Auto news on Youtube Nov 2 2017

It would be easy to get lost in the details of how to implement the chatbot, and temporarily

forget the bigger picture.

In this quick video, I wanted to briefly touch on a few subjects that are worth considering

when designing a chatbot.

The first question we should ask is in regards the positioning of our chatbot.

Let's start with its role.

Is it going to act as an information kiosk, a customer support agent, a marketer, or a

sale representative?

Our flower shop bot example will not really look into customer orders or provide post-sale

support.

It will however provide information and in the process, also help us sell some flowers.

Keep the role in mind when designing chatbots for your own business.

Next, let's consider its purpose.

In our case, we want to help our customers in making the right choices when buying flowers,

and reassure them that the flowers will be delivered on time.

In the process, perhaps selling more flowers and cutting down on a few calls to our hypothetical

shop.

Another element worth considering is proactivity.

How proactive or reactive do we want our chatbot to be?

A more proactive chatbot will tend to solicit information from the user, engaging and guiding

them rather than passively waiting for a question to come along.

Depending on the type of chatbot you build, one approach might be more favorable than

the other.

In many cases, a combination of the two may be the best of both worlds.

Asking questions and guiding, when it's relevant, and sitting back waiting for the

user when it's more appropriate.

Tone and personality are also important.

Do we want our chatbot to sound more formal or more friendly?

Just adjusting a few words in a sentence can lead to our chatbot sounding a lot nicer/inviting

or, conversely, to come across as completely distant.

This matters because chatbots with a friendly tone and a bit of personality to them tend

to be perceived more favorably.

We start thinking that the chatbot might be more useful, even intelligent, than it actually

is.

There are no universally right or wrong answers here, but thinking about what we want our

chatbot to accomplish and how we want it to come across will go a long way as we approach

building the chatbot.

It will enable us to choose the right dialog flow, words, and tone, as well as deciding

what to leave in and what to exclude from the scope of our chatbot.

In a previous video I mentioned that our scope is common questions and not all possible questions.

When you define the positioning, so the role, purpose, and proactivity, for a chatbot, you'll

quickly find out that the scope is even narrower than that.

Which is a good thing.

You don't want a chatbot that is a jack-of-all-trades.

It's better to have the chatbot handle fewer types of questions but do a great job with

those.

What we implement should reflect the positioning, tone, and personality we set out to provide

to our users.

In terms of good chatbot design, there are three fundamental rules that you should follow.

First rule: Avoid using yes or no in your replies.

If your chatbot fails to interpret the question correctly, a yes or no answer can be misleading

or provide the wrong information.

For example, you might be tempted to set the response for "Is delivery free?" to "Yes,

it is".

But what happens when the user asks the bot, "Is delivery free or do I have to pay?"

Watson will pick up the right intent, but the "Yes, it is" answer won't work here.

There are situations, perhaps because of negatives introduced in the input, where the chatbot

will respond yes, when a human would correctly respond no.

It's far better to provide an answer that covers the topic of delivery price, without

including an absolute yes or no element within it.

So "Deliveries are free." is definitely a much better option than "Yes, it is".

This brings us to our second rule.

When possible, incorporate part of the user's question in your response.

For example, say "I understand that you're looking for help with an order.

Unfortunately, I'm unable to look into orders that have already been placed."

And then perhaps we would have the chatbot refer the customer to our phone number or

email address, in order to help them perhaps speak with a live person about their concern.

This will further cement, in our user's mind, the relative intelligence of our chatbot,

even though it technically failed to directly provide what the customer was hoping to find

out.

The third rule is in regards to the length of the chatbot answers.

Succinct and accurate answers are best.

We want to be informative, but not reply with a wall of text.

When a wall of text is an absolute must, to convey the information, then deflecting by

sending the user to an informative page is a better approach.

In the next video, we'll finally get to build our flower shop chatbot from scratch.

For more infomation >> Chatbot Course - DO and DO NOT of Dialog Design [Build Your Own Chatbot] - Duration: 4:54.

-------------------------------------------

How To Build A Bigger Biceps Peakgymnastics: - Duration: 3:34.

How To Build A Bigger Biceps Peak

My work at the Human Performance Lab at the University of Tampa allows me to investigate new ways to increase muscle activation through exercise.

One thing Ive discovered is that if you want bigger biceps, doing spider curls a different way can get you to those big peaks faster.

Spider curls are usually done off the steep side of a preacher bench, but something told me that the biceps might get more stimulation if the curls were done off the top of an incline bench.

As I describe in my article on increasing lat activation, electromyography (EMG) analysis, which measures muscle activation, is a great way to measure the efficiency of different exercises.

I used EMG to confirm my hunch: Making the biceps do more work in the starting position increases muscle activation.

The Power Of Constant Contraction.

When you do spider curls the normal way on a preacher bench, your upper arms hang perpendicular to the floor.

This puts your biceps in a relaxed position.

To really dial up the muscle activity, flex your shoulders to about 90 degrees (as if you were doing the first part of a front raise) and start your spider curls from there.

This one change enables you to initiate the curl with the biceps already contracted, rather than relaxed.

This initial contraction increases the biceps time under tension, an important factor in muscle hypertrophy.

If you start with your arms simply hanging straight down, you lose all that tension—and all those gains.

This elbows-forward position also increases the peak contraction by getting the biceps into a much shorter position than when your arms hang at your sides.

My tweak is based on the fact that your biceps brachii doesnt just enable elbow flexion.

While both the short and long heads act as elbow flexors, the long head also flexes the shoulder.

Starting in the new position works both the short and long heads of the biceps, so you get two gains for one exercise!.

Gain Big Even With A Light Start.

The first time you try this new starting position, be sure to use a lighter load than normal.

 Maintaining this new flexed starting position is hard! But not to worry.

Even with a lighter weight, youll be maximizing tension where itll do the most good.

Spider curls are harder when you do them this way because the incline doesnt provide the same upper-arm stability you get from the preacher bench.

This instability forces your muscles to engage more, which creates more muscle activation—and that means more muscle growth.

Okay, there you have it.

Now crush some arms!.

For more infomation >> How To Build A Bigger Biceps Peakgymnastics: - Duration: 3:34.

-------------------------------------------

Native Advertising Insider Ep 2: How to Build a Performance Native Advertising Campaign - Duration: 3:37.

So to me, when it comes to content marketing it all starts with

the audience that you're trying to reach and the problem you're try to solve for them.

To create really good content you have to recognize that you have a variety of

different types of customers. And they may be buying a product for different reasons.

So once you identify the groups of customers or types of customers or

personas that you have, the next question would be to ask: What kind of problem

does my product solve for these customers?

To know if your content works or not, I think it all starts with an assumption.

With your best guess. The next thing is to deliver your content and see how

it sticks with the audience. Do people actually spend time reading it or not?

And you can gauge that as the validation for your assumption.

I think for me to gauge the success

of content that we produce

I definitely look at click-through rate, that's one of the metrics because it

indicates just top level interest in a particular content that we have produced.

But then after that, I'm only looking at time on side.

To me average time on site

anywhere above for example two minutes or a minute and a half for maybe

slightly shorter pieces, it's definitely a great way of gauging relevancy of the

audience that I am targeting.

There's nothing wrong with asking people to do a certain thing after a person reads a

piece of content. For example subscribing to a newsletter or

downloading an ebook or requesting to talk to a sales representative.

At the end of the day is a decision of a user or reader to do it or not to do it.

If you're one of those brands that are so great at telling really awesome

stories that give goosebumps, maybe you can get away with telling one story or

showing one video to a user for them to make a decision about your brand. But for

us mortals, most of the time you actually have to talk to a consumer multiple

times. There are various mediums. So that's why

it's great that you got the user to a piece of content, but that's just

very initial interaction. That's why you need to think of how can you retarget

users with more content and get them to read maybe articles or watch videos that

are more bottom of the funnel where they talk more specifically about the product

versus all the company values that they were originally to interact with.

The question of attribution of content is definitely not straightforward.

I don't think there is a silver bullet for attribution that currently exists,

otherwise I think everybody will be using it. I think you can use a

combination of different analytics tools. You could certainly use platforms that

try to help you map the user journey from the first point of interaction down

to sale. You would probably have to supplement it for maybe on site

analytics like Google Analytics but I would say it's a combination of multiple

things. At least at this point in time.