Automating Domain Portfolio

https://slides.com/anvius/berlin-2017

@anvius 2017

Text

DOMAINING EUROPE 2017 BERLIN

0. About me

@anvius 2017

Mathmatician, in domain market from 1997 (20yr OMG).

BigData and Machine Learning applied to domains.

My company is small (8 employees) with no office and using videoconference daily.

I live and work quietly in a sea village, Machine does everything. :-)

DOMAINING EUROPE 2017 BERLIN

1. The problem (2011).

@anvius 2017

We was renewing over 5000 domains per year.

Sporadic sales only.

No annual profit (therefore no business)

DOMAINING EUROPE 2017 BERLIN

2. Der prozess (not Kafkian)

@anvius 2017

We needed:

  • Clean portfolio and select most profitable (probably) domains.
  • Design strategy to buy and sell.
  • Automate most processes.

DOMAINING EUROPE 2017 BERLIN

...2...cleaning...

@anvius 2017

Domain target

  • Liquid
  • Subliquid
  • Notional

Domain type(of)

  • Acronim/numeric
  • Product/service
  • Fantasy
  • Future brands
  • Type-in
  • Hacks
  • Notional marketing

DOMAINING EUROPE 2017 BERLIN

Customer type(os)

  • Big companies
  • Small tecnological companies
  • Minimal companies and individual companies.

Wittgenstein: A language is not "real" without scope.

3. Valuation.

@anvius 2017

Find relationship between data and price.

Using Machine Learning for training and analisys.

We are using linear regressions with each dataset.

We can mesure error gap.

DOMAINING EUROPE 2017 BERLIN

4. Sell and buy.

@anvius 2017

We can mesure:

  • Amount of domains we will sell.
  • Expected profit we will do.
  • Quantity of people interested in our domain.
  • How much they can pay.
  • How much time portfolio will live.

DOMAINING EUROPE 2017 BERLIN

5. Next steps.

@anvius 2017

  • Sell API for Valuation, Brands, Email to sell, etc...
  • Expand to other languages: Chinese and English
  • Accelerate growth with more domains (more than 100% per year)

DOMAINING EUROPE 2017 BERLIN

Summary

@anvius 2017

  • Automation has a multiplier effect that tends to infinity for (almost) everything.
  • Machine learning can be more reliable than a human because it has no feelings toward a domain, so decision is more accurate.

DOMAINING EUROPE 2017 BERLIN

@anvius 2017

Twitter, Github, Telegram: @anvius

Email: anviusr@gmail.com

* All images from UnSplash.com

The end.

DANKE - TNKS - GRCS.

Questions...

DOMAINING EUROPE 2017 BERLIN

Berlin 2017

By Antonio Villamarín

Loading comments...

More from Antonio Villamarín