Product data and content syndication

Product data & content syndication involves managing product data & content & the smooth distribution to different omni channels of retailers

Typically, most retailers would have their own technology to manage product data and at least 3 out of 5 organizations would use a PIM (Product Information Management) system to capture, store and workflow product data.

Along with that there would ideally be a DAM (Digital Asset Management) system to manage digital assets like images, videos, rich media content and the ERP (enterprise resource planning) system. Of course, there are other important systems like a CRM, CMS, PLM etc. which would also be part of the technology eco system of any ecommerce player.

In the gamut of syndication, PIM & DAM play an important role, as the core ideology of syndication is to create a hub of product data and content. From the hub the content can be distributed seamlessly into several channels. PIM & DAM are essentially hubs for product data & content respectively.

Many technology service providers nowadays are trying to bring the best of these 2 systems together to create the ultimate hub or a Unified repository of data and content so that it serves as a good platform to provide a CaaS (Content as a Service) solution.

But there are several challenges to do that, especially for big retailers with millions of products and digital and rich media assets. So, an integrated solution is still a viable option that most retailers are going for.

What is the need for syndication?

In the ever-changing retail landscape where customers are constantly looking for improved experiences, it is essential that the retailers cater to their requirement and give them the experience that they want.

The digital consumer’s experience now goes beyond only an ecommerce website. There are several other touchpoints which have evolved over time like mobile apps, mobile web, internet of things, chatbots etc. And regardless of several touchpoints, they expect same experience and consistency in content, accuracy of data across all these channels.

This consistency in content and thereby the experience need to be supported by robust process workflow and technology to collect, transform and distribute content seamlessly across channels. In short, this is what syndication aims to achieve. Syndication with a headless CMS strategy are almost synonymous, wherein syndication is a concept and headless CMS is the technological tactic to achieve the same.

So, to sum up following are the top advantages of a syndication strategy

  • Faster time to market
  • Consistency in content across various channels
  • Seamless ingestion and distribution of content and product data into retailer’s ecosystem from multiple sources
  • A consistent and holistic product and content experience to end consumers.

What lies in the future?

Although the concept of syndication has been doing the rounds for some years now but many retailers are still in the journey of properly adopting it.

Natural language generation

A bird’s eye view of Natural language generation

Natural language generation is an AI software process of converting structured data into natural language. It has the capability to generate narratives from input data which is of value for humans.

Before delving further into what is Natural language generation (NLG), it is beneficial to know about Natural language processing (NLP) and where does NLG fit in the space.

NLP is something which has been doing the rounds for quite some time in the Artificial Intelligence gamut of things. In fact, NLG and NLU (Natural language understanding) are sub-categories of NLP.

The illustration below followed by a brief definition of the three terms will help explain the concepts better.

What are NLP, NLU & NLG?

Natural language processing is an AI software process of processing language in a human way. Through machine learning models, language is interpreted and used in ways like a human would do in order to get some activities done.

Natural language understanding is a software which interprets structured or unstructured text inputs, voice inputs and do a set of activities of human value based on those inputs. Example, chat bots or voice assistants like Alexa, Siri, Google home.

Natural language generation, a sub-set of NLP is a software process which reads, interprets structured data and converts into natural language which is of value to humans. A simple and widely known NLG adaptation is Gmail’s text predicting email composer.

Image from google

So, while composing an email in Gmail it tries to predict what the next set of words that the user will write and prompts them. It looks quite simple but there is an AI algorithm (NLG based) in place which is doing that. Now, at times it may not be accurate or may not predict at all but it is still a bit better than having to type every single word to compose an email.

Text prediction is just a tip of the iceberg for NLG use cases. Let’s look at some other uses of the technology

What are the uses of Natural language generation?

  • Writing narratives from analytical data : Natural language generation based platforms can automatically create linguistically rich insights from a set of data. Using the NLG capable AI softwares, companies are able to write analytical reports from structured statistical data. This data would normally be read, analyzed , interpreted by humans and then analytical inferences written. That may need them days or even weeks to come up with reports on lets say a company’s quarterly or annual financial results. The same can be done using NLG based softwares at mass and at speed. So the human workforce can focus on more high value tasks.

Check out the following NLG based COVID-19 dashboard from Arria, an NLG service company.

The narratives in the illustration above is NLG driven. It is automatic, scalable and keeps updating itself with change in data.

  • Other uses include generating blogs, newspaper articles, editiorial content, sports update etc. However, commercial acceptance of the same by humans would need more advancement in the technology.

What does the future hold for NLG?

By 2022, 25% of enterprises will use some form of natural language generation technology.

Gartner

NLG is improving fast. From the early 2010 when its usage was limited to spelling checkers, small scale text predictions etc. it is now moving into writing a whole content by itself.

The illustration above provides different NLG models deployed by several compines in the course of past 2-3 years.

You would notice each of the models launched mentions a number. Eg, OpenAI GPT- 110m, Google AI BERT-340m and the latest in the chart which is Turing NLG – 17b.

These numbers indicate the number of parameters the machine learning models have been pre-trained on.

So, in essence higher the number of parameters in the model, the better is the prediction and quality of natural language output. Turing NLG launched its beta version in early 2020 with 17 billion parameters which significantly beat NVIDIA’s 8.3b parameter’s model, the previous best.

But Turing NLG, has recently been outcrowned by Open AI‘s new model, GPT 3. Open AI, which is an AI research and deployment company and funded by Microsoft has been a key contributor in developing and improving NLG models over the years. Its previous 2 launches are GPT 1 and GPT 2.

But GPT 3 comes with a significant upgrade with 175 billion parameters! The best till date.

So, what are these models capable of doing? Here are a few things among many,

These models are trained on stuff that is published in internet. For example, GPT 3 contains trillions of words from the internet. It has every Wiki article, every books published online and millions of contents in the web. It is therefore capable of

  • answering human questions in text
  • translate into different languages
  • providing improvised content

Here’s a blog written entirely by the GPT-3 beta version. Check it out.

Who are the top providers of NLG services in the market?

The enterprise level adoption of the Open AIs and Turing NLGs may take some time. But in the meanwhile following are some providers of NLG services, not in any order of ranking and not an exhaustive list.

Forrester New Wave NLG 2018 review

Conclusion

As our world become more data driven, with quintillion bytes of data generated every day it is a no-brainer that AI and its uses, in our case NLG, will help consume the data quicker and provide output which can be used by humans for their own value.

As NLG evolves further, it will be interesting to see how soon can we have our own bots having effective natural language conversations with us.

What is Search engine optimization

Search Engine Optimization is a method in which websites become visible to users of search engines

SEO or short for Search Engine Optimization is a method in which websites tend to be more relevant and search-friendly in various search engines viz, google, bing, baidu etc. It helps websites and its pages rank better in search engines. Although, it is called search engine optimization but it is fairest to say it to be Google optimization as still 90% of searches are done in Google. The rest of the searches are divided between the other search engines viz, Bing, Yahoo, Baidu and others.

What is a search engine?

The wikipedia definition of Search engine is a “A web search engine or Internet search engine is a software system that is designed to carry out web search (Internet search), which means to search the World Wide Web in a systematic way for particular information specified in a textual web search query”. Pretty boring right! To put it simply, it is a portal through which anyone can search anything in the web using a few keywords. Search engines use these keywords to understand what are the most relevant websites, blogs, articles etc. that are matching or are the closest match to those keywords and then list them down for the user. There are quite a few search engines available in the world. The most famous & popular being Google and some of the others are

Bing: by Microsoft, about 2% of the global searches

Yahoo: a very popular one and an old one as well

Baidu: used in ChinaYandex: owned by Russia

What is web optimization?

Optimizing your website so that it ranks higher in the search engine results is the primary goal of web optimization. In this context, it is necessary to understand what is a keyword and what is a keyword strategy.
A keyword is any word, phrase etc. which an user uses to search for content in a search engine. For example, if you search for “men’s running shoes” in google you may get billions of results all talking about something related to men’s running shoes however, it is the google algorithm which decides which results are the best and the most relevant and ranks them in the SERP (Search Engine Results Page).


Now all the website pages should have an intent or purpose and that ideally should be linked to a keyword. That is part of every SEO strategy. Content is the king! and the more relevant the content is to the user search the more chances are the website will rank higher in SERP. However, linking every page to its intent and keyword is part of white hat SEO techniques.

Page experience – google ranking factor

Page experience is the next big google ranking factor which is being introduced by google in 2021. Read on!

Google came up with an announcement on the 28th of May that they are going to introduce a new search ranking factor which will be based on page experience. So, the better the experience an user has when visiting a webpage, the better are the ranking chances for the website, provided of course the other ranking factors are kept the same.

Page experience is going to add on to the aspects of the Core Web Vitals, which was announced by Chrome early May.

Core Web Vitals and page experience

Core Web Vitals are a group of metrics which are also used to measure page experience. Although the actual page experience metric is deduced from additional factors as well, but page speed, interactivity and layout are part of Core Web Vitals. Following is a visual representation of the different aspects that make CWV

Largest Contentful Paint (LCP): when the largest content of the page is rendered. Simply put, how soon is the largest content (ideally the most important) visible to the visitor of the page. Any time within 2.5 sec is considered good while more than 4 sec is considered poor. A simple example shown below.

FCP is the First Content Paint & LCP is Last Content Paint

First Input Delay (FID): The time between when user first interacted with a page (clicked a button, link etc.) to when the page responded to the interaction. A delay of less than 100 milliseconds is considered good and anything over 400 milliseconds is considered poor. A very tight margin for error!

Cumulative Layout Shift (CLS): this metric measures the visual stability and the amount of unexpected shift of visible page content. A CLS score of less than 0.1 is good and more than 0.25 is considered poor.

Well, that was a summary for Core Web Vitals. Now, what are the factors for page experience? The image below will help explain the various aspects of page experience and where CWV rests in them

Another important change which will be brought in along with page experience update is the relevancy of AMP for ranking mobile pages. Till date AMP enabled mobile pages have an edge over other mobile pages in google search results but with the roll out of the page experience update it will no longer be a ranking factor for mobile pages. So, page experience will solely be the ranking factor for web and mobile pages alike.

When will the page experience roll out?

Google has given a six-month heads up as the roll out will not happen until 2021. Google has provided a bigger than usual wait time for an algorithmic update due to the current COVID-19 situation. So, the businesses have some time to do the housekeeping of their sites.

Is page experience good, bad, ugly?

Some or maybe most of the aspects that define page experience already in some way impacts ranking factors. Page speed, user interactivity are in the picture for quite some time. But it seems google wants to package this with proper defined metrics and drive the importance even more by rolling out a separate update.

In the age where most of the businesses are spending a lot in improving their customer’s online experience, a metric which will drive their site’s visibility in google will only help the businesses focus on the same in a more structured way. At the end of the day it will help improve their customer’s online experience. So, it cannot be bad. But we will wait and see.