Data Consultancy specialising in the intersection between data analysis, machine learning, digital analytics and online marketing.

Get in touch

I always find it difficult to describe what I do; as a freelance consultant I'll do pretty much anything if the price is right.

Businesses mostly pay me for my data analysis skills, particularly with a focus on paid search marketing. I know more about data science than most PPC experts and I know a lot more about PPC than most data scientists.

Example Projects

  • Using machine learning to identify different customer segments for a large online publisher.
  • Setting up experiments to rigourously measure the incremental uplift of brand search.
  • Helping agencies by writing AdWords scripts to automate parts of the measure → learn → apply cycle.
  • Modelling the true incremental value of marketing spend for a luxury hotel.
  • Building custom analysis tools and dashboards for a large manufacturer.
  • Personalised training sessions on machine learning, the R programming language and PPC strategy (not all at the same time!)

You might be wondering why I don't list PPC account management here. My account management services are relatively expensive for a solo practicioner which means they are only suitable for quite a small niche; for this kind of relationship to work well for you, the business needs to be big enough that I can make enough difference to give you a good return on my fee yet small enough that I am able to manage the account by myself.

If you think this might be you, then we should definitely chat.

Greatest Hits

A curated selection of the best content on the site:

R for Web Analysts

A soup to nuts guide for getting started with the R programming language and Google Analytics.

I provide customised training for organisations to help develop the contents of this document towards a particular business need.

Multi Level Bidding

One of the challenges of biddable media is setting the correct bid based on not very much data.

This page demonstrates one approach to helping with this which uses the account structure to "pool" data together which gives a better estimate of the conversion rate.

The hierarchical model approach can also be used in other ways.

Lines of Best Fit

A very simple introduction to adapting machine learning models for custom use

Another example of how this can be useful is my post on different test models

Read more in the blog archive.