Choose your customers. Choose your clients.

“This is a choice, a huge one in the life of the freelancer, the entrepreneur or anyone who seeks to engage with the marketplace.” – Seth Godin

Key concepts when working either as a client or at an agency.

Link

“Jobs to be done” and B2C product development

The job, not the customer, is the fundamental unit of analysis for a marketer who hopes to develop products that customers will buy.

Clayton Christensen on “What Customers Want from Your Products

Quote

How a recommendation engine supercharges your content strategy (and how to avoid syndicated junk)

Do you find “recommended content” useful? It usually irritates me, little more than distracting sludge.

The content syndicator gives a free widget (a list of news articles or video clips) in exchange for a share of advertising revenue. Implemented correctly, it can be lucrative and add value by filling a gap in one’s editorial coverage.

Most of the time though, the paid content widget does little more than splash irrelevant link bait across the page. It kills the perceived value of a respected brand and makes editorially-questionable sites appear all the more cheap and spammy.

For example this nasty Sponsored Content by Taboola widget is from the bottom of a sports web site :

image

Completely irrelevant!

Quality sites don’t want this kind of junk, but giving customers recommendations is a great way to add depth and value to one’s product.

But how would a recommendation system work?

Recommendation Engine Requirements

I’ve been thinking a lot about the requirements of an effective recommendation engine for use in an editorially-driven product.

At the most basic level, there is content-based filtering. This kind of filtering uses aspects of content itself to group like items and determine what is most appropriate to recommend. Some criteria would be:

Time-based

  • Most recent
  • Editorial ranking

Relevance to a given topic

  • Category
  • Keyword
  • Popularity
  • Additional metadata?

Diversity

  • Limits on number of recommendations from a given subcategory

Content types

  • Articles
  • Videos
  • Broadcast streams
  • Photo galleries
  • Other “atomic” units of content

Collaborative filtering generates another kind of recommendation and uses multiple inputs to determine the most relevant content. Most of the time this means the user will have expressed a preference for a certain kind of content (film, book, sport team, player, league, etc) and the system matches those preferences with the preferences expressed by other users to arrive at a set of recommendations.

Used together, a recommendation engine that combines collaborative and content filtering, user accounts and personalisation supercharge your content strategy :  increases engagement, lengthens time spent within a product and delights users with relevant, useful content.

How a recommendation engine supercharges your content strategy (and how to avoid syndicated junk)

The idea is that you need a ton of website visitors, then some of them become become leads, and then after you do something (the usual recommendation is to bombard the leads with marketing automation) they relent and pay you money, thus becoming a ‘customer.’ 

I hate this, because it’s shortsighted.

– by Ben Chestnut of MailChimp & TinyLetter

I see this sort of scattershot, cynical approach from marketers all the time. It’s irritating, illogical and counterproductive.

Love your customers.

Image