doing this for the first time then there will be a multitude of time wasted in fixing the various challenges associated with these integrations, if they can complete them at all.

What to look for:

• Required integrations are already in place and evidenced
• Data flows and management docs can be provided for all necessary areas including Carrier Data, Order Data and Customer Service integration
• Demonstrations are provided for live in-channel agent handover or hand-back (or simple asynchronous raising of tickets if required)
• Clear order data integration approach provided, with documentation, if new integration is required for order data 
• Carrier data integrations are evidenced, with full description of type of integration. 
• Any integrations to enable in-flight amends to customer orders are evidenced

A good response: “We’re already integrated with your carriers and with your post purchase CX provider. We’re also integrated with your ecommerce solution for product data, however if you have an in-house platform then we have an easy Order API endpoint, fully documented, to make things as easy as possible.”

5. Who builds the conversation models and designs the flows, you or us?

Unless you have a dedicated team, experienced in producing high quality conversational contexts and UX and ready to continuously improve these based on ongoing test results and usage data, then you do not want to take on conversation flows and content design. Seriously, you don't.

So if your supplier starts talking about conversation dashboards or drag-and-drop flow creation, then perhaps raise a red flag. The expertise simply doesn’t exist (yet) in the industry to warrant total client ownership of conversations at any stage of the process. The quality of the end experience will either be low or and the internal cost will be high both up-front (to get mobilised) and ongoing (to ensure continuous improvement).

What to look for:

• Supplier has all the relevant conversation design for post-purchase in place already and is able to take you through these
• Client input into the content and style is flexible, enabling but not mandating input into the conversational content and CX
• Supplier augments the conversation model directly from platform feedback, to ensure the best possible experience on an ongoing basis
• Supplier will support the development of client expertise to reduce the reliance on supplier over time

A good response: “We have an experienced UX team who have built into the platform the necessary capabilities to resolve the majority of customer requests. We’re in a position to analyse the thousands of daily data points from real customers and we continuously augment the platform in response to this live data. We provide these flows and improvements as part of our SaaS platform and you will benefit every time we enhance the experience.”

6. Do you provide everything we need to get going?

There are many suppliers in the market who provide the various granular elements of the necessary solution for you. By the way they market themselves you could be forgiven for thinking that most of these providers will deliver everything you need. Unfortunately the opposite is true - and many times you’ll find out about this too late simply because you’re procuring this for the first time and don’t yet have the expertise to know what’s important and what you need.

What to look for:

• Engaged delivery team ready to deliver not only the tech and any additional integrations but also the project management, bespoke UX, data and stakeholder management.
• Ability to manage the onboarding of the digital messaging channel(s) in question.
• A platform architecture that includes the necessary elements to deliver the full customer experience in-channel, not just “open up WhatsApp”.
• Ability to deliver agent handover in-channel using expected flows into the necessary CS system(s).
• A team who have a clear shared interest in your success as a client.
• An engaged team post-delivery - ensuring detailed post go-live support, insight and SLAs. 
• Continuous improvement, directly informed by platform and cohort behaviour in-channel, importantly including analysis of unmet needs.
• Existing integrations with the necessary carriers and/or logistics partners.

A good response: “Yes we provide everything you need to provide robust, engaging and successful conversations with customers, in their channels of choice, to drive CS cost saving, customer retention and augmented CX. We’d welcome the opportunity to show you the necessary elements of the platform and we don’t rely on any third parties, except the channel providers themselves.”

7. How do you measure success?

In order for digital CX projects to mobilise, let alone be successful, the ways in which success will be measured should be established from the start. These cover the initial hypotheses for success all the way through to the specific, granular data points which reveal the true impact of the platform.

What to look for:

• Hypothesis driven success measures, aligned to your key strategic priorities
• Full client access to clear platform statistics, tailored to your specific KPIs
• Outcomes-based short-term project management, with a long term view of client success

A good response: “There are three primary areas from which we believe success is derived: project delivery, channel impact and client ROI. We establish and agree these from the start and ensure that all activity contributes and supports these metrics.”

‍

Part 4: Post Go-Live Ops & Support

1. Do we get access to data and how?

As is always the case, data for its own sake is useless unless you can make decisions/drive action from it. Adoption and behavioural KPIs are of course important yet arguably more important is what’s NOT working, what opportunities exist to satisfy even more customer needs and what’s trending right now that could be a big opportunity for us?

What to look for:

• Data can be provided via configurable front-end dashboards or at a lower level directly into clients’ in-house data storage and visualisation systems.
• The platform provides insight into not only the success metrics but also the failure cases to reveal where customer needs are not yet being met.
• Clustering analysis is performed on the data to determine key trends and inform priorities

A good response: “We provide you with all the access to data that you’ll need to make decisions and inform action. We provide this access either in the form of real-time data visualisations or straight-through API access. Not only that, we also analyse the data, both automatically and manually, to ensure that we provide the most useful insight to our clients on a regular basis.”

2. How do you handle customer privacy? 

Your customers will expect privacy, so both tracking customers’ orders, and data analytics, should be using anonymised customer IDs wherever possible.  

What to look for:

• Customer and data security and privacy are number one priorities
• PII is redacted from the logs to reduce data risk 
• Full GDPR compliance
• Opt-in compliance for key messaging platforms
• Platform to align to client Privacy Policies and Ts&Cs.

A good response: “Customer privacy is taken very seriously. We are fully GDPR compliant and any logging is redacted for PII.”

3. How do CS agents resolve those queries not satisfied by the automated platform? 

There will always be a proportion of the more complex inbound queries which require some human-led customer service. You hope this will be a small proportion, but it will still exist. Ensuring that the experience of this agent handover is as frictionless as possible is key to making this work well for the customer.

What to look for:

• The platform can be configured to enable multiple options for customer service handover. 
• Options range from fully integrated in-channel handover to live agent platform (e.g. Zendesk) through to asynchronous service approach such as the creation of support tickets 

A good response: “We have multiple integrations in place already with the key CS platforms. We can enable agent handovers in-channel through these integrations for the best possible customer experience. Additionally, we can enable asynchronous CS ticket creation if that is the preferred method for your customer service team.”

4. How does your architecture handle other countries/geographies/languages? 

Multilingual capability needs to be engineered in from the foundations up; it can’t be layered on as an afterthought. Additionally this isn’t just about translation but localisation i.e. not just what is said but how references and processes work in that region. Google translate is great when ordering your dessert on holiday - but I’m afraid it won’t cut it here. 

What to look for:

• We do not use 3rd party language tools because these simply provide translation, not the in-depth localisation that should be in place to create an appropriate experience in different regions.
• The architecture enables localisation in a way that differentiates between customer needs and manages the conversational content according to geography.
• Check that geolocation of customers is available if you require multi-region capability to reduce dependency on multiple platform instances

A good response: “Our systems have been built to be able to manage multiple-languages and multi-region localisation as necessary. We do not use a blunt 3rd party translation tool but work to ensure full localisation to support your needs. We can accommodate your needs as you decide to extend this capability out across regions. We can use multiple methods to determine the geolocation of inbound customer requests, depending on channel and variation of customer experience/branding across locations.”

5. How do you handle product images? 

The way that images are handled and displayed varies greatly across ecommerce platforms, digital asset management systems and conversational messaging channels. Inconsistent aspect ratios and cropping approaches are used, making it difficult to create any sort of consistent output for the end customer. 

What to look for:

• Active image handling is used to ensure that product details are presented in an acceptable way.
• Where customers have multiple items in an order they can receive a single aggregated collage image to align to the UI of the channel.
• The system uses AI driven subject recognition and image processing to render the photos at scale without any further intervention

A good response: “We have a proprietary collation and rendering tool which intelligently resizes source image data and creates image collages for multiple orders. This means that the visual brand experience is high quality and the presentation consistent, across multiple channels.”

6. Can you both use and understand emoji?

Customers will interact with your conversational platform in many varied ways, and this includes emoticons and colloquialisms. Many people use them as a useful shorthand and many people find them a helpful way to engage, so the platform had better understand not only emoji but also spelling and grammar errors, TLAs and other ways to communicate.

What to look for:

• The system not only handles emojis but also can use them, if appropriate, for the outbound elements of the conversation
• Additionally, the NLP is able to handle misspellings, grammar issues, TLAs, net-speak, colloquialisms and all sorts of other ways that people get their message across

A good response: “Our platform is able to cope with loads of different ways to communicate including emoji and shorthand phrases. Not only that but the natural language generated by the platform also uses these where appropriate in its output.”

7. What does your roadmap include for the next few years?

While you’re currently looking for a platform solution to deliver specific benefits right now in relation to post-purchase, you will want to expand the capabilities of this system once you see the impact it can have. However in doing so you do not want to increase the supplier list or create a fragmented experience. The supplier you work with needs to be a partner in your conversational comms and their ambitions should align with yours.

What to look for:

• The platform is not considered a point-solution but an evolving tool capable of meeting your growing needs in relation to automation across your digital channels.
• The roadmap is clearly presented and includes enhancements in relation to the post-purchase experience. In addition, the roadmap also recognises that there are soon going to be expectations that conversational commerce will move upstream to support not only retention but also conversion activities. 
• The future vision is one which will grow and support your conversational needs as a retailer. The platform roadmap represents the delivery of a holistic and consistent conversational capability to provide immediate resolution of customer needs at all stages of the customer journey.

A good response: “Our platform is heavily focused on your ecommerce post-purchase experience right now, including CX for order fulfilment and returns plus customer retention. That being said, we are already building the elements of our roadmap designed to satisfy more upstream features to support conversion and acquisition. These include capabilities such as subscriptions, multi-buy, product feature information and social validation through group sharing etc.”

And that's it. 27 questions you need to get on and ask your provider of conversational IA in ecommerce right now. Not an easy list to get through, yet not exhaustive by any means.

Hopefully this demonstrates just how many considerations there are in getting this stuff right - we'd like to be the people to help you succeed in, and reap the benefits of, Conversational Customer Experience (CCX) in ecommerce.

Please contact us here if you would like to ask any questions about this or any other aspect of conversational commerce.

‍