January 18th, 2012
The BlitzLocal team attended the Facebook Mobile Hack event in NYC on January 18th and took some notes on the newest updates to the mobile platform.
Read here to see what the Facebook engineers had to say, and hear what has worked for several case studies.
Intro: Facebook Platform for Mobile
How has the web changed since the early days?
- Photos!
- Facebook tagging (social)
- Games
- Not focused on rendering the best graphics, but social aspect
- 200 million people playing games on Facebook Platform
- The rise of mobile
- Across all levels of mobile devices
- 350M users of Facebook mobile
- Twice as engaged as desktop version
- Social + Mobile
- Friends, Newsfeed, search, notifications, requests, bookmarks
- Photography, websites, music, communication, games, books
Web Apps (HTML5) Development: Matt Kelly & Vikas Gupta
- How to facilitate sharing with friends
- Problematic when sharing over different types of devices
- Apps don’t exist across platforms
- Hard to share content between them even if apps exist
- How can Facebook fix this?
- The social channels
- Requests: user to user direct messaging
- Make sure they are super fast! Almost real-time
- News Feed: one to many sharing with friends
- Posting, view on wall, view on news feed
- Open Graph: Lightweight, seamless sharing
- Wherever your app lives or works, Facebook distributes properly so there are no holes.
- Bookmarks
- m.facebook.com
- Gives users a specific spot to get back to your app
- Mechanism for engagement
- Native facebook for iPhone
- Payments
- Use credits to monetize apps
- Build social from the ground up
- Search, login auth, directed to app
- Social apps work everywhere
- iPhone, laptop, android, ipad, android tablet
- HTML5 works across all platforms
Native App + Open Graph: Aryeh Selekman & Christine Abernathy:
- iOS and Android + Facebook platform
- Open Source Native SDKs
- Developer app settings
- Fields to configure iOS and Android apps
- Linking and app distribution works appropriately if these fields are filled out correctly
- Single Sign On (SSO) login without typing
- Login with facebook button
- Understanding Native Distribution
- If an iOS or app exists, all requests/News Feed Stories/Timeline stories will link directly to native app (or apple store if not installed)
- On android, requests/News Feed Stories/Timeline stories wil only direct to your mobile web app
- 5 Best Practices
- 1. Build a mobile web app
- HTML5: distribution on m.facebook.com on all webkit enabled touch browsers
- native wrappers
- take advantage of SSO capabilities
- 2. Implement SSO
- 3. Implement requests for app discoverability
- Rate now, invite friends, send requests
- 4. Leverage existing friend graph
- Promote activity and interaction
- Re-engagement
- Push notification through native channels to let you know your friends joined
- 5. Get your app on timeline
- Open Graph and Mobile Apps
- One API – distribution to ticker, Timeline, newsfeed
- Allows you to define what people do in your application
- Reading, listening, watching, etc.
- 4 Steps to get started using open graph
- 1. Define your actions and objects
- 2. Design your Timeline aggregations
- Pictures, maps, represent actions and data that people send through
- 3. Markup and expose your objects
- Everything represented by underlying URL
- Where do objects live
- 4. Publish actions
Native Distribution for Mobile Apps
Case Study: PhoneGap
- Pattern
- Embed a chromeless browser in a native app
- Create a bridge between the browser and the native code providing access to native APIs
- Write a web app
- Package the web app with the native code and deploy to devices
- Write once debug everywhere
- Take note
- HTML, JS, CSS included in an app package
- HTML loaded on file:// URI scheme, no cross domain request restrictions
- Engineering wise, approach is simple to extend to new platforms
- Support Platforms
- iOS, Android, BB, webOS, Symbian, Windows Phone (mango), Samsung Bada
- Mobile first!
- HTML5: write native code easily, scale like an app so width is design width – no pinch zoom etc
- CSS3: webkit transformations
- The future
- Tooling
- WebGL
- Facebook, Linkedin, Walmart use case
- Continue polyfilling HTML5
Case Study: Washington Post Social Reader
- Coding for the futures
- Everything you write effects possible futures
- Architect for the foreseeable futures
- Short term futures
- Building mobile second
- Use mobile to rethink boundaries
- Roll mobile learning back into the webapp
Case Study: Thuzi
- Hospitality app
- Social by design
- Send invites, RSVPs
- Share great offers with my friends
- Provide reviews of the experience
- Capture the moment for a special occasion (Timeline)
- Local by design
- Dining is a local experience
- Find a local restaurant
- Invite friends, redeem offers, find out what’s happening
- Native by design
- Want to ensure you have access to the newest native APIs
- Want to have the fastest app possible
- Want to guarantee formatting correctness
- More choices for monetization – iAds, etc
- Many existing open source libraries and blog posts and tutorials
- Are not dependent upon plugins or other 3rd party series for push notifications
Tags: events, Facebook, Facebook Mobile Platform, Open Graph, Social media
Posted in Facebook, Learning Center | No Comments »
January 3rd, 2012
Didn’t get a chance to check out the Trada Webinar on making Facebook ads work for you? No worries! Here at BlitzLocal, we watched and learned, and took some notes for you. They started with an educational slideshow about the different types of Facebook ads, and finished by answering questions from the chat window. Here are our notes, and some additional article links for extra information.
The Bottom Line: Facebook Ads are the most versatile, targeted way to advertise online, and they have incredible reach.
Challenges with Facebook Advertising
- Banner Blindness
- You have to hire a graphic artist to make all the ad creatives
Facebook Glossary:
- Connections: The number of individuals who liked your Facebook page, RSVP’d to your event, or installed your app within 24 hours of seeing your ad or sponsored story. Basically, a connection is a conversion.
- Unique Reach: The number of individual people who saw your sponsored stories or ads.
- Social Reach: The number of people who saw your sponsored stories or ads because their friends liked your page, RSVP’d to your event, or used your app.
- Frequency: The average number of times each person saw your campaign’s sponsored story or ad. This is helpful for measuring ad fatigue.
Basic Ad Formats
- Basic Ads
- “Like” ad: Links to tab on Facebook page
- Event ad: Links to event
- Application ad: Links to application
- Standard ad: Links to specific URL
- Sponsored Stories
- Page “Like” Story: Mary-Jane likes your page, Page “like” Story lets Mary-Jane’s friends know about the like
- Page Post Story: You published a post to your page’s fans. Page Post Story allows this post to show up in fans’ news feeds
- Pages Post “Like” Story; Alex liked one of your page posts in the last 7 days. Page Post “like” Story lets Alex’s friends know about the post like
- App Used and Game Played Story: Lauren played/used your game or app. App Used/Game Played Story tells her friends about this action
- App Share Story: Hayes shared a story from your app in the last 7 Days. App Share Story lets Hayes’ friends know about the share
- Check-in Story: Lisa checked in or claimed a deal using Facebook Places. Check-in Story lets Lisa’s friends know about it.
- Domain Story: Mike liked/shared content from your website or pasted a link from your site to his wall. Domain Story lets Mike’s friends know about this action.
The complexity and potential of targeting on Facebook
- There are so many ways of targeting that it can be confusing
- Age, likes, interests, birthday, apps, education, timeline content, friends, event RSVPs, fans
- When choosing targets, focus on two things:
- Narrowing your audience
- Demo and Geo Targeting
- Geography: Country, State, Province, City or Zip targeting
- Demographics: Gender, Age, Birthday, Relationship Status, Language
- Workplace and Education Targeting
- Workplace, Education, Preferred Language
- Likes and Interests Targeting
- Favorite TV Shows, Movies, Books, Music, Hobbies, Religion, Political Views
- Thinking outside the box – Let’s say you want to sell golf clubs
- Nick (who plays golf) is an obvious target
- Barbara doesn’t like golf – but she likes the Palm Beach Country Club
- Chaz doesn’t have the word golf anywhere on his profile, but he plays golf for a living: he’s a sales guy!
- Try the obvious targets, but Facebook’s best advertisers use non-linear thinking to target ads.
Campaign Organization Tips
- An “Account” in Facebook is similar to a “Campaign” in Paid Search
- A “Campaign” in Facebook is similar to “Ad Group” in Paid Search
- Warning: Don’t create campaigns with many different targets and ads. Keep your campaigns small.
- Do not put all segmented target groups in one campaign – as your ads are competing within the campaign.
- Prevent ad fatigue, or banner blindness by changing ads frequently
- As soon as CTR trends down, submit new content!
Q&A:
What is a good CTR?
What images work best?
- Logos traditionally don’t work well
- Images of people are effective
- Format design keeping small size in account
What is a preferred or optimal frequency?
- Keep it small!
- 6,7,8 is bad
Any issues with click fraud?
- There is a department at Facebook dedicated to click fraud with a sophisticated monitoring team.
- More difficult to produce in Facebook than in search where you can just search to make a specific ad appear.
How will timeline affect ads/business pages?
Fans much more likely to take some type of action than non-fans.
Targeting workplaces is a great tool for B2B marketing.
What sort of company uses page post story? How breaking should the news be? What types of posts work best?
- Entertainment, news businesses using these most effectively
- Have emotional connection so people will click
Is there advertising on Facebook mobile?
- In testing, not released to general public yet.
Notes taken by BlitzLocal Analyst Matt Prater
Graphics obtained from: http://www.slideshare.net/TradaPaidSearch/facebook-ads-you-can-make-them-work and www.facebook.com.
Tags: advertising, facebook ads, facebook ads platform, facebook advertising, Pay Per Click, PPC, Trada, webinar
Posted in Facebook, Learning Center | No Comments »
June 13th, 2011
Sentiment analysis, though not a recent term, is nevertheless generating a lot of new buzz within the online business world, due in part to the proliferation of social networks that are generating huge streams of user chatter about companies, products, and services. Social media pundits have drilled businesses with the message that their customers are talking about them, whether they like it or not, and whether they are participating in the conversation or not. As these businesses begin to accept this fact, they realize that they need to do several things if they want to take part in those conversations:
- Locate the conversations
- Gauge the overall sentiment of those conversations
- Participate in the conversations
- Work towards increasing positive sentiment
Tools to help businesses locate the conversations that are taking place around their brands, products, and services have proliferated over the years, so achieving #1 is generally a piece of cake.
Participating in the conversations simply requires good old-fashioned customer service, as does increasing positive sentiment, although a fair amount of training may be required to teach employees how to navigate different social networks, and help them understand just how public their interactions really are. Despite the extra training, numbers 3 and 4 are still hands-on, human-touch activities that that businesses already understand how to handle, and they generally don’t handle being automated very well. Customers would rather have no interaction with a company, than to have nothing more than automated robots attempting to solve their problems.
So if number 1 is easy to solve with automated tools, and numbers 3 and 4 are handled primarily with human-touch activities, what about the second need – gauging the overall sentiment? This is an area that into that fuzzy, not-quite-sure-how-to-handle area. People and businesses are trying to automate this process because applying a sentiment grade to huge datasets is beyond the capacity of manual human power. At the same time, trusting a machine to correctly analyze sentiment is fraught with problems.
Let’s first consider the ways that a machine might analyze sentiment.
Methodology 1: analyze human-scored data, in which humans first analyze and rate sentiment of each piece of data, and the machine then analyzes those ratings to produce reports and aggregated summaries.
Methodology 2: the machine applies a human-scored dictionary of phrases to incoming data, then analyze the data to produce reports and aggregated summaries.
Both methodologies require manual human action first, and neither is a perfect solution. Either humans must score data, as in methodology 1, which is not scalable, or humans must score dictionary phrases, as in methodology 2, which is much less accurate because the phrases cannot take things like mixed messages or sarcasm into consideration. “I love my ugly shoes”, has both positive and negative dictionary words (love and ugly). How should a machine evaluate that? Or “Wow, that’s the most awesome mullet I’ve ever seen!” might sound positive to a machine, based on words such as “wow” and “awesome”, but that algorithm will miss the fact that the user was being sarcastic.
For now, the best solution seems to be the human/machine/human process. Humans must first create phrase dictionaries (or use ones created by others), possibly run actual data through a group of humans who score the data as the basis for which machine learning can rely on, then let the machine classify new, incoming data using the human-scoring process as a way to improve the machine algorithm. The entire process should be repeated as often as deemed necessary. Eventually, the classifier should improve, but humans will always be needed to oversee the results.
Clearly, this process is not merely a matter of humans vs. machines. Both are required to properly handle sentiment analysis.
“It’s hard enough to determine sentiment via email– think of when you may have misinterpreted tone in your own personal communications. Now try to determine sentiment when you have only 140 characters or when a third of these interactions come through a mobile device. Irony, sarcasm, and humor are hard to convey via a 4 inch touch screen. Thus, automated sentiment analysis is problematic, at best, but can at least give you a general sense of how your customers feel. Then manual checking of sentiment can sit on top of your automated monitoring.”
– Dennis Yu, CEO of BlitzLocal
So how should companies gauge sentiment if automated tools aren’t sophisticated enough to pick up on sarcasm, for instance, but humans aren’t scalable to meet the growing flow of data?
There may not be a “correct answer” to this question, but I believe a smart strategy is to choose the best tool to meet your particular needs, and then use intelligent, business and marketing savvy humans to “analyze the analysis”. This human-based analysis of the analysis will ensure that the data reports are not only correct, but are also actionable. After all, knowing that 93 customers are unhappy is a meaningless metric. Understanding which customers are unhappy, and why they are unhappy, leads to being able to take the necessary action to accomplish goals 3 and 4 – participating in conversations and improving sentiment.
Let’s take a look at a few ways that sentiment analysis an be turned into direct action:
Customer service: find specific complaints and negative sentiment, so those users feelings and issues can be addressed, before they cause widespread brand damage.
Promos and offers: determine how a new promotion is doing, enabling quick decisions on what to promote further, and what to discard. Sometimes, a promotion backfires, and should be pulled (and perhaps explained) before negative sentiment explodes.
Understand conversion problems: make note of potential blocks within the conversion funnel that is creating negative sentiment, thereby causing potential sales to fall off the cliff.
As I mentioned early on, there are many tools to help locate and monitor the conversations taking place about your business. These tools range from the simple Google Alerts tool to the more robust and complex Radian6, with most small businesses choosing a tools that fall somewhere in the middle, such as socialmention or Trackur.
Some tools that are either entirely focused upon sentiment analysis, or include sentiment analysis as one feature of a broader social monitoring tool, include:
Which tool is chosen will depend heavily upon company needs and budget. Regardless of the tool used, however, the question businesses need to ask is, “Do we have the manpower needed to properly assess the results, and then take action to enhance or improve sentiment?”
A smart marketing team, in conjunction with a customer service team, should do “an analysis of the analysis” to determine the next steps to take. Understanding the causes of sentiment and making smart marketing and customer service decisions based upon that understanding requires people who are trained and highly skilled in these areas. These people may be in-house, or may be a part of an outsourced social media marketing agency such as BlitzLocal provides, but what is clear is that while machines may be necessary to process large datasets to calculate sentiment, humans are a crucial factor in not only refining the analysis, but in taking necessary actions based upon that analysis.
Tags: sentiment analysis, Social media
Posted in Learning Center, SEO and Marketing | No Comments »
May 26th, 2011
When I googled “social media vs. seo” (in quotes) this morning, 12,400 results were returned. Restricting the same search to only the past year, over 800 results are returned. Obviously, many people are pitting the two against one another, and have been for some time.
There’s little sense, in my mind, in postulating whether social media is better than SEO or SEO is better than social media. Making these two marketing channels adversaries is akin to pitting shirts against pants/shorts. For those people who don’t wear dresses or jumpsuits, they’ll likely be wearing a shirt and a pair of pants. Asking them which is better will probably only result in having them give you a look that implies they think you’ve lost your mind.
Cross-Channel Marketing Rules
When promoting anything, whether it’s a business, a website, a product, or a brand, promotion works best when it covers a broad spectrum of marketing channels. Television ad campaigns work best when they are reinforced by radio campaigns, magazine campaigns, newspaper campaigns, email campaigns, etc. Cross-channel marketing succeeds because the layers of reinforcing messages work together to create a connection with the target audience.
Waves of Data Engulf Users
People today are absorbing external stimuli in huge waves of data, much like a whale ingests enormous swarms of krill. It is unlikely that a whale would be able to pick out the one particular krill that tasted peculiar amongst the millions he ingested. Likewise, the data we absorb each day is so large and diverse, that recognizing and remembering one marketing message amongst the continuous data flow is rather low. It’s much easier for a lone marketing message to be lost in the flow than it would be if it were repeatedly reinforced by entering the data stream in various ways, from various places, and at various times.
Be Where They Are
The goal should always be to interact with users where they are. If they are searching via a search engine, then an SEO campaign can help reach them during that process. Of course, a PPC campaign also works hand-in-hand with SEO to cover both the organic and paid listings that are returned in the search results for the queries relevant to your brand, service, or product.
Users are spending a considerable amount of time on social networks, so a social media campaign is needed to reach them there. Like the concept of using both SEO and PPC campaigns to cover organic and paid search listings, social media campaigns will include organic interactions such as tweets, Facebook Likes and wall posts, etc., in addition to paid interactions such as promoted tweets and Facebook ad campaigns.
Value In Each Piece Reinforces Overall Value
Each channel will reach users at a different point in the marketing funnel, so the value should be calculated and compared in relation to other marketing efforts aimed at that same audience. In most cases, a search user is likely to be in one part of the marketing funnel, and a Facebook user is in another (or may even be outside of the funnel altogether). Pitting the value of an SEO campaign aimed at the search user against the value of a social media campaign aimed at the Facebook user will only result in a confusing matchup. There is value in both, with each being one important part of an entire cross-channel promotion.
Give each marketing channel its own level of respect, evaluating the effectiveness of each using data that is relevant to that particular channel. Analyze how well the campaign is working within its channel, adjust, re-analyze, adjust again, until you are satisfied that it is accomplishing the goals you set for that particular channel.
Make sure each channel’s message is consistent with the messages from other channels, so that every channel reinforces the others. As the daily influx of data streams through each person, those reinforcing messages will build upon one another and help the entire package of marketing messages stand out in the user’s mind.
The key isn’t to compare seo vs. social media, but to use the right messages in the right formats to ensure the best chance of reaching users at the right time and in the right place. The more often you can accomplish that, with consistent messaging, the more successful each campaign will be, no matter which channel it was focused upon.
Let us know if you’ve found a particular combination of cross-channel marketing especially successful, or if you’ve had problems connecting various channels to create a tight overall campaign.
Tags: cross-channel marketing, facebook advertising, PPC, seo, Social media
Posted in Facebook, Learning Center, SEO and Marketing | No Comments »
May 24th, 2011
When preparing a Facebook audit for a brand, a cookie-cutter approach won’t work. Each brand has its own factors that must be considered when evaluating campaign strategy, but there are some common strategies you can use to create killer audits for any brand.
Many Brands Within The Brand
Some large brands consist of many sub-brands. Let’s consider P&G as an example brand. They have many sub-brands, 23 of which have over a billion dollars in annual sales. Some of these sub-brands include Tide, Bounty, Pampers, Duracell, etc. Ideally, this type of megabrand should have a portfolio of sub-brands that tie back into a central hub page. The central hub is often weak, which isn’t necessarily a bad thing, but strengthening the hub will usually result in strengthening the brand overall.
When measuring the power of the megabrand, especially when comparing against their competitors, be sure to include their overall portolio (i.e. P&G vs. Unilever), as well as brand-to-brand comparisons of the sub-brands, such as Duracell vs. Energizer.
Targeting and Relevancy
When designing sample Facebook ads, remember that success with Facebook relies on being SUPER relevant to users. Let’s take a look at a few examples of how this can be applied.
Utilize Brand Connections
If the user is already familiar with the brand via a related connection, target those users specifically. For example, Tide has sponsored a car in Nascar for a long time. Fans of Darryl Waltrip and Ricky Rudd are then likely to also be fans of Tide, so targeting the fans of those Nascar champions is an easy win.
Leverage Existing Campaigns
We can leverage the power of a brand’s existing advertising. For example, if the brand is Yoplait yogurt, you know that they’ve spent a huge amount of money on the pink lids for breast cancer ad campaign. By targeting Susan Komen, breast cancer, and related interests– and then pairing that targeted base with a message that Yoplait supports breast cancer research, you can make their existing non-Facebook ad campaigns work double-time here. Send users to an appropriate web page that has a Like button on it (preferably on the Facebook page), to solidify that connection.
Harvest Celebrity Endorsements
Someone else has already done the footwork needed to attach a celebrity’s fan base to the brand, so harvest that base. For example, Carrie Underwood uses Olay, so target Carrie’s fans with relevant messages and landing pages for the Olay brand on Facebook. There is a LOT of celebrity traffic – a lot more than people who say they like laundry detergent or batteries or skin care products – so harvest that mass of low-hanging fruit.
Milk The Competitors
There’s nothing wrong with actually reaching out to the fans of the competition. For instance, if they like the Energizer Bunny, it’s possible they’ll also like Duracell. This can be hit or miss, as some fans are loyal to a fault and won’t have any desire to switch, but you won’t know without testing. Many fans are on the fence and can be pretty easily persuaded to flip.
Manage The Audit Process
The simplest tool to have on hand during the audit research phase is the spreadsheet. First, research a list of the interests that are related to each of the brands. Create a spreadsheet consisting of one row per interest, using four columns:
- Interest
- Audience Size (shown in Ad tool)
- Relevant Brand Fan Page URL
- Relationship Between the Interest and the Brand
The relationship column should include things like:
- Is the interest a competitor of the brand?
- Is the interest a non-profit that the brand supports?
- Is the interest a celebrity that endorses the brand?
- Is the interest a current advertising campaign that the brand is using?
When possible, ensure the spreadsheet includes at least a dozen interests. If the brand is a huge megabrand, you might end up with perhaps a hundred interest targets, all in the same spreadsheet.
With smart, highly relevant targeting, you can show the benefits of a Facebook campaign to any brand – big or small.
Tags: ads, advertising, audit, brands, facebook ads platform, facebook advertising, target
Posted in Facebook, Learning Center | No Comments »
