What is real-time call tracking with conversation intelligence?

Call tracking with conversation intelligence is when you use machine learning to track specific attributes or activities from conversations. These could be conversations between an agent and a customer, sales rep and prospect, medical professional and patient, or interviewer and candidate.

The machine learning model tracks calls, not in the sense of volume per agent, or where they have originated from, but rather literally tracking the conversational content of the call by listening in and analyzing what the participants are saying in real time.

You can use real-time call tracking to predict the type of call that is taking place, whether it’s sales, service, or complaints. You can also use the data analysis to optimize the call experience by offering callers relevant information at the perfect time. Plus, you can use it to identify issues that may frustrate the customer or caller and avoid them, like making sure callers aren’t given repeat offers.

Once you’ve ingested the call data, you can use conversation intelligence to set up real-time custom trackers to track contextually similar words.

Who uses conversation intelligence call tracking?

Many different industries can power their technology with conversation intelligence platforms. It’s not just specific to call centers, but can also be applied to sales, telehealth, HR, logistics, engineering, compliance, legal, finance, and more.

You can set up call trackers for any business that has telephone conversations with customers, and these can be operated in real time or asynchronously. You can use conversation intelligence to analyze the conversations and optimize results – like improve sales, increase customer satisfaction, analyze trends like topics of complaints, and coach employees on what’s effective.

Here are some examples of how you can use real-time call tracking in different industries:

• Contact center: Track for service issues, bill shock, complaints, escalations, and order tracking. You can use the insights to connect the customer with the best agent, and to suggest workflows — these can be triggered in the background automatically. For example, if a customer says “my bill is too high” the machine can suggest a 20% discount.
• Sales: In a sales context, you can track product mentions, the names of competitors, as well as set up promotion/discount trackers.
• Telehealth: Track and identify symptoms, like a fever, to then ask more relevant questions. Trackers can be used as a triage system, so certain keywords could surface suggested questions to determine whether the patient has been in contact with anyone with a certain disease, or if they have been vaccinated. You can also use the insights from Trackers to divert a patient to the appropriate specialist.
• Human Resources: If a candidate is talking about a certain experience, you can suggest more probing questions, or divert them to the next round based on keywords they used.
• General business meetings: When a user says “Let me share my screen,” you can use the call tracking to automate the action of screen sharing, as shown below:

Why is conversation intelligence call tracking the smarter approach?

Conversation intelligence tracking is different from other forms of intelligence tracking because it’s smart enough to use contextually similar keywords, not just keywords specified in the model. This added level of sophistication leads to a more accurate tool for business use cases.

Intelligently automating manual processes and gaining analytics from huge volumes of data saves you significant amounts of employee time. In fact, the amount of conversation data that is analyzed during call tracking goes far beyond what a human would be capable of. The machine learns over time, never gets tired, and can do all of this in real time — and without error or bias.

You also save time because you don’t need multiple applications. Symbl.ai’s API platform has the whole process covered, all in one place. And, your customer experience is optimized thanks to the speed of execution.

Example of Trackers being used in a Customer Service industry

For example, you can use conversation intelligence to help coach agents. Managers can see and track what’s happening in real time with multiple agents on a dashboard, so you can see if there’s a complaint or if something needs escalating. You can also track whether an agent uses empathy or is polite in real time, asking the model to give live suggestions, such as a reminder to say “thank you” more. You can also get a percentage score on an agent’s performance, or if necessary, a transfer recommendation to a different specialized agent.

How to use Symbl.ai’s conversation intelligence to track calls

1. Define your trackers

Use Symbl.ai’s Tracker/Management API. In each tracker you can define an array of keywords or phrases.

2. Stream the conversation

Stream the conversation using Symbl.ai’s Streaming API  with Trackers enabled.

Here’s an example of how to pass Trackers in the config object for the startRealtimeRequest of Symbl.ai’s JavaScript SDK. This example also shows how to consume the results of the detected Trackers in real time.

const connection = await sdk.startRealtimeRequest({
     id,
     insightTypes: ['action_item', 'question'],
     trackers: [
         {
             name: "COVID-19",
             vocabulary: [
                 "social distancing",
                 "cover your face with mask",
                 "vaccination"
             ]
         }
     ],
     config: {
         meetingTitle: "My Awesome Meeting", 
         confidenceThreshold: 0.7,
         sampleRateHertz: 48000,
         trackers: {
             "interimResults": true
         }
     },
     speaker: {
         // Optional, if not specified, will simply not send an email in the end.
         userId: "john@example.com", // Update with valid email
         name: "John",
     },
     handlers: {
         onTrackerResponse: (data) => {
             // When a tracker is detected in real-time
             console.log('onTrackerResponse', JSON.stringify(data, null, 2));
             if (!!data) {
                 data.forEach((tracker) => {
                     console.log(`Detected Tracker Name: ${tracker.name}`);
                     console.log(`Detected Matches`);
                     tracker.matches.forEach((match) => {
                         console.log(`Tracker Value: ${match.value}`);
                         console.log(`Messages detected against this Tracker`);
                         match.messageRefs.forEach((messageRef) => {
                             console.log(`Message ID: ${messageRef.id}`);
                             console.log(`Message text for which the match was detected: ${messageRef.text}`);
                             console.log(`n`);
                         });
                         console.log(`nn`);
                                               console.log(`Insights detected against this Tracker`);
                         match.messageRefs.forEach((insightRef) => {
                             console.log(`Insight ID: ${insightRef.id}`);
                             console.log(`Insight text for which the match was detected: ${insightRef.text}`);
                             console.log(`Insight Type: ${insightRef.type}`);
                             console.log(`n`);
                         });
                         console.log(`nn`);
                     });
                 });
             }
         },
     },
 })

3. Capture the trackers in real-time

Symbl.ai analyzes the data from your spoken conversations within your calls in real-time and tracks various word patterns, such as “I would like to place the order but we need X and Z information first”.

Learn more about how to consume trackers in real-time here, and learn how to get trackers using Symbl.ai’s Conversation API here.

4. Get the tracking results

The last step is to build workflows or suggestive inputs in your software based on specific tracker events.

Conversation intelligence is accessible to all. Symbl.ai is a developer-focused platform of APIs, which puts the power of conversation AI into the hands of business. Learn more about Symbl.ai or sign up for your free account and get started today.

Additional reading on real-time call tracking:

• Guide to call tracking and the power of AI for analyzing phone data
• Using Trackers API (Beta) | Symbl Docs
• Streaming API Reference | Symbl Docs

The post Real-time Call Tracking with Conversation Intelligence appeared first on Symbl.ai.

The use of voice and video for oral communication has seen an increase in popularity over the last decade, especially for workplace collaboration, customer service, and live events. But with the complexities of human speech and the growing need to analyze it for business insights, standard transcription just isn’t cutting it anymore.

Let’s take a quick dive into the world of transcription, why standard methods are falling short, and why intelligent transcription is the next step.

What is transcription?

Transcription is the process of converting audio and voice into a text transcript. This is also known as speech-to-text, or automatic speech recognition (ASR).

With speech recognition technology you can add real value to communication. With transcripts, note-taking can be faster and content more searchable and accessible. Also, transcripts allow for better knowledge sharing, can summarize conversations to catch valuable business insights, and help reduce costs of recording and storing conversation data.

There are two levels of transcription available on the market today:

1. Standard speech-to-text transcription: Most providers can offer you basic transcription. This is often domain-specific and you’ll need to use data to train the AI model.
2. Intelligent transcription: This uses contextual understanding for an almost “human” understanding of a conversation, including sentiments and even cultural context. It’s also domain agnostic, which means it can comprehend speech and conversations that occur in any context.

Why intelligent transcription is better than basic transcription

• Sync and async formats: Standard transcription doesn’t offer you multiple formats, such as asynchronous and synchronous (telephony, WebSockets), whereas intelligent transcription offers both.
• Higher accuracy: Transcription accuracy is a challenge for standard transcription because there is no contextual information. Intelligent transcription is able to add real value as you benefit from features such as punctuation (making it more readable with sentence separation), and entity recognition (e.g. it can identify names, times, dates, and places).
• Speaker separation: Intelligent transcription is able to offer speaker separation for multiple speakers. This is not easy to emulate in standard transcription. With intelligent transcription, even if you cannot identify separate speakers, you can emulate this by using different channels and speaker events.
• Multiple output formats: Intelligent transcription can offer you output format options: both SRT format (standardized for video captions and subtitles) and Markdown (easy to publish in an HTML paragraph structure).  Standard transcription only offers SRT format.
• Unlimited transcription time: Standard transcription isn’t very easy to get in real time as most providers offer you limited audio length. For example, Google offers five-minute time chunks of transcription, which means you have to start a new five-minute section just before the previous one ends to preserve the flow. With intelligent transcription there’s no limit to the length of time available.
• Optimization: The biggest challenge is how to optimize your use of transcriptions. With standard transcription, you get the text script and this creates oodles of data with exciting potential. With intelligent transcription you realize this potential with insights, action items, questions, topics, and trackers.

Who’s using intelligent transcription?

Intelligent transcription has a wide range of use cases across different industries. Here are some examples:

• Many start-ups in the productivity space use intelligent transcription to give them an edge over competitors and to save time (which can make up for lack of funds or resources). By using highly accurate transcription in meetings you can create newsletters, or content from the information. As sales growth is very important for a start-up, intelligent transcription facilitates swift market research.
• Customers use intelligent transcription during live events and webinar platforms, as well as insights, so they can post their event and summarize using a real-time API.
• Start-ups working with asynchronous communications like voice notes for workplace collaboration benefit from intelligent transcription as they can segregate data by topics.
• Intelligent transcription is popular for voice-based social networks and podcast providers because it increases content accessibility by catering for format preference, and also increases audience potential by catering for disabilities. It also helps with SEO, searchability, link building and publicity.
• Contact centers and sales domains use intelligent transcription to track and coach agents by identifying what phrases work best with customers and drive better results.

4 ways developers benefit from using Symbl.ai’s intelligent transcription – “Transcription Plus”

Symbl.ai’s Transcription Plus offers you intelligent transcription services with all of the features and benefits discussed above: higher accuracy, unlimited transcription time, sync and async, multiple output formats, speaker separation, and optimization. Making it easier for you to use and provides real value to business.

Symbl.ai’s Transcription Plus provides you with the ideal solution because:

1. It’s easier for you to integrate with Symbl.ai’s plug-and-play APIs. You can be up and running within minutes. Transcription Plus works with a wide range of different audio channels, such as web RTC, telephony, and recorded files – streaming through Websocket and using the Async API post live audio.
2. Most transcription suppliers (such as AWS, Google, and Azure) offer different APIs for punctuation, diarization, topic or keyword detection, etc. This increases the complexity of the system and the pricing. Symbl.ai offers all conversation insights and intelligent transcription through one API: Transcription Plus.
3. Transcription Plus reduces a lot of your engineering requirements. As well as working with multiple audio channels, Transcription Plus also supports JSON, text, SRT, and MD output formats.
4. You’ll enjoy straightforward post-processing of your transcriptions. For example, you can use Transcription Plus’ speaker separation to have clear and accurate differentiation within conversations and Q&As.

Transcription Plus makes it faster and easier to integrate your voice or video products, making conversations more searchable, accessible, and valuable than ever. Contact Symbl.ai to start transcribing today.

Additional reading:

• Using Symbl.ai Conversational Intelligence APIs for Recorded Meetings
• Intelligent Transcription System Based on Spontaneous Speech Processing
• What It Really Means to Add Context to Your Conversation AI
• Trade-offs in Building Speaker Separation Into Your Application for Advanced Speech Analytics

The post What’s missing in standard transcription solutions? appeared first on Symbl.ai.

What is Agent Call Coaching?

Imagine a world where the only data available to a call center manager was the percentage of resolved customer calls. For example, let’s look at two agents. With this kind of data, you can easily see that Chloe can resolve 67% of her calls, but Natasha only 45%. However, you cannot determine the best way to resolve it without deeper insights to reveal why this is happening.

How do you know the best way to help Natasha succeed? What explains the difference between Chloe and Natasha’s results? Is Chloe more polite than Natasha or more empathetic?  Maybe she’s better at handling specific objections, which helps her improve conversions or resolutions. Any of these reasons could explain Chloe’s success relative to Natasha’s. Understanding the specific reasons for these differences could help you provide Natasha with more personalized call coaching to improve her call handling. But, how do you figure out what they are?

Call coaching is often limited in its effectiveness because without deeper insight into your team’s performance, you have nothing to inform the kind of training you might use to improve it.

Firstly, not all calls are recorded. The industry standard is between 2% and 5%, which is a small sample size. You may be missing out on important call information, which could be vital in training or making improvements to certain aspects of calls.

These calls could be selected from a range of different call types, making it difficult to focus on one department or one type of call because of the small sample size.

With the small number of calls selected, the quality team will listen to them. Their job is to analyse the calls and identify improvements for the relevant departments. Depending on the size of your business, the quality team could be large, adding costs to your business.

Listening to several calls can be time-consuming as you have no idea what each call is about or how the call was handled. Therefore you would need to listen to all recorded calls.

You also can’t give feedback in real-time. This means it could be days or weeks before feedback is given on a certain call rather than going to an agent and telling them how they can improve the call they’ve just handled.

If you want to make things easy when call coaching, using artificial intelligence (AI)to monitor and extract insights from your calls is a game-changer. It allows you to be proactive in your coaching because the analysis can be done in real time, while the agent is on the phone with the customer and makes it easy to drill down into certain calls without listening to them all.

Symbl.ai’s Conversation Intelligence can help do this.

Symbl.ai’s conversation intelligence takes call coaching to the next level

Conversation intelligence uses AI to analyze text or speech to gain insights into calls. You can integrate Symbl.ai into your video or audio applications without having to build the complex machine learning models typically required for AI.

There are different types of data you can gather from analyzing calls but these are the most helpful to call coaching:

Business-Specific Insights

Custom Trackers help you track when certain words or phrases are said in a conversation. You’ll be able to spot trends or see how well a conversation went. This information will let you address if any agents aren’t using the right language with a customer.

For example, you can create a list of certain phrases or keywords around a context and Symbl.ai will be able to look for contextually similar phrases. When it comes to detecting phrases in a conversation, the offset value will show you if there’s a match.

Let’s look at an example of a tracker list. Tracy, the head of training for the sales team is looking to see how well her team does at the call to action at the end of a sales call. She uses the CTA Tracker to gain insights. The CTA Tracker uses contextually similar phrases to:

“Do you want to buy it?”
“Is this of any interest to you?”
“Do you want me to process this order”
“Who shall I make the invoice out to?”
“When would you like this delivered?”
“Let’s discuss pricing.”

Questions

Symbl.ai can automatically identify and generate  questions or anything deemed as a request for information from any conversation. Monitoring what questions are asked over multiple calls can help identify those most commonly asked. This information can be used to train new agents or run workshops for existing agents so they know how to answer these questions correctly.

Action Items

In our conversations with people, there can be times when we have agreed to do something. This could be arranging a meeting for next week, calling a customer back within 48 hours, or checking the stock level for a certain item.

Armed with a list of action items from calls, you can identify common tasks and determine how to best provide support to ensure they are completed. You can also see how the team member did at actioning these tasks and give positive feedback for those completed promptly.

Use Cases

1. Refining customer service responses

By using Trackers, you can help your team members improve their responses to customers. This can help you improve customer satisfaction levels, increasing Customer Satisfaction(CSAT) and Net Promoter Score (NPS).

If you set up a tracker list of positive phrases that are common in successful conversation such as “Thank you,” “You’ve been helpful,” or “It’s been good to talk to you,” you can analyze the call to view how often contextually similar phrases are being used.

If you find a team member that has low scores, you can spend time improving the language that the agent uses to speak to customers. You could also let a low-scoring team member shadow or listen to calls from a high-scoring team member to learn from.

2. Address questions before they’re asked

If you find that someone on your team is constantly being asked the same question over multiple calls, then you can address this by adjusting the script to anticipate that question.

This will allow the team member to be more proactive and address the customer’s problems before they even occur in the customer’s head. This enables faster resolutions to customer issues to improve the overall customer experience, resulting in increased customer retention.

3. Monitor performance

You can use an Action List to track the performance of your customer support team, monitoring each conversation and comparing them with the number of support tickets. This will show you how well your team is able to resolve issues on the phone instead of having to raise a ticket or escalate to a higher level of support.

An Action List can also give you insight into whether there are tickets being raised as tickets that could be dealt with on the phone. This introduces the opportunity to know where additional training is needed to help your agents resolve more issues during their calls, improving customer service and reducing the number of tickets raised. 

4. Increase conversions 

Easy access to in-depth call data can help you increase your conversions. If you’re analyzing calls in real-time, you can also enable suggestive inputs for cross-sell/upsell opportunities, and other possible services or products that might interest the customer or solve their problems.

Reviewing action items can also identify customer pain points providing insights for innovation – ideas for new products or services that you can develop to help with cross-selling or upselling.

Conversation Intelligence in your apps for coaching purposes

With Symbl.ai you have multiple options for integration. You can either analyze calls in real-time via the Streaming or Telephony APIs, or you can take recorded phone calls and analyze them asynchronously.

Symbl.ai’s SDK makes it easy to integrate into web conferencing platforms like Zoom and is available in JavaScript or Python. Each of these approaches is described in Symbl.ai’s SDK documentation.

The Streaming API takes advantage of JavaScript’s Websockets. This is the best option when you want real-time results while processing live audio. The Streaming API can be integrated directly via the browser or server.

The Telephony API can help you get results from real-time calls based on PSTN and SIP protocols. You can make use of the Symbl.ai’s SDK to help with integration. You can find tutorials for this approach in Symbl.ai’s Telephony API documentation.

One of the easiest ways to analyze your calls is via the Async API. It can provide analysis of text, audio, or video files. When you upload the files to the Async API, you get a ConversationId which allows you to retrieve the insights you’re after in your calls, such as questions, action lists, and much more.

Take your call coaching to the next level with conversation intelligence. Symbl.ai can help you improve call coaching to empower your agents with better skills and improve customer service. Sign up for a free account today to explore all the possibilities!

Additional reading

• 9 Quick Tips to Improve Your Call Center Coaching
• Symbl.ai’s official documentation 
• Get started with Symbl.ai in 15 mins
• Join Symbl.ai’s Slack community
• Symbl.ai blog

The post Using Conversation Intelligence with Real Time Call Tracking for Better Call Coaching and Results appeared first on Symbl.ai.

What is intelligent transcription?

Normal transcription is the process of converting audio and voice into a text transcript. This is also known as speech-to-text, or automatic speech recognition (ASR). Intelligent transcription uses contextual understanding for an almost “human” understanding of a conversation, including sentiments. Intelligent transcription can be done real-time or from a recording (async). It’s also domain agnostic, which means it can understand words from different domains and apply that knowledge to contexts it hasn’t specifically been trained for. With speech recognition technology, you can make transcriptions intelligent, adding real value to communication, to enhance note-taking, compliance, and making your content more searchable and accessible. Also, intelligent transcripts allow for better knowledge sharing, can summarize conversations to catch valuable business insights, and help reduce costs of recording and storing conversation data.

What’s intelligent transcription used for?

Intelligent transcription is highly applicable to everyday life. Some real-world examples include live captioning, accessibility, and transcribing whole video and audio files into text.

Live captioning

Live captions help with focus, engagement, and information retention. They can even boost your SEO because search engines index text, whereas they don’t index audio. Live captioning is often used for online learning because it allows student engagement in real-time. It’s also invaluable in sound-sensitive environments, like if someone is taking a class in a noisy home or with a sleeping child. In fact, studies have shown that the majority of people watch videos with the sound off!

Accessibility

Intelligent transcription makes videos accessible to viewers that are hearing-impaired. Not only does this inclusion increase the potential audience, but it also enables compliance with accessibility laws to ensure, for example, that local governments don’t discriminate against groups of people when live streaming town hall or board meetings. The same is true for businesses removing barriers when sharing content or live meetings when it’s estimated that 30% of the workforce has a disability.

Transcribing whole video and audio files into text

Intelligent transcription has the advantage of making your video and audio content quicker and easier to distribute, giving you the ability to reach a wider audience. For example, you can take the transcription of your live webinar and create written content or thought leadership material from it, or broadcast the contents of a business meeting beyond the team that attended.

How is Symbl.ai’s Transcription Plus different?

Symbl.ai’s speech-to-text intelligent transcription product, better known as Transcription Plus, offers you conversation intelligence with your speech-to-text. Transcription plus is specifically designed to reduce engineering requirements for developers. Let’s take a closer look at a few more features and capabilities:

1. Real-time and async, with multiple inputs: Whether it’s video, audio, text or phone, Transcription Plus offers you multiple formats, such as asynchronous and synchronous (telephony, WebSockets).
2. Multiple outputs: You can choose SRT format (standardized for video captions and subtitles), Markdown (easy to publish in an HTML paragraph structure), JSON, and text.
3. Optimization of your video and audio content: Detect insights, action items, questions, topics, and trackers.
4. Formatting: Auto paragraph generation (this is Markdown format).
5. Accurate punctuation and entity recognition: You can benefit from features such as punctuation (making it more readable with sentence separation) and entity recognition (to identify names, times, dates, and places).
6. Custom vocabulary: You can program your model if you need specialist understanding or have frequently used words. For example, you might want the word “sell” to be transcribed as “sell” more often than “cell.” Here you would use speech adaptation to bias the transcription to recognize “sell.”
7. Speaker Separation/Diarization: You can create speaker separation for multiple speakers, or if you can’t identify separate speakers, you can emulate this by using different channels and speaker events.
8. Supports multiple languages and accents: Symbl.ai supports more than 20 languages including English, Russian, French, Italian, Hindi, Japanese, and Spanish. Symbl.ai also offers speech recognition models that are fine-tuned for different accents, for example, to understand the differences between spoken American and British English.
9. Sentence level sentiments: Sentiment analysis at the sentence level lets  you determine whether the speech is positive, negative, or neutral.
10. Unlimited streaming length: There is no limit on the time available for Symbl.ai’s live transcription. Other standard transcription providers, like Google, only offer five-minute time chunks of transcription, which means you have to start a new five-minute section just before the previous one ends to preserve the flow.

How to get started with Symbl.ai’s Transcription Plus

Symbl.ai offers you Messages, which is the endpoint of the Conversation API, specifically for speech-to-text transcription. “Message” refers to a continuous spoken sentence. To get started with Symbl.ai’s Transcription Plus, there are two ways to hit the ground running. When you want to ingest or process files, this can be done in real-time or async with each of these options having a different Symbl.ai API endpoint. Let’s take a look at each:

Real-time  

With real-time,  can get your speech-to-text transcription in real-time using Symbl.ai’s Streaming API for WebSocket Protocol, Telephony API for SIP/PSTN, or with Symbl.ai’s SDKs for JavaScript or Python.

Asynchronous

With an async file, you need to use Symbl.ai’s Async API. This provides a REST interface that helps you submit any recorded or saved conversations to Symbl.ai.

Symbl.ai’s unique “ConversationId”

When you process any conversation through Symbl.ai, whether it’s from Async API, Javascript SDK, Python SDK, Telephony or Streaming API, you’ll always receive a unique conversation identifier (called the “ConversationId”), which consists of numerical digits and is unique to your conversation. Your ConversationId is the key to receiving conversational insights from any conversation (async, real time, or text) processed with Symbl.ai. Here’s a simple API call as an example that grabs the speech-to-text transcription from the conversation. You can process any text payload with the Async Text API, or audio file with the Async Audio API.

Sentiment analysis with Symbl.ai’s ConversationId

You can also get sentiment analysis – the interpretation of the general thought, feeling, or sense of an object or a situation – using your Symbl.ai ConversationId in the Sentiment API. All you need to do is pass query parameters sentiment=true. Here’s an example of the sentiment analysis API response. “Polarity” shows the intensity of the sentiment. It ranges from -1.0 to 1.0, where -1.0 is the most negative sentiment and 1.0 is the most positive sentiment.

{
     "messages": [
          {
              "id": "6412283618000896",
              "text": "Best package for you is $69.99 per month.",
              "from": {
                  "name": "Roger",
                  "email": "Roger@example.com"
              },
              "startTime": "2020-07-10T11:16:21.024Z",
              "endTime": "2020-07-10T11:16:26.724Z",
              "conversationId": "6749556955938816",
              "phrases": [
                 {
                     "type": "action_phrase",
                     "text": "$69.99 per month"
                 }
              ],
              "sentiment": {
                 "polarity": {
                     "score": 0.6
                 }
             }    
         }
       ]
}

The Symbl.ai platform offers APIs for all your conceivable, intelligent transcription needs. Get in touch today to get started.

Further reading:

• GET Speech to Text
• https://docs.symbl.ai/docs/concepts/speech-to-text
• Captions increase viewership, accessibility and reach

The post Getting Started with Symbl.ai’s Intelligent Transcription: Transcription Plus appeared first on Symbl.ai.

What is intent tracking?

Conversation or user intent data is the information you can get from conversations that gives an insight into the speaker’s intent. For example, if a telecommunications customer calls and says, “I want a data plan,” or “I’d like to port my number,” the intent is clearly to obtain a data plan or transfer a phone number.

As humans, we understand this request and its intent within the conversation. But at scale, this can be a challenge. Conversation intelligence allows you to identify and track customer intent from conversations by tracking contextually similar keywords. This type of data will reveal the specific intents that are driving customer conversations at scale. And in real time, the same intents can be used to derive timely suggestions for your team members during their conversations with customers.

The intent data you can get from conversations is particularly useful for sales, marketing, customer experience, and automation teams. You can use conversation intent data to build a clear understanding of things like your target’s pain points and buying motivations, topic searches, and intent velocity (how active they are around the topic). Intent tracking follows and analyzes intent data, which helps you suggest the next best action (NBA), upsell/cross-sell opportunities, and trigger workflow automation.

Traditional and new intent tracking methods

Considering someone’s intent during a conversation is not new. Gathering data for actionable intent insight has existed for a while, and many larger businesses already consider intent in their marketing and sales strategies.

For the last decade, intent tracking has been predominantly achieved using natural language processing (NLP), but now you can use the more sophisticated natural language understanding (NLU) to track intents.

With NLU, you can track intent through contextual similarities. This includes an additional layer of timing, relevance, and context within campaigns. For example, the insight that intent data provides can help you understand exactly where a customer is in their buying journey based on what they’re asking about. It also provides clarity on the sentiment attached to certain intent (positive or negative), giving a more comprehensive view than traditional profiling or behavioral data.

Using intent recognition systems with conversation data

NLP of conversation data is useful when a human speaks to a machine or chatbot. This works well for conversations that are structured, such as a series of scripted questions with yes/no answers. However, the limitation is that the person speaking needs to use essential keywords to get the correct information from the machine. Otherwise, the machine won’t understand what they’re saying and the results won’t be reliable or accurate.

In a natural human to human conversation, sentences won’t be structured or follow a standard format, which leaves NLP struggling to process keywords across multiple sentences. This is where NLU comes in – it power AI that passively listens to customer conversations and provides you with actionable insights.

Contextual similarities and associated keyword frequency with Symbl.ai Trackers

Traditional systems focus on specific keywords, but there are almost always several other similar keywords that could be used in natural conversation. Understanding intent contextually is more accurate because you can track anything that makes similar sense to the chosen or required keyword(s).

With Symbl.ai’s Trackers API you can recognize contextual similarities at scale, tracking custom intents by configuring key-value pairs with a few example keywords – all without requiring any training data. The more keywords you have, the more context you’ll create.

For example, if you have Symbl.ai’s Tracker setup for order confirmation, with keywords such as “Order is confirmed,” then contextual words such as “done,” “booked,” complete,” and “good to go,” will also register as words to track.

Business use cases

Sales 

Symbl.ai’s Tracker can be used in sales to identify qualitative attributes of agents; like whether they’re polite, how they handle objections, and making sure they stick to the script. You can track each occurrence of keywords and contextually similar phrases, such as “thank you,” and “let me help you with that.” This helps to identify strengths and weaknesses in certain agents to understand what’s working well and where some agents might need additional coaching.

You can also use Symbl.ai’s Tracker for large-scale tracking of conversations and even tag what’s happening, such as whether there was a sale, no sale, or a “maybe” sale (e.g. a request for more information). From this data you can track why something worked or went wrong.

Workplace meetings

You can use Symbl.ai’s Tracker to identify specific projects or issues within a team. Symbl.ai’s Tracker can be used in real time and can also be really useful for async collaborations, such as with voice notes. In either scenario, the Symbl.ai Tracker API will help a project manager keep up to date with how a project is progressing. 

How to use Symbl.ai Trackers

There are three simple steps to get started with Symbl.ai Trackers:

Step 1: Create a Tracker. The first step is to create a Tracker with a set of phrases and keywords using Symbl.ai’s Async API.

Step 2: Enable Trackers. When you have made a real-time or async event, you just enable trackers and then they’ll begin. Using the conversation_id you get from step 1, you can GET the Trackers for the conversation.

Step 3: Check the output. At the end, go back and check the Trackers’ output.

You can also update and delete your trackers.

Here’s an example of Tracker keyword instructions to identify any mention of a promotion:

Start tracking conversation intents in your business today using Symbl.ai’s Trackers API.

Additional reading:

• Trackers (Beta) | Symbl Docs
• Using Trackers API (Beta) | Symbl Docs

The post Are you still tracking intents the old way? appeared first on Symbl.ai.

Why we are so excited about TADHack 2020!

• We released our developer platform in March 2020 and this is the first hackathon we are sponsoring to inspire and equip you with the tools that we have built with early customers and developers to analyze conversations at scale.
• To create evangelists, champions out of developers by getting you excited to use Symbl and build unique conversational intelligence experiences. We are always growing our family .
• Receive product feedback and see different use cases that your creative juices will yield. Tell us how we can do better, we are always improving!
• Yes, to generate more sign-ups and fuel growth – we are a fast growing startup and want to keep it that way 😀

Register for TADHack

Recapping – What is Symbl?

Symbl’s APIs unlock machine learning algorithms that can ingest any form of conversational data to identify actionable insights across domains and channels (voice, email, chat, social); without the need for upfront training data, wake words, or custom classifiers.

See what all you can do with us 🙂

Also sharing below a list of key concepts that will help you understand the capabilities of the platform – like what are non-definitive action items ??!!

Real time Transcription
Symbl generate real time speech to text that can be used a live captioning or a searchable transcript with single or multiple audio streams. The transcription is generated with word level timestamps, speaker information and can map to all the other AI capabilities using message ID. The transcription is available in multiple languages and you can use custom vocabulary to enhance the accuracy of speech recognition. Read more here.

Contextual Summary Topics
These are the most relevant topics of discussion from the conversation that are generated based on the combination of the overall scope of the discussion on this topic and the relevance of the topic. Summary topics are not detected based on the frequency of their occurrences in the conversation, they are instead detected based on context and hence map to a group of message IDs or a defined scope in transcription that they refer to. Read more here.

Action Items
An action item is a specific outcome recognized in the conversation that requires one or more people in the conversation to act in the future. These actions can be definitive in nature and owned with a commitment to working on a presentation, sharing a file, completing a task, etc. Or they can be non-definitive like an idea, suggestion or an opinion that could be worked upon.  Action items are not biased on keywords or type of conversation and are generally identified to fit most use cases. All action items are generated with action phrases, assignees and due dates so that you can build workflow automation with your own tools.

Follow-ups
This is a category of action items with a connotation to follow-up a request or a task like sending an email or making a phone call or booking an appointment or setting up a meeting.  All action items are generated with action phrases, assignees and due dates so that you can build workflow automation with your own tools like Calendar, Project Management, CRM etc. Read more here about action items and follow-ups 

Action Phrases
The action_phrase type represents the actionable part of an insight. Read more here. 

Questions
Any explicit question or request for information that comes up during the conversation, whether answered or not, is recognized as a question. Read more here.

Speaker Diarization
Detect and separate unique speakers in a single stream of audio/video without need of separate speaker events. Read more here.

External speaker events or independent audio stream integrations:
Speaker Events can be pushed to an on-going connection to have them processed while using a single stream of audio for multiple speakers. Read more here.

This might come handy! 

1. Visit Documentation 
2. Sign-up on the platform 
3. Getting Started with Postman 
4. YouTube Videos 
5. Tutorials and How to Guides 
6. GitHub Repo

We are excited to see more experiences and use cases that you will build and are excited to support you during / before / after the hackathon!!

During the hack →

• Have fun! This is the new normal of hacking – so have fun at home! We wish to get together soon in a place once things settle. Build something you want to build – and we will support you through the process.
• Try something new – new language / use case / solve a problem for you! We have most of our team available for you – to ask questions or learn.
• Drink lots of fluids 🙂 Wanna make sure you are hydrated.
• Demo it – tell us the why and what  ! Even if it’s not working, show the “art of the possible” 😉 Getting NLP and NLU right is hard, don’t worry about scaling generally, focus on one specific scope and we can take it from there.

What can you do after the Hackthon?

If you want to keep digging into this space even after the Hackathon, we have some cool things coming that you might want to try out! If you are interested in the early versions of these API, request early access by sending an email to devrel@symbl.ai with the subject “Count me in!”, and we will follow-up with you!

• Sentiments by Contextual Topic
• Hierarchical Topics outline parent and child topics with timeline of conversations
• Conversational Analytics like Talk time, Pace, Overlap
• Symbl JS Elements

Happy TAD-Hacking 2020!

Register for TADHack

Missing something that you might want us to figure out how to add? Write to us at: devrel@symbl.ai or join our Slack channel. Also, we will reach out to you after the hackathon to get your home address for sending some swag and love in your mail!! See you all virtually.

The post TADHack 2020: Hacking with Symbl appeared first on Symbl.ai.

Think of it like ‘IF — (specific) Conversation, THEN — (specific) Action.’

We’re building Symbl’s conversational intelligence platform to specifically enable simple, low-code workflows to derive outcomes from conversations immediately. Let’s explore a few areas where conversations can lead to instant actions and help automate processes .

Workplace Productivity

Tools like Slack, Zoom, Loom, Jira, and Google Drive have made business conversations and interactions easier, whether employees are working together in the same office space or remotely.

Here are a few workflow examples to boost workplace productivity through real-time actions in natural conversations:

• Automatically assign a task to someone on Slack/Jira from a meeting or stand-up
• Talk about sharing a file in a meeting and have it sent automatically to Google Drive
• Make all your team demos searchable with transcripts and topics
• Instantly share updates from interviews and stand-ups

Inside Sales

Many sales conversations happen on conferencing platforms or phones that require tracking on CRM systems and immediate follow-ups.

Here are a few easy workflow examples that streamline sales pipelines in real-time without manually entering data:

• Schedule any follow-up calls automatically on the calendar
• Update opportunity-specific details to the CRM instantly
• An aggregated view of real-time sales discussions on specific topics, mention of competition, etc.
• Automatically progress leads to the life-cycle stage to set up demos, reviews, etc.

Contact Center

At high volumes, it’s hard to maintain the best customer experience and agent training quality in contact centers. But with continuous conversational intelligence, contact centers can reduce average handling time and increase customer experience.

Add these additional workflows to provide an exceptional experience for customers and contact center employees:

• Use conversations to automatically enable document verification with RPA
• Set up and communicate customer visits and appointments automatically
• Build a live, searchable knowledgebase using historic customer service and on-going support conversations
• Automate backend processes with no/minimal user intervention such as invoice generation, collection, etc with RPA
• Automatically recommend relevant knowledge base articles to help agents have a smarter conversation

Symbl flows

There are endless possibilities across industries and use cases of using workflows that can be triggered from conversations. Early customers and developers on our API platform are currently building similar workflows. We want to enable our customers and developers to pre-build, low-code steps for similar workflows. We’ll soon publish some of them as steps for Flows on the Flow Manager (beta).

Here are some sample Flow implementations we’re excited about:

• Customer service conversation using Twilio Voice → Extract insights → Update on Salesforce CRM
• Tech teams on any meeting platform→ Extract Insights & Actions → Update Jira tickets
• Inside Sales call → Extract Insights → Update opportunity on Hubspot

Interested in building workflows through Symbl?

Sign up on platform.symbl.ai and learn how to create flows

If this blog inspired you to create your own workflows, let us know
by writing to dev@symbl.ai

In the future, we’ll be able to speak things into action. Let’s build that future together.

The post Extend Conversations to Workflows appeared first on Symbl.ai.