Thoughts around conversion and effectiveness of marketing campaigns


Before I start with my version of nonsense on this topic, let me warn you this is just a note on the idea- not exhaustive, not detailed.

Okay, so how this cropped up in the first place was - I read David Ogilvy’s ‘On Advertising’ which is a fantastic read. It talks about the concepts around advertising and setting up and running an agency. While he wrote about O & M’s brilliant campaigns (like the Rolls-Royce ad below) he also mentioned about tracking the effectiveness of the campaigns.

After all, his clients wanted to know if his marketing budget was used well or it just ended up with the agency.

During those times, coupons attached to the copy was one of the ways to measure the effectiveness. The coupon would give the customer say a 10% discount and the company could quantify the effectiveness of their marketing spend.

Another metric could be the number of request for brochures to quantify the ad spends.

But what could be such parallels in digital space?

People talk about  web analytics and dashboards, but as most of these pundits will say (almost like issuing a disclaimer I feel) that the tools and results are only as good as your ability to use them.

You collect tonnes of data, but your data is not segmented well. There are duplicates,

So how could you track if your social media campaigns are actually effective? For one of my projects, I thought of a simple model that could do this.

Let’s say you want to measure ‘How many people who access my free content on YouTube actually go ahead and buy the paid stuff from my site?’ What you could do here is at the end of your YouTbe video have a coupon code that gives you a 5% discount - this will give you some sort of idea on your conversion rate.

But most businesses have a mix of  different delivery modes (written content, blogs, videos, social media marketing, search engine traffic etc.)

So if you were to check how to measure which source gives you the best conversion, what you’d do is work backwards from your payment link.

Before landing on the payment page, where were the users - Facebook, Twitter, Quora, LinkedIn, or maybe your own website?

This could help understand the trigger for conversion - is it premium features or is it the uniqueness of the content? (let’s say someone sees something that only you offer and then moves to the payment page)

A lot of analysis can be done, but for this what is needed is not data alone, you need a model too.


  1. Your model or method makes sense
  2. The data you use is clean
  3. You are looking at the complete picture and not just one section

For example, let say you have method A with a conversion rate of 80% and method B with a conversion rate of 10%. Which one is better? A?

Not necessary, let’s say you measure the conversion as ‘People who have visited my page via this, buy something from my site’

80% of visits from source A, buy something and only 10% of visits from source B buy something.

But the question one should be asking is what is the exact number of visits from the source? Let’s say A has 100 site visits and B has 10,00,000 visits. This means that more people who come via B will buy stuff because of the high volume  of the source.

So in short

What I’m saying is that all of this digital marketing, SEO nonsense is a scam if you don’t think through the model. After all - it’s Lies, damned lies, and statistics