Product, UX, Marketing, Sales, Customer Support- How to know who screwed up!

Ok this is going to be a quick post. Using your top level working knowledge of GA you can understand broadly what is simmering underneath.

The first number everyone looks at is the Top Line. How much revenues are coming in.

Top line is simply  - No.of orders * Average order value.
No. of orders = Traffic * Conversion rate


Consider a typical vanilla day, with no big marketing promotions/ discounts running on the site.

1. If the number of orders is a constant ( normalized to other 'such' days) and top line goes down, clearly Sales and Product pricing is the issue. Or probably a new low value popular product cannibalized the sales of other high value products.

2. If the "No of orders" is going down multiple scenarios originate :

A. Conversion Rate Stays The Same

Check Traffic. It is in the RED (wrt what its supposed to be )

The Marketing head probably decided to switch off one or more channels of traffic due to poor ROI. What is good or bad ROI is for another blog topic.

Or may be something did get screwed.

To drill further, check the Traffic - Source Wise.
Leave the rest of the forensics to the Marketing dude.


However, before you make up your mind to fire your Marketing dude it is worthwhile to check out the graph for Direct traffic + Brand organic traffic coming in. If thats on a downslope chances are all other marketing channels are following a similar trajectory.

Direct traffic + Brand search traffic is a good reflection of your Brand pull. This Brand pull is not 'always' in the hands of the Marketing Officer. It might be getting heavily influenced by Customer Support or Delivery or Quality issues or because of heavy competition. Brand then becomes the responsibility of all the business functions.

Generally, first the Return Traffic for "Direct traffic + Brand search traffic"segment will show signs of weakness. As negative word spreads the direct traffic takes an even more steeper slope.

B. Conversion Rate Dropped

Conversion rates are a function of -

- Quality of traffic- Quality refers to your traffic-product fit.- Intention to buy from your store. If the fit is poor you will see higher bounce rates. Controlled by the Marketing dude.


- Quality of the stuff you sell- If your products are not worth buying, people are just not going to buy. Simple. Check the conversion rate for the Returning Traffic only for just Direct & Organic traffic sources. That's a pure indicator.


- Pricing- If your customer segment is very price sensitive, product pricing becomes absolutely crucial. Users check multiple stores for the same or similar products. This becomes tricky for a product manager since you have to balance conversion rates and average order value. As a store manager you want people to buy costlier stuff but stuffing more costlier products also mean less people buying ( aka lower conversion rate)
Check the conversion rate for the New Traffic. Again, for just direct and organic traffic sources. That should be a good indicator.


-Ease of purchase- If users have to spend a lot of time finding what they want, they will get tired and bounce off. This probably would not be an overnight change, but overtime this can be very significant.
This is where product personalization, recommendation engine & site search robustness come in.
Directly impacts conversion rates.


-Social Elements- Adding features that  social influence, urgency, social proof, etc is known to engage customers further and develop brand trust.


-Customer Support- Keep a regular check on the Customer C-Sat and D-Sat score. One of the most crucial and revealing number. If business numbers are not helping enough to make sense of the current situation it is best you call the customers themselves ! Check the support@yourdomain.com and feedack@yourdomain.com emails. There is a good chance the solution to all your problems already lie there. They usually are...

Application of Interest Graphs- Uncovering The Mystery

For the uninitiated let me try to explain very quickly what they mean. 

The Interest Graph is a representation of the relationship between people and things.

In general, the ’things’ are things of interest to the people or things people care about. Actively or Passively, Passionately or as a Fringe interest. They can be- celebrities, brands, objects, topics, events, industry, people, anything really.

                

This concept probably arose in contrast to the Social Graph, which is a representation of the relationship between people and other people, as popularized by Facebook .

                                              

                   

It’s often a 1-way relationship or an asymmetrical relationship as opposed to a Social Graph which is typically a 2-way "friend" relationship.

  

Also, now the talk of the town has become something called the Taste Graph.  

Taste Graph is just a way of saying “refined interests”.  Saying something like you like sports, movies or music is a broader Interest graph, but saying you like Tennis and Dark Humor Movies and Rock Music gives more depth to your Interests aka a Taste Graph. You can ofcourse go deeper and say you like Nadal and Tarantino and Jim Morrison, refining it further.

Ok. So what’s the fuss about?

Knowing what you are really interested in can be put to a lot of use. Useful to You! 

Creating such an engine however requires a lot of heavy thinking and designing rather complex algorithms and tree structures. 

Using this data to give relevance to users is not an easy job and certainly not cheap. Not always accurate too. 

However, in most cases the juice is worth the squeeze. 

Here’s why:

1. Digital marketing- Creating awareness about products you might be interested in based on your browsing history. Based on your interest graph/ taste graph, ads can be served that are more relevant to you and have a higher probability of catching your attention. Hence generating better CTR's and better ROI.

2. Product Recommendations- Amazon is the best example. The site has all the elements of the equation: your purchase history, product search history and even product correlations (people who bought this also bought that).

Note how in each of the above cases you don’t explicitly mention your likes and dislikes. The system "learns" based on your behavior and passes recommendations. This is a method of what is called 'implicit data collection'. 

On the other hand asking for your preferences and then personalization is a way of collecting data 'explicitly'. 

However, using explicit information from Social Networks to provide recommendations might prove tricky as the bio created might be a projected image instead of a true image.

Other Interesting Applications:

3. Content Discovery: There has been a lot of movement in this category with regards to personalized recommendations .What news topics might interest you, which songs you are likely to like, which movies you are likely to enjoy and all of this with a fair bit of accuracy.  

However there are some significant challenges here. What if they don’t have enough data about you to make good recommendations. Even if they did, there is a risk of the signal being merely incidental or the recommendations thereafter getting too repetitive.

Maybe, they should just ask you. Also, you have no reason to project an image to an algorithm! Companies such as Hunch have attempted to address this by asking users to share interests explicitly, with the incentive that users will get more accurate recommendations in return. 

However, this approach has some psychological hurdles. The amount of data that one has to volunteer in order to get a decent recommendation or deal is so high that he/she might as well just search for the product directly. The return on the effort put does not fit in. No ROE if you will. 

What can therefore be a high enough incentive to collect user data by directly asking them?

Welcome to the World of 'Social Discovery'

In the easiest of languages 'Social Discovery' means one user finding out relevant information about another. 

Can two Interest graphs talk to each other and activate 'Social Discovery'. For instance- 

4. Employee - Employer Discovery: An employer is always interested to hire that perfect guy with the right skill sets and the right culture fit. 

Most individuals as we know are mostly, at least, passively looking for a new job. 

Some not happy with the pay, some unhappy with the work hours, some with the work culture, most cant stand their immediate bosses and a hundred other reasons. A research conducted says that in the services sector that number stands at a staggering 83% !! Sorry state of affairs- yes. But does this also represent a big opportunity to match the right employer to the right employee. Clearly, a keyword search based system to match doesn’t seem to be working out that well.

Its probably time for companies to start creating their own Interest graphs and publishing it to attract the best fit talent. For individuals to start creating their own about.me pages with information about their professional (Linkedin) AND their lighter side (Twitter handle) 

Can systems then map the signals present in the two Interest graphs and send auto suggestions...!!

Can this finally mark the end of the ubiquitous resume !

5. Dating & Matrimony - 

Probably the most romantic use of Interest graphs can be to solve the biggest problem of them all. Finding the right life partner!! Now we are really talking! 

Certainly a product built on the Interest graph fundamentals seems to have the best chance of driving those initial conversations, help break the ice between the individuals, so crucial for the success for any product trying to ignite the chemistry!

Btw, the social graph layer to back identities might prove mighty useful here.

Well, Leaving the rest of possible applications to imagination and discovery.

Like it or not, the web Is changing every single day. It keeps on growing bigger and we keep on getting more addicted and dependent on it. The internet needs direction. We need to start moving from searching the internet to using it. 

The personalized web some argue is going to be the future of the web. And the personalized web is actually just one interest graph away..