I decided to understand the evolution of mobile technology and how it may come to redefine how we discover information or rather how information will discover us. I will be honest when I decided to take the challenge I put before myself I was not sure what I would find or how I would come to fulfill it. But I have to admit this journey has taken me beyond the realm of technology and its application to something deeper, something rooted in down-to-earth biological system.

I started at something banal, looking for apps that exist today. And I can say it was quite unsuccessful, I mean what was I to look at – a series of apps (and there aplenty) and present them as my findings? However, it was the seed, the beginning of a discovery that has led me from the quest to find the next big thing to understanding the basis of creating the next big thing. And that is transformative – to my thinking and to my challenge.

Here is how it all began and it all started coming together, one small piece at a time.

I started at the top – searching to find….? But I found a list of what seemed like a disarray of apps and services.

I then re-looked at my assignment and started with recommendation/discovery apps and then progressed to virtual assistant apps.

Recommendation apps – because it is based on personal interest graphs and not just social interactions

Virtual Assistant apps – because it leverages (or is trying to) natural language interface and syncing all activities

Potentially, all the recommendation apps become the resource – a sort of database and the virtual assistant the delivery mechanism.

To summarize quickly based on my learning so far, the way to tie these together for marketers is to:

1. Collect interest data from recommendation apps (like Foursquare that plans to let brands and merchants access to user data to target them elsewhere)
2. Synthesize and assimilate information on their consumers through email database, website behavior, social interactions.
3. Optimize website to be ready for semantic and natural language search for the virtual assistant app adoption.
4. Target with personalized offers (e.g. user plans to travel from one place to another as on the calendar – instantly sync with an existing app or find the best offer for taxi/car rental options)

But I would like to describe more in detail how I came to the above mentioned.

I looked at some recommendation apps such as Nara, Sosh, Ness 2.0,  and Stamped to see what they captured and how they made decisions for the user (predictive analysis).

For virtual assistant apps, they have been talked about so much; I considered Siri, Sherpa, Osito and Google Now among others.

Together, I think they are attempting to create highly personalized and intelligent answers system.

So what does that mean for not just consumers but also marketers? Does this signify a shift to a more one-to-one recommendation?

Granted the ideas are not altogether new, they have been around for sometime. Especially e-commerce sites have been doing this. Amazon is a great example of the collaborative filtering. Interest based recommendations personalized to my past behavior.

And I think the answer lies in Amazon’s magic and Google’s intelligence with knowledge graph and interest graph.

To reiterate, the recommendation apps/social networks become the resource – a database and the virtual assistant the delivery mechanism. And why do these apps matter? I mean is there not enough data already to provide the interest graph information to marketers?

Yes and no.  To some extent the interest graph information can be found from Facebook, Twitter, YouTube and other social activities. But there is more granularity that these recommendations apps can provide because they are based on the individual’s taste and preference and not just his/her social connections or interactions.

I think there is merit to this approach over using only social recommendations – what my friends like I might like too. The reason I say so and I am still uncovering is that all what my friends like are not necessarily my interests. I think there are more interest-based options on the open web.

More recently, Forrester released their findings on Database of Affinity that proclaimed Google as the winner.

What caught my eye is the fact that Google has more interest signals based on the online activities the users perform including the dark social activities (sharing via email). This together with the collective knowledge of the Internet serves the right answer at the right time. Relevance and interest come together.

For marketers, the implications lie in not only investing in branded experiences but also truly leveraging the technology that these  apps offer.

And more importantly, I think it is no longer just optimizing for keywords on branded sites but also optimizing for answers. It is learning to understand what questions are being asked and how best we can supply with responses. Thinking how Siri or Google Now may eventually bypass links to sites and answer the question – but they will rely on someone to offer those responses.

Enter natural language interface and the theory of Noam Chomsky of how to understand language is learned and processed. It is a little wonder then everyone is after the perfect algorithm. But it goes beyond linguistics, for Artificial Intelligence it is also about motion, gestures, facial recognition and the immediate environment.

Take for instance the Smileage app that is integrated with Google account, knows when you go on a drive, tracks your route, lets you take photos and tag passengers, post it on your behalf. It also pulls in information on the weather, time and events across the web to create a personalized video.

Or have the knowledge that if a consumer who has viewed a recipe on an app, added the ingredients to a shopping list app and her virtual assistant is synced with her activity such as physical grocery shopping to provide this consumer the relevant offer at shelf as she checks of her list.

This presents marketers an opportunity to serve content, targeted ads and offers, based on all these factors.

But it does little if marketers don’t realize the opportunity to becoming experts in their area – optimizing content, tying the data from external sources and meeting the wants of the consumer.

The app unlike other video or gif apps functions on constraints. Only 6 seconds, no way to edit the footage (what you see is what you get), no filters, no audio editing.

These constraints immediately reminded me of Lars Von Trier, the Danish director. Von Trier is well known for imposing constraints or obstructions (in fact he thrives on it). The filmmaker works within these limitations, which inherently breeds new creative thoughts in filmmaking.

Let’s review his film 5 Obstructions.

5 Obstructions is a documentary film in which Lars Von Trier challenges his mentor and friend, a fellow filmmaker Jørgen Leth to remake his movie The Perfect Human five times with a different obstruction each time.

First obstruction, required Leth to create the film in Cuba with no set and only using 12 frames to answer all the questions in the original movie.

Second obstruction was to make the movie in the ‘worst place’ in the world, but without showing the place onscreen. The meal from the original film was to be included but not the woman.

The third obstruction since the second attempt was deemed a failure and Von Trier asked Leth to either re-shoot in the same location or to to remake the film as he chooses.

The fourth required Leth to make the film as an animated movie

Finally, the last and my favorite, Vons Trier has made the movie and Leth must add the voice over  in his style but using the script that Vons Trier has also written.

There are lessons to be learned from this. Intelligent constraints can indeed help fuel creativity and innovation. This HBR post drives home the point of how it is important in business to place constraints to find creative solutions.

5 Obstructions teaches us the lesson that we should to take what we know and reinterpret it in different ways.

“Think different”  by Apple is one such example. Apple reinterpreted personal computers. Using constraints of design and user experience the company introduced the world to the beauty of computers.

And in today’s content marketing era, real-time trends are the constraints. Oreo is one such brand that understands the real-time constraint to create content. Oreo’s Daily Twist campaign captures it beautifully and more recently, its responsive marketing during the blackout at the Super Bowl will be mentioned over and over again for sometime to come.

Another way to look at this would be crowd sourcing content by placing restrictions to see how consumers can reinterpret the brand. Not only does this provide marketers content it also shows them how consumers tend to think about the brand. It can lend more insight and understanding of the community.

It is important to place smart limitations to spur the creative thinking. Do you agree?

Now and again, I am entrusted with the task of creating a social media plan for brands that I work on.

There is a lot of information and so much to analyze that sometimes it can be daunting to put together an effective communication plan.

This morning, I came across this very interesting article in the HBR blog by author Scott Edinger. And I wanted to capture my own interpretation of the same especially, from a brand’s perspective. So here is my take on the three elements for effective communication to craft a social media plan:

Ethos – Credibility or in the words of consumer, why do I care?

Pathos – Empathy or from a consumer’s viewpoint, why do I share?

Logos – Reasoning or from a consumer standpoint, what’s in it for me?

At the start of any brand communication plan, it is important to understand why a consumer would choose to believe in the brand.  This is in part defined in the positioning of a brand. A positioning statement will define the target, the solution and the product with a compelling benefit.

The rational and emotional aspects of the positioning concept also help craft the social media plan.

Here is a template of a social media plan that I used for a brand that I work on.

Business Objective : Build a community loyal advocates.

Community Objective: Recognize those who are passionate about the brand.

Target – Adults (males and females) between the ages 21-35, who are tech- savvy, entrepreneurial and trendsetters.

Why do I care?

Ethos – or credibility. The reason to believe for this brand lay in the fact that the product was crafted using the best locally available ingredients and that the family that created the product was still involved in the production. There is more trust when there is transparency and with a family business that resonates more.

Why do I share?

Pathos – or empathy. A consumer would want to share information/content about this brand and the product because it is unique and hand-crafted. It satisfies consumers’ need to belong and curate their identities by introducing something new to their friends and connections.

What’s in it for me?

Logos – or reasoning. A consumer needs to have key benefits for following a brand and talking about it. This can be tangible or intangible.

In this case, it was the ability to connect personally with the other like minded individuals who were equally passionate about the brand and communicate with the family that owned the brand.

Brand Purpose – The reason why the brand exists? In this example, the brand’s purpose was to bring unrivalled experiences from the woods. It elevates the brand from the realm of tangible/rational benefit to personifying it as a curator of experiences.

Key Strategic Idea: In this case, I built the following statement to help define creative approach to the social media plan.

“Facilitate a community that will enable consumers to discover unique ideas to create memorable experiences.”

Brand Voice: It builds on the personality based on the brand purpose. The brand voice and tone help in developing content for ongoing engagement.

Discussion Topics or Content Buckets: These are key content categories that help with copy and image strategy.

Analytics: To rinse and repeat based on success

The key to a good social media plan are the three pillars of communication. What do you think, what is your approach to crafting an effective social media plan?

Graph Search
What is it? It allows users to query on Facebook to find results related to friends and connections. It is NOT web search like Google.

“Graph search is the ability to provide you answers and not links that MAY have the answers.” Mark Zuckerberg

But the semantic nature of this search functionality does have some merit. It is still not where I would go to search for information. Something Zuckerberg recognizes and therefore includes a web search for those needs. The problem is I and the majority of users prefer Google’s search.

But that said this does present some key opportunities to brands who off late have been begrudgingly accepting the decline in reach and engagement.

While the current feature will not include ads, in the very foreseeable future I see that as a very distinct possibility.

Before I move on to the actual features if the search, I would like to prefix that brands on Facebook more than ever need to optimize content for Graph Search. Users can search for content and interests that present a tremendous opportunity to organically and not so organically be in the results.

Brands will need to employ a three pronged strategy.

Build

Continue to build an audience (results are based on connections a user has, so it goes to reason a brand with more fans will have greater likelihood of being featured in the results)

Optimize

Optimize content that could be searched (e.g photos of cats, photos of popular places or iconic monuments. Using such images that people are likely to search for increases the chances of discovery in the results)

Promote

Promote the posts (remember, a key piece of this search is whether content searched for is liked by the user’s network. More popular the content the better the chances of being shown)

At this moment, the search information will not be made available to brands or businesses. But it is only a matter of time. In fact, this may well be the first step toward a browser. Ads too will be rolled out in all probability. Ads such as sponsored stories can only get better with this.

So here we go with the actual details of the Graph Search.

There are 4 types of Graph Search.

1) People Search
This is similar to the “Find Friend” toolbar that let users find friends by different cities, schools, employers etc.

It has now been relaunched as part of People Search

Users can search for Friends in NYC and the list will show them all friends in their social network who live in NYC.

Users can refine the search by gender, age, location, employer,relationship, hometown, current city, school, friendship and name. This will bring up result that match all the criteria.

Companies can use this to recruit employees. Friends of current employees is a good place to start with recruiting.

Companies and HR can use this to see who are the current employees whose friends are connected to another company. E.g. Employees of NASA Ames Research who are friends with Facebook employees.

How can brands use this?
A brand looking to hire prospective employees can leverage this to show a sponsored story when a user searches for a friend by an employer. It is important to encourage employees to fill out the information in their profile so they will show up in the results. This is directed at LinkedIn, to some extent.

2) Photos Search
Users can search for photos of friends e.g. Photos of friends hiking or Photos of friends who have been to Yosemite Park or Photos I like etc.

Results are based on two factors, proximity and the strength of the friendship .
Proximity – the connection level (first, second – friend of a friend etc.)
Strength of friendship – the closeness of the user and the friend(s) (in terms of interaction)
Top results are based on the two factors. And thus why continuing to build fan base and optimizing content supported by paid media is important.

Only content that has been shared with the user will show up in the results I.e. content that is private or limited to only certain friends will not be shown in the search results. Facebook is taking the privacy issue seriously.

What does this mean for brands?
Fanbase, content optimization and engagement will have to be part of Facebook posts.

When the ads feature for search does roll out, brands will be able to pay for ads to be shown with content that is in the results, similar to Google Adwords or even the existing Facebook Sponsored Stories. E.g. Photos of cats my friend likes could display a sponsored story with cat photo from a cat litter brand that has been liked by friends.

Ad would contain an endorsement of the brand by the users’ friends. It uses peer influence as the tactic to generate interest and action.

3) Interest Search
‘Movies friends like’ will show the movies that the users’ closest friends like. There is an option for extended search that will show related content that friends enjoy to help introduce users to join new things.

What does this mean for brands?
Brand will have to invest in continuing to build a fan base because search results for a user search such as ‘cat litter my friend likes’ would depend on whether there are enough friends of the user who have liked the brand or the product.

Search results will also drive more likes possibly as users discover brands that friends have liked.

4) Places Search
This allows users to search by location e.g. Mexican restaurants in Miami that friends like.

The results displayed will be of those Mexican restaurants that are liked by friends.

FB search does have an edge over Google’s Search Plus Your World ( a product Google rolled out in Jan 2012), which also leverages the social graph in the search results. This is because the people a user may have circled (or friended in FB terms) on Google Plus may not be their close friends. Most connections on Facebook tend to close relationships and acquaintances in the real world.

The con is that Facebook assumes that a user’s friends will have liked the restaurant page but many local businesses don’t have a FB presence and this could incentivize them to build a presence so they can be displayed in the search results organically. Although, there will be an investment to build the fan base in order to leverage the connections or friends of fans.

What does this mean for brands?
The closest I can think of for this, especially for a consumer brand that is not a retail store, is ‘offers my friends like in Miami.’ This should then show what offers are locally available. Again, for this to happen a brand will need to create and promote an offer. While this is not available in the current version a future roll out cannot be ruled out!

This is all about local search and brands maybe able to target ads in the future, as mentioned, based not only on location but also friends’ affinity or likes. E.g. ‘search for nasal allergy relief friends like’ will likely show brands that friends have endorsed.

For those brands that don’t have a FB presence and are unwilling to make the required investment will have to go down the advertising lane but without the friend endorsement a la sponsored stories, it is unlikely that the ad will have as good an impact as one that does. It will however add to brand awareness and can be used in the long-term to see if that type of Facebook ad drives site traffic and conversion to make an informed decision to launch a dedicated Facebook page.

Please note: similar correlation can still be done between Facebook marketplace ads and the conversion/traffic on site metrics.

Last but not the least is the Web Search in partnership with Bing. When graph search does not always produce the desired result, Bing’s partnership will provide relevant results via the web. So it is still important for brands to invest in SEO and SEM to be displayed on Bing as with Google.

Graph search is rolled out to some of the users today and will be rolled out eventually to all users. You can take a look here: https://www.facebook.com/about/graphsearch

In Sept, 2012 Facebook made a change to the EdgeRank algorithm, which impacted the reach for most Facebook pages. Different third party research show that reach indeed declined. The degree to which it decreased varied across brands. Socialbakers along with the digital agency We Are Social found that reach went down by 40%. In fact, when delving into the data of pages that my agency manages I saw a similar trend. Pages of different fan size saw drops that varied from 20-50%.

However, what is crucial here is that while organic reach was affected viral reach saw steeper decrease. I saw this in the data for the pages I analyzed. I looked at both the page and post level data. In fact, on one of the pages, which has maintained a PTAT of 3-4% before and after the algorithm shift, the average post reach almost halved.

I point this out because this page has had consistent engagement on posts with an average virality of 4-5%. After the algorithm shift, the virality has increased but only because reach has decreased. So the post level PTAT has remained more or less steady indicating that we are reaching fans and non-fans who are engaging with the content regularly or have a higher affinity for the brand. And this is what We are Social also concluded from their study with Socialbakers. But the lowered viral reach is troubling. The whole point that engaging fans leads to their friends, which can influence them is setback. This post by Jon Loomer discusses the viral reach decrease. And this is where paid media can come to be useful – reaching less active fans and friends of fans who otherwise may not be engaging as much with the page content.

But interestingly, one of the pages had a media buy for fan acquisition around the time the algorithm change happened. Here I noticed that the organic reach on the posts increased with the paid media. Just to clarify there were no promoted posts. Viral reach remained unaffected. So the posts were actually entering news feeds of these newly acquired fans. The paid media also resulted in an substantial increase in the average lifetime post level PTAT.

All this also seems to fit with what Napkin Labs reported, only 6% of fans engage on a regular basis. These are the superfans, who need to be nurtured to influence their social network. Napkin Labs study found that top 10 most engaged superfans received 2.3 times more likes and 1.8 times more comments than less active fans. It is all geared towards quality over quantity. I have found that it is true, using Socialbakers I am able to identify the top ten engaged users and turns out they are the ones who regularly comment on the posts.

But reach is still important, per this article and Facebook’s study. Reach drives revenue – this study conducted with Datalogix shows that campaigns optimized for reach can get 70% higher ROI. Facebook is trying to convince marketers to look at the platform as a mass medium. Exposure to quality brand messages as sponsored stories can help drive offline sales.

What does this mean?

Brands more than ever need to focus on not just building huge fan bases but also sustain the fans. While size of a page can give it more credibility with a new or an existing consumer, the emphasis is on nurturing the relationship with active fans.

Engagement is driven by content and superfans. Identify these superfans and reward them with recognition and exclusive information.

Fans and users are more empowered and can and will remove brand posts from their feeds that are spammy. More than ever a brand needs to pay close attention to what resonates with the fans.

Paid media is important to keep the brand top of mind and also reach friends of fans. Word of mouth and peer recommendation influences purchase decisions. Facebook content strategy should consider paid and earned media. Facebook released a white paper that explains how a brand can reach 5x more people with paid ads.

Brands cannot expect to reach simply with organic posts. What do you think? Will this change your strategy?

So on a cold winter day while I was waiting for my train, my mind furiously trying to predict which subway would show up first, the E or the F, I had a thought. An app that could help me and the other commuters, an app that could take the guessing out of the rush hour game.

My travel itinerary includes switching subways to reach Grand Central where I hop aboard the Metro-North to arrive in Connecticut. I am constantly trying to make the best guess on which subway will be faster, whether one queue will move faster than the other. See, the difference in saving a few seconds can mean whether I get on the Connecticut bound train or not.

My trip is all of two hours one way, so it can be very frustrating when I reach the station to only see the Metro-North pull away. Not only that when I take the E I have to transfer to the 6 at 53rd and Lex. And inevitably there is always a queue I have to navigate to get to the upper level. I have to make the bet the line that I pick will move quicker than the other.

My commuting experience and the numerous articles I have been reading on big data sparked the idea that cold winter day. An app that would analyze all the happenings on the track, on the platform, up the escalator, down the stairs, every movement, each foot step carefully documented to predict what my commute should be. I was fascinated by the Ted Talks video below, which inspired me to believe that such an app could be possible – one that could even out the traffic while reducing the congestion. It is an app that could finally take the “What if” out of the rush hour equation.

Brands on Google+ can create communities, this is unlike Facebook where groups can only be created using the personal profile of  the admin, which can severely limit the branded presence in groups. “Administrators have to use personal Facebook profiles when posting to the Group; they cannot comment or post as the brand itself.” Read more

But more importantly, communities has value for marketers. Below is a list of benefits of Google+ Communities.

Ten Benefits of Google+ Communities for Marketers:

1. Creating a community allows brands to identify consumers who have a specific shared interest.

2. Identifying popular topics and using it for SEO – brands can create content around  topics important to community, which could improve SEO.

Community members can post to a specific topic. If a topic has more posts it could be inferred that topic is more popular.

Note: Google+ does not show the number of entries for each topic, so at this moment it seems this may require more manual sifting. However, I cannot imagine there won’t be some analytics tool that would be able to help with that going forward, or Google may itself roll out this feature.

3. Gaining more followers, communities don’t require a user to also follow the brand’s Google+ page,  however by joining a community a user maybe more likely to also follow the brand’s Google+ page.

The more the Google+ followers a page has the more likely the brand will be able to capitalize on SEO. This is because when a brand publishes the content on its page the more followers it reaches the greater the chances of the content being shared and  discovered by other consumers. Read more

4. Making the brand more social, communities allow an easy and accessible platform for like-minded consumers to connect with each other.

For the Gen Y consumers, community and relationships are of great importance. A brand that facilitates that kinship is more likely to develop loyal customers in the long term.

5. Allowing consumers a chance to be highlighted, communities offer the ability for consumers to share content with other members.

Posts by community members can be selectively shared on the brand’s page, making the connection between the brand and the consumer more intimate.

6. Gathering resource, content shared by both the brand and the community members make the community a go to place for expert information.

It helps cement the brand’s position as an expert in the category. It makes the brand more reliable and trustworthy.

7. Discovering new content, brands can find unique content that community members publish.

Communities offer crowd-sourced ideas that brands can leverage.

8. Using communities to get feedback on new product ideas.

Unlike a page where fresh content replaces older content fairly quickly, communities allow for topics that consumers can visit anytime and interact with the respective posts. (Hat tip to +Jerry Daykin)

9. Identifying the most active community members, these consumers who post regularly and respond to other community members are brand advocates and potential influencers.

10. Seeing new or changing trends within the community, as members gravitate to a specific topic or a new one, brands can have a better understanding and insight.

These are some of my first impressions on the benefits of Google Plus Communities for marketers. What do you think?

Even before the web profiles, Instagram’s API did provide brands a unique opportunity to integrate Instagram photos within their site or blog. For quite some time now, search engines have been using social signals as a weight to indicate trustworthiness and relevance. While one can argue that social signals can be manipulated as well, for this post I will consider the quality shares – ones that are by real consumers who actually care about the product. And so that brings me back to the question how can a brand boost SEO with Instagram?

For one, just the web profile could in the future become searchable and indexed. I mean how often has one typed in a brand and received the Facebook page also within the top results? Plenty, right! See below, I did a search for Tide and voila there it was the Facebook page. So, I do think at some point Instagram profiles will also begin to show up in search results.

Now how does that affect SEO you may ask. And I have a theory, most consumers are  less likely to visit a site on a regular basis – aside from those of retailers. So, getting consumers to connect with the brand at different touchpoints enables brands to communicate.  And once a consumer has opted in to receive messages from a brand it is a lot easier to direct them to the site for an offer or a promotion. Yes, the impact is a little indirect. But also consider the organic listing as free advertising another impression if you will. If a brand page shows up in search result that is always a benefit.

But more importantly images are also returned in the results. As you can see above the results also include images that match my search tide. Leveraging Instagram photos for image optimization can be yet another way to help with organic listing.

Free People has integrated Instagram within the denims section of the site. Certain designs are promoted with hashtags and  images posted on Instagram with the specified hahstag are incorporated within the site. These images can be optimized with captions and Alt tags. This makes them more searchable and therefore are more likely to show up in results for the keywords associated with them. Additionally, there is always the benefit from user-generated images being shared on other social networks.

Next, I think Instagram’s Photo Map feature could also become integrated in the search results. A search for Nike brought in stores nearby and a map with the locations. It would be great to have the brand images show up within the map for these retail locations. Images are more engaging and likely to persuade the consumer to the next step in the path to purchase. I think this is especially important for mobile search. Below is a screenshot of the map that showed up in the results when I searched for Nike. While such a feature may not be incorporated at least for now, since Google and Facebook are at loggerheads, it is possible nonetheless.

These are some of the reasons why I think Instagram could help with SEO or organic listing and why brands should take a serious look at in investing on the platform. What do you think?

Ingredients
1 medium long squash chopped
1 medium red onion diced
2 medium plum tomatoes chopped
1 tsp cumin seeds
1 tsp coriander powder
1 tsp red chilly powder
1/2 tsp turmeric powder
1 tsp garam masala
1-2 small green chillies chopped
3 cloves garlic chopped finely
1 inch ginger chopped finely
Salt to taste
2 cups of water for cooking
Chopped cilantro for garnish

Preparation Time: 5-7 min
Cooking Time: 30-40 mins

Instructions

Peel the squash, cut longitudinally into quarters. I prefer to remove the seeds as well at thus point. I do this by scraping the top section. Chop the rest into cubes. Keep them small so the squash cooks quickly.

Add a tbsp of oil to a large pan or wok and let it heat, don’t let it smoke. To this add cumin seeds, sauté a bit and then add diced onions. Cook till onions turn pink and translucent and then add onion and garlic. Stir for a few seconds and then add chopped to tomatoes and green chillies.

Let the tomatoes get soft and then add the spices i.e. turmeric, red chilly powder, coriander powder and garam masala. Mix well and let the tomato-onion paste cook till oil separates.

Now add the chopped squash and water. Also at thus time add salt.

Let it come to a boil, cover and cook on a medium flame. Check regularly, stir to keep it sticking to the bottom of the pan.

Check if the squash is cooked after half-hour. Use a fork to pierce and judge the softness. If more water is required, add some as needed. Cook till squash is done. Garnish with coriander. Serve hot with rotis or paranthas.