Measuring ROI and Success with Analytics

SONY DSCOur 11th Analytics class was about ‘how’ to measure an online strategy’s return on investment (ROI) — which sounds much simpler than it often is.

Some of the many challenges are:

  • Assessing SEO  – It’s hard to estimate the SEO’s ROI because you can’t estimate the full impact of the long tail. A workaround is to multiply the search volume for top key words by 3.3 (or 30% of the possible clicks) but this is just an estimate.
  • Defining Social Media ROI – You can’t always definitively tie a social conversion to a specific financial metric.  Subsequently, you need to assign a financial value for each social engagement metric by testing and validating social activity. For example, if you generate a $1,000 sale for every 10,000 Twitter re-tweets (RT), each RT is worth $10
  • Attribution Analysis Quandary – There is no definitive answer on which brand/campaign touchpoint ‘wins a visitor over’ and prompts them to complete the desired action (e.g. purchase, sign-up or send an email inquiry). It may be the fourth visit to the website or seeing the item in the store after after reading about it once on the website. As an alternative, Sofia outlined a blend of Media Mix Modeling and Marginal Attribution Analysis. Specifically, this means measuring your baseline, allocating part of the budget to one marketing channel, run tests with/without it and track impact until the value declines. Repeat for each channel until you find the most cost-effective combination with the highest returns.
  • Offline Impacts – Offline promotions and external events (e.g. new competition, scandal, market crash) can skew results. You need to take extra steps to track offline promotions, such as setting-up vanity URLs to track coupons, and find rationale for traffic patterns.
  • App ROI – Apps aren’t cheap to make or run. To measure their ROI, you need to consider the Total Cost of Ownership (TCO), including development and ongoing operation costs, which can fluctuate over time.

So while measurement is a finite science, measuring online ROI is part art because it requires some subjective decisions.

Sofia provided comprehensive best practices and a number of formulas, which I think will give me long-term value — or ROI on the cost of her college course. However each campaign and scenario is different and there is no ‘cookie cutter’ solution.

What does this mean to me?

For example, the business model for my senior project generates revenue by selling online advertising. If I need 10,000 page views per month to sell a $1,000 ad but I currently only get 9,000 views per month, I might use a keyword upgrade to drive traffic to my site. The resulting formulas might look like:

For SEO Revenue:

18,000 (people searching for my keywords) X 10% (average success rate, where I define success as clickthroughs with page views) = 1,800 new page views

$1,000/1,800 = $0.50 value per page view

For ROI:

$1,000 (July SEO Ad Revenue)/$200 (Contractor to identify keywords and incorporate them into site) = ROI of 5

Illustration Source: Lisa Solonynko /Haml via Morgue File.