IMUG – Understanding Multilingual Tweets

internationalThe most surprising for me was the fact that Twitter content is often translated by users (Twitter uses Bing machine translation to accomplish translation of tweets) – and the volume of translation requests was four times larger than anticipated.

People want to translate tweets…

Twitter will allow to use only 140 characters, but they can be in 140 languages.  Twitter is internationalized; Twitter interface is translated into multiple languages using crowdsourcing.  Set of core languages used for 99% of online communications is fully supported, some of the languages from the remaining 1% may or may not be supported.

Mixed-language tweets are also common (of course, immediately, I had to look for them in two languages I speak – mixed language tweets were not hard to find…  Titter determines language of the tweet by the most prevalent  language used in the message.


Some tags might be spelled identically in different languages – to determine the language of the tag, Twitter uses an algorithm, but it is not perfect.  Hashtags can use multiple languages within individual hashtag; they are not translated (hashtags are not sent to Bing machine translation).  Translation can take more space – for example, a 140 characters tweet in Chinese can become a couple of paragraphs of text in English.

Even if Twitter message has just 140 characters, this message has quite a few elements in general and many of these elements can be affected by language of the user.


Twitter is monitoring the network for any “tweeter storms,” which can happen during different events, including social revolutions.  Twitter makes sure that its infrastructure can support  increased activity.

I tried to analyse my own Twitter account with help of Fliplingo, and, surprisingly, found non-English followers.  Even if the percentage of international followers is small in my account, I did not expect to fine anybody at all.  My native language did not even make it into top 5!  Language mix is probably related more to topic than to the speaker (or the personal background of Twitter account holder 🙂 ).


googleGoogle hosted this lovely event, and it was my first opportunity to visit Google campus. I must admit I was very curious after years of reading books about Google history and numerous mentions in business books. My guess, every business book under the sun mentions google as an example – of something.

The food was excellent!  However, I must admit that all events I attended in the Bay Area during last three months provided wonderful food choices 🙂

Google lobby met us with sounds of live piano, and the conference room had chairs of different colors: blue, red and yellow.  I thought the sitting arrangement was intentionally “on-brand”  🙂

Tweet: Testing Click to Tweet – am I the only one who have not used it yet? 🙂


Marketo Pumpkin Carving Contest

pumkinsYes, I happened to be the right audience for Marketo Pumpkin Carving Contest – I love marketing, and even if I have never tried carving a pumpkin, it definitely looked fun 🙂

First, I needed a pumpkin.  I was determined to find a purple or blue (or some other color pumpkin closer to Maketo color, as, of course, I wanted to create something in between Marketo and Halloween – I am a fan of both.  And the idea itself was beyond hilarious!

Hm…  I never carved pumpkins before…  and they happened to be quite hard to touch.  How can it be carved? My guess, some power pumpkin carving equipment exists, but that was a bit too extreme, and my skill level in the area was absent – or somewhere around one big misconception that pumpkins were as malleable as big orange soft cantaloupes.

End of contest for me?  Not so quickly!  😉  It would be as fun to try to carve the cantaloupe and, with a little help of Photoshop, make it properly purple.  This approach may not qualify for the contest rules, but it would be even more entertaining – plus, I can eat the cantaloupe.

monitorOK – cantaloupe and Photoshop became partners in my non-pumpkin carving efforts.  As usually, cantaloupe was delicious and easy to cut, though not as easy to put back together to pretend to be a pumpkin.  I also tried adding an additional Halloween accessory from  a company’s event.

Definitely, my “pumpkin” had to be purple – the contest idea came from Marketo.  The result was not quite what I expected, but the process was entertaining enough to consider the whole exercise worth the effort. It was a fun Friday eventing!  🙂


I was not completely sure about the Halloween accessory, but, why not?  🙂

Tweet this: This is another test of the click to tweet 🙂


Traction Conference – 2015

TractionExcellent event!  The format of short presentations and no need to decide which good speaker to miss in a break-out session was perfect.  Each speaker has something “practical and tactical” to share beside deep fundamental thoughts.  Hopefully, this format will become more popular in other events.

All available conference slides can be found on SlideShare.

Credit Karma

Credit Karma entered marketplace when the idea of paying #30 for a credit check was a common practice.  The company decided that providing the credit information for free would generate eye balls, and the business model can be discovered later.

  •  Concentrated on creation of a fantastic product (small team – product focused).  Currently the product is used by about 40,000,000 people
  • Company had just 6 people during first 2.5 years; now about 400 employees

First, we need to prove it works, and then, we need to prove it works at scale

  • Off-line marketing worked the best; for consumer products that are ubiquitous off-line marketing is an excellent option
  • Superball ad on the topic shown by a major bank generated traffic to Credit Karma
  • Company tried partnerships, and discovered they did not work as well as expected: “take a projected number of partnerships and divide it y 50…”
  • An “Aha moment” – company had significant repeat traffic.

Survey Monkey 

How to convert a user from a free to paid model:

  • Usage based (pay for more use)
  • User based (pay for added users)
  • Feature based (pay for additional features) – most difficult, but the only method that does not limit virability of the product

As viral spread is important for Survey Monkey, all features that hurt this opportunity reserved for premium packages (white label, etc.).

Survey Monkey constantly A/B tests.

Metrics are very important.  At the early stages of the company’s growth a person was hired to publish business metrics for the entire company.  Survey Monkey currently has meetings open for the entire company where metrics are presented and discussed.

Which metrics are vital, and which metrics are “noise?”  Any metric important for the specific business can be beneficial, the danger is “too many metrics.”


The Inside Story on How SurveyMonkey Cracked the International Market

When Selina Tobaccowala joined SurveyMonkey, 85% of its business was done in English. It was solidly a domestic company making slow inroads overseas. A little over five years later, they are supporting 17 different languages and 28 currencies. Currently, the domestic market is 55% of their business, but they’re aiming for it to be just 25% of their business.

At the start of 2013, nine out of the top 10 internet properties in the world were based in the U.S. (think Google, Facebook, etc.), but 79% of their users were international.

A year later, only 6 of the top 10 were in the U.S. and 86% of their users were not.

The company has 150 different variations of its pricing page to support various languages, currencies and product packages.

In Asia, traffic to SurveyMonkey is 75% mobile at this point.

Mobile does not convert as well as desktop.  Expect 50% conversion rate; though it varies in different regions (higher mobile conversion rate in Europe).

Churn is the most important metric for SurveyMonkey (as it is easier to retain a customer than to acquire one).  Survey Monkey discovered that engagement decline precedes churn – good time to send a note to the customer.  Retargeting churned customers on Facebook generated results.

When to survey?

Survey after a specific action – the person is more likely to take another action immediately after a previous one.

  • After purchase:  what features drove the customer to purchase?  Use this features on the pricing page for similar people.  Response rate of this type of the survey about 20%-30%
  • After cancellation: why cancelled?
    • Going to a competitor – watch!
    • No reasons to use the product – give reasons
    • Can not pay by using current payment method – offer more payment methods


70% of users are outside of US.

Evernote has a very passionate user base, which wants to bring the product to their companies. Evernote objective is to help these users:

  • Market to IT, so they are aware about the product when employees start talking about it
  • Market to users – explain how to approach their company’s IT, and what IT would want

What to do to go global?

  • Follow the data and let data lead you (Evernote was adopted in Japan before the company even considered any localization for this market – something in the product appealed to Japanese users)
  • Need to start from day 1 by building a global product

Android or iOS?  Android is huge!  Android helped to propel company to international expansion.

Evernote challenge – the product can be used for everything.  The company had to focus and give examples for specific use cases.

Mistakes in International markets that companies make:

  • Don’t take one model and move it around the world
  • Be as local as possible in international expansion


Everybody in the Bay Area knows Slack, not not necessarily outside of it…  Slack brings several applications together and, based on surveys, reduces email volume and meetings.  It also reduces the number of tools used.

WOM for slack is very good

  • Trying to add methodical approach
  • Sales are to “help through the buying process”

Slack has about 250 people and 1.25 million users.  Customer support is the largest team. (This was surprising for me as Slack supposed to be simple…)

Most important metric for the company is NPS (Net Promoter Score).

Slack’s team concentrates on making other tools work better with Slack.

“External email will live forever, but email inside the company may not survive..”


“Growth Hacking” is not a favorite term…  Popular “growth hacking” stories worked at specific situations and generally anekdots, rather than strategies to follow.

Biggest mistake a company can make is try to grown a bad product: if you tell everybody about a bad product, everybody will know about it. How do you know that you have a good product?  Retention.

How to improve retention:

  • get feedback, understand what works
  • find a segment of users for whom the product works
  • get users to experience the core value (Aha moments).  LinkedIn core value is growing network.

The business needs to define one core metric

  • aligned with your business
  • aligned with user’s value

Example: YouTube metric “Watching # of videos” made sense for the company, but not for the users, as users might have tried several videos they did not like and stopped after a few seconds.  YouTube improved metric “Minutes watched” includes customer value.

Targets of different teams should be summed up to this business metric.

Growth requires continuous prioritization and feedback.


Growth team at LinkedIn is over 100 people now.


Morgan Brown

Rapid experiments help growth (assuming the company had a good product).  However, from the outside of the company these changes are not visible.

Generate as many ideas for the experimentation as possible, prioritize, and establish the testing cadence.  Important: share the results of experiments with all.  If the team does not have access to learning, they can not come up with better ideas.

  • speed is competitive advantage
  • disciplined speed is a competitive advantage
  • you are not in the business of testing, you are in the business of learning



Only data-driven people will be put in charge of brands in the future.


Companies align themselves with something “more.”

  • Slack – be less busy
  • Facebook – give people the power to share

The world wants more of those. Companies need to invest in “unscalable moments of delight.”

Interesting example: Betabrand – a company, which crowdsources all clothing designs and does not use professional models – only real people.



CMO+CIO = Modern Customer Experience
CMO+CIO – Partner or Perish.  If you are not tech-savvy marketer, marry a technologist.

Interesting example of a tech product – Sproutling  – a new type of baby monitor.



Email is fundamental for the company.  Interesting, the company does not use any popular marketing automation system available on the market.

Engagement program: airbnb to guests.  Before the program started the company was silent between booking windows (what might happen once a year only).

  1. Scope the opportunity (does it make sense doing)?
  2. Pick a metric – watch this metric
  3. Run tests to prove you can move the metric

Opportunity example: first time bookers.

It is possible to move faster if scale globally first, then optimize the program.  Scaling allows to gather more data for optimization.

The company defined its own “Quality Score” for the email  – a function of positive activity and negative activity (unsubscribe).

Results of some fo experimentation: sending emails two weeks after the action was not as effective as sending emails immediately after the action.

Measurement: 90% get email, 10% do not.  Is there difference in booking by the end of the year?

Anecdote from the conference: an email to the event registrants with the subject line “Are we meeting this Thursday?” was opened 100% (the event was Thursday, and the email was sent Wednesday).


Optimization is a compound interest for growth.

Conversion optimization is a process rather than a collection of tactics.  

Don’t copy your competitors, they don’t know what they are doing. 🙂

If you can not describe what you are doing as a process, you don’t know what you are doing.

  • discover what matters for your users, and prioritize
  • we do not need more data – we need better data

Not “data driven” – “people driven,” as people are interpreting the data.

If you torture data long enough, it will confess to anything


  • Insight: people are concerned about security
  • Solutions: many!  You can test what would work


How to Come Up with More Winning Tests Using Data – article describing the approach.

Price Intelligently

Very insightful presentation, suggesting to spend more efforts on monetization… and highlighting that most of our attention is typically spent to acquisition.


However, the efforts spent on Monetization are more effective…

price-2Pricing might be different based on  a specific persona – what needs to be understood and considered.


Companies do not pay enough attention to pricing.  Typical SaaS company spends 8 hours on pricing (I am assuming per year).

Recommendation: evaluate pricing every 3 months, and change every 6 months.


Price Intelligently blog has more insight and recommendations.

Wealthfront (retargeting)

It is easier to get an active user to do more than to get an inactive user to do anything.

  • Beneficial to resend emails to those who clicked, but did not complete the action (rather than resending to all)
  • If the person did something once, ask for the same action again immediately (if business includes this type of activity).  Example:  LinkedIn asks users to endorse others.  After they do, LinkedIn asks for more endorsements immediately.  People are willing to continue endorsing for some time.


Interesting approach to business: start with academia > students will become familiar > students will bring the tool into companies where they will work.

Business took off during the economic downturn.  Currently, the product is used by 8,000 brands (including, and has 1000 employees worldwide.

During the economic downturn everybody retreated, and Qualtrics was hiring and expanding.

SalesForce believes that the largest mistake was not to invest during the downturn of 08-09.

Innovation is critical:

The model that got you here won’t get you to the next stage.  You need to be comfortable with changing the model to grow.

Enterprise buying is changing: Marketing can buy a tool independently…  Sales can by the same tool independently…   Purchasing can buy the same tool…  functional heads often make their own buying decision.  

We are comfortable when one department is trying the product.  They will spread it around.

Advice for startups targeting enterprise: think about a price… and ask for twice that much.


How to get customers who stay?

  • watch your signup funnel
  • focus on user quality, not quantity
  • ask for referrals (Mixpanel gives extra credits to companies who put a Mixpanel badge on the site).  Even in referrals, quality matters.  A company (venmo) asked to refer just two friends, rather than “all friends” – as a result, the referrals were targeted to friends who would be more likely to become customers.
  • A/B test your message
  • Don’t waste valuable marketing real estate.  If you are afraid to experiment on the web site, experiment on the email.



Communications are based on activity:

  • Users who use Shazam 10 times a day love frequent notifications.
  • Users who use Shazam once a month would be annoid by this type of communications.


“Start with Why” – this book – and a concept – was mentioned during the presentation.

Some of the speedbumps the company can avoid in building its culture:

  • Communication (over-communicate – over-share, and listen…)
  • Alignment (what is the most important objective? )

EventBright perceives company culture similar to Maslov’s Pyramid of Needs.  Where on the bottom are basic needs (salaries, etc.), next “connectedness” the company provides to the individuals, and then the “why” the company’s business matter.


For EventBright the “Why” is “Bringing the world together through live experiences.”


(Interesting: SendGrid prominently places educational materials on the home page – below the fold.  The site visitor has a call to action to follow beside anything related to the product.)

SendGrid is sending as many emails per day as Twitter sends Tweets.

  • Product/Buyer fit (target audience were developers, who did not want to bother with everything related to the email process; it was a specific problem and a specific buyer – similarly SalesForce started with sales reps a the buyer)
  • Truly listen to your customers (customers wanted services…  company invested into service; now support and customer service are the largest groups)
  • Build a team (unique and differentiated culture: there are people who would not like your culture; the culture should be unique, and having people who do not like the culture is completely normal)
    • Hiring is a privilege: do not loose sight on who is making hiring decisions – not everybody should, per SendGrid
  • Don’t get burned by the cash burn: plan for business cycle change


The company found a “hole” in the market for employee information – too many tools and needs to handle employment process.  “We wanted to make employment hassle-free for companies.”  Zenefits is one system to manage, the rest of the systems are connected to it.

This approach could threaten many existing markets.

Currently company has 1,600 employees.  How the company accommodated growth: “we hired great recruiters.”

recruiters are unknown heroes of Silicon Valley – they redistribute the talent

zenefitsInternational growth depends on the countries.  In the US, the best margin for the companies with less than 25 employees, because nobody else targets them.  In Canada, the companies to target are less than 100 employees.

Fantastic event!  The main themes were:

  • Start with “Why”
  • Listen to customers; pay attention to data
  • Process over tactics

Marketo Users Group – Survey Campaign

Silicon Valley Marketo Users Group is an excellent place to learn innovative ways of using Marketo and also discover what other marketers found interesting, challenging, or beneficial.

spearDan Reed from Spear Marketing Group gave a wonderful presentation about survey-centered campaign, integration of a third party survey tool with Marketo, and overall result from the marketing effort.  Dan shared his deck with detailed instructions for the SVMUG members.

The company wanted to collect data on an important industry topic in the area of its service offering, use this data for future multiple marketing materials, and generate leads.


Very interesting: both “phases” of the campaign (generation of the data for the report, and promotion of the report itself) are presented as one process – what makes complete sense.


A few points from the group discussion on surveys and materials generated from them:

  • used to establish the company as a thought leader
  • not necessarily offering “something”
  • sometimes, first 100 are eligible for $5 starbucks
  • filling out survey usually $5, typically, $50 is offered to setup a meeting (those who tried $50 approach suggest that the volume is good, but result is still unknown)
  • providing free access to enough data can result in amplification by other publications (for marketing, eMarketer can pick up, etc.)

Technologies that group members tried and found interesting (the rule of the SVMUG encourages sharing any technologies, except company’s own product 🙂 ):

  • PersistIQ  (sales communications)
  • SendBloom (sales communications)
  • Datanyze (data tool)
  • Uberflip (content management – for marketers)
  • Highspot (sales enablement tool)
  • PoliteMail (Outlook-based internal communications tool – ha ha – I wish I knew about it three years ago)
  • Sigstr (managing email signatures for the entire company – I wish I knew about it …  a decade ago 😉 )
  • Spiderbook (search for qualified prospect via semantic analysis of online data)
  • Persado (generation of subject lines, etc. to increase response)
  • Brandcast (website management?)

An interesting point from the group: webinars are stressful…  however, the option to pre-record the “live” webinar and “play” it during the live session allows to avoid possible technical problems.  The Q&A part is done live, as in the traditional webinar.  In addition, the speaker, who is “presenting” the recorded talk, can communicate with the webinar attendees on Twitter.

European Innovation Day

European_innovation_dayThough the event was dedicated to European entrepreneurs, many discussions revolved around entrepreneurship in general and the environment entrepreneurs need to thrive and bring innovation to the market.

Currently Silicon Valley generates more new $B+ successful ventures than other locations, and this gap is growing.


Silicon Valley was successful in creating this environment, but Silicon Valley is not a place, it is a recipe. What are some of the elements of the recipe, which make creation of new companies and new product possible?

  • Entrepreneurial culture – reliance on diversity (currently, over 60 languages are spoken in Silicon Valley)
  • Failure is not considered a stigma, but rather experience
  • Access to capital (investors also provide mentorship)
  • Access to technology – roles of Universities, which adopt policies to help the “recipe” (for example, an idea generated by students can be easily commercialized without any legal issues)
  • Large companies are breeding ground for management talent, where it is considered acceptable to work for a couple of years in a large organization, then move to a smaller one
  • Favorable government policies, as “without regulations there is no free market”
  • Liquidity (IPOs and M&A)
  • Infrastructure (law firms, accounting firms, etc.)

Over-regulation kills innovation.  Under-regulation results in chaos.

44% of European startups are acquired by US companies.  In Europe, startups are not spon by the university system. Venture capital play lesser role in European funding system; government step up to help the process.


European regulation is currently complex “we have everything 28 times.”  European Commision is addressing this problem through “Digital Single Market Strategy.”


Why do we need to pay attention?  There is more and more transactions with Europe.  In general, all successful companies are global companies.

Interesting point on the startup growth: at one point, a company that hired generalists in their fields need to switch to hiring more narrow specialists, and these specialists need to be managed differently.  This is a shift in the company’s structure, at which point some of generalists (and even some of founders) might not feel comfortable and might decide to leave the organization.  This shift needs to be carefully managed.

IMUG – Skills Desired in Localization Careers


Interesting coincidence – a topic related to localization careers was discussed in a lovely LinkedIn office.

The localization industry jobs descriptions are very diverse, and may not match a “typical” resume.  Often, the person can still get the job with resume that is dramatically different from the job description.  Not all job descriptions are written by hiring managers, what can add to requirement confusion.

Why this job description confusion exist?

  • no certifications
  • no career paths
  • no standard job titles

There are three categories of jobs in the industry: the linguist, the localization manager, and the localization engineer.  I must admit the last type of jobs was surprising to me – I did not realize the depth of specific knowledge required by developers to be effective in the industry.


Industry experience is not a surprising requirement, as in-depth knowledge of complex subjects would be needed for the translation.


Interesting: translation skills barely made it into the top ten requirement for managers, while vendor management and localization best practices were considered more important.


Rather unexpectedly, the process and workflow improvement skills were very important for localization engineers.  Could it be explained by the complexity of the localization process, which inherently includes many diverse considerations, people, and time zones?

tech-skillsThe industry is moving to the direction of requiring understanding of any tool in a category rather than a specific one.  This was also the recommendation to the hiring managers.

The blessing and the curse of the industry is its relatively small size and specific skillsets.  From one side, it makes people with the skillset more valuable and hard to find, from another side, the reputation of the company or the person becomes very important.  One company was loosing candidates because of its reputation.  The company stopped hiring and concentrated on making its current employees happy.

LinkedInAs any other area of business, localization industry is changing based on the new tools and new opportunities these tools bring to the market.  Most likely, the fluctuation will continue in the nearest future. Or, maybe, the pace of change won’t slow down for a long time.

My guess, some certifications will appear, but the career paths will continue to weave around new opportunities and tools.

It will be interesting to watch the industry to evolve!