Ratios and Leading Indicators – what drives me

Every industry has a specific, almost canonical metric it pays attention to and benchmarks its participants against one another. I wanted to take a few seconds to think out loud about the basic ratios that guide me on a daily basis.
Some business cultures emphasize one financial and operational ratio over another. In Switzerland and Asia, inventory turnover is a big deal. In NASDAQ-listed tech stocks its price-earnings as a measure of value and expectations (P/E). In other industries it’s revenue per employee, while others focus on average revenue per user (ARPU), and then there are operational, specific ratios such cost-per-click.

I live in a world of expense-to-revenue, or “E2R“, expressed variously as “e:r,” “e/r,” and “e|r.” As a relatively green marketer, I assume this is the prevailing canon for measuring marketing effectiveness and efficiency: for every dollar spent in market, what revenue is received, and what budget should be fenced off based on the top line to support a marketing expense. In dummy terms, “How much did I spend to make $X?”

A 1:1 e-to-r means I spend a dollar to make a dollar; that comes out as a 100 percent E/R. We can call that a wash. A negative E:R is when I spend two dollars to make one dollar. This is a 200% E/R and means I’m looking for a new job. A good E:R is when I spend one dollar to make five, a 20% E/R, or as Warren Buffett would say, buying dollars for twenty cents.

This applies in interactive marketing on ad tactics and assumes perfect insights into the purchase of an impression (based on cost-per-thousand impressions, or CPM), yield click-throughs (measured by click-through-rates or CTR), and ultimately ARPU, or average revenue per user (or unique visitor). If I am able to track the user from the first purchased impression to the final checkout, then I can credit the tactic with the sale.

Still with me? This is predicated on that user accepting the tracking beacon, or cookie, from my metric system, a cookie that the user gets when clicking on the search term or the banner ad I’ve purchased for N. If the user permits cookies — in my case one deposited by a script on every page of our website that will permit our metrics tool, Omniture SiteCatalyst to follow that user across multiple visits — then, if our commerce guys have successfully done their job, the user will buy something and I will be able to credit the original marketing tactic with the sale.

Sounds hard and imprecise? It is, but compare it to a newspaper ad. If I buy a full page ad in the Daily Planet and it is seen by a theoretical 100,000 readers, and sales from the zip codes where those 100,000 readers live goes up by 10 percent can I declare the ad a success?

This is the old handgrenade-and-horseshoes school of marketing — the famous I-know-half-of-my-advertising-works-but-which-half” school of marketing.

So back to what I worry about, which is interactive advertising: the interesting thing in assigning expense-to-revenue ratios to a particular tactic is knowing what the intent of the original message was, and second, knowing what it yielded.

Those tactics are pretty simple. Search engine marketing, which is very precise in that I am purchasing clicks through an auction model, but which is usually promotion driven by non-brand tactics like “cheap notebook” or “fast PC.” Then there is display advertising — banners, buttons, IAB standard graphical units — and that is arguably either pure branding or can be pure promotion, but generally are a hybrid of the two. Success in assigning an E/R ratio depends on accuracy in following a customer through multiple visits to my store, because in my market, where I am selling products for $500 to $3000, the customer is researching, comparing, and considering before hitting the purchase button.

We can do this, the rest is pure optimization, arbitrage, and discipline. It’s actually kind of cool and very fun, and a nice feeling at the end of the week to say, “I bought a dollar for twenty cents.”

Author: David Churbuck

Cape Codder with an itch to write

0 thoughts on “Ratios and Leading Indicators – what drives me”

  1. Agreed, and I’d never vendor bash. OMTR is good stuff. Accuracy is a challenge regardless. There are multiple methods for reconciling (skinning) the “paradox of uniqueness” and first or third party cookies (the cat). I’m a bit keen on UNCA’s NetInsight and WSSI’s Visual Sciences products, which use alternative and advanced data collection methods not as prone to detection by tin foil hat wearers, of which I must include myself. 😉

    The truth is out there.

  2. Judah,
    I think web analytics, at least in the case of large volume sites, needs to be based on statistical sampling and not canonical precision, as there is, as we both know, no such thing as absolute precision as long as people justifiably reject tracking beacons from their sessions.

    It should be possible to extrapolate from a verifiable sample, a baseline for planning purposed, but in the end, it will be impossible to achieve 100 percent accuracy in an expense-to-revenue assignment without 100 percent cookie acceptance.

  3. (slapping head)

    Sigh. It’s ALL a proxy for NPV. Net present value is the name of the game.

    Cable companies can discuss revenue per subcriber (ARPU) as can cell companies or most subscription based businesses…

    Retail discuss same store sales while airlines use passenger miles…

    They’re all proxies, shortcuts, dumbed down simplified metrics that (hopefully) are one of the largest — as in most significat — variables buried within the NPV calculation.

    Yours, my dear David, is a simple ROI spun backwards.

    What neither ROI captures nor your E2R is this…

    Would you rather have an E2R of 20% on a $10,000 investment or an E2R of 17% on $150,000? That’s where NPV would assist in making the right decision. Chasing E2R without looking at the NPV means money left on the table. E2R does not hold ad infinum. You can’t find an E2R of 20% and push $50 billion through it.

    The timing of your cash flows are compact so the TVM (time value of money) isn’t a huge variable here but NPV is much more important than either ROI or your basterdized version of it called E2R.

    I wonder if ROI is so complicated that marketers needed to do it backwards and give it a new name.

  4. I think it all comes down to “revenue per thousand” and whatever clever derivatives we can derive from there and agree on whether based on expense or hedge. I think both of yr angles have utility as lenses to look on web business.

    I am a big sucker for “profitable revenue.” I live IRR and hurdle rates, and one of my goals is to continue to apply it to web analytics. My grad skool was bread and spread on npv, which to me is simple to define: the profit of tomorrow’s actions in today’s dollars.

    As for stats, again, I won’t vendor bash. 🙂 I think that with more advanced data collection methods (read: not just page tags imho), the minimal the use of holy stats (see Unica and VS).

    For me, the problem with the application of stats in web analytics is the drill down, in the ad hoc analysis features. It is very conceivable that as you drill down to discover rich segments you sample samples of sampled samples. Tends to lead to lower confidence in the predictable powers of yr analysis.

    Yes, it remains arguable as to what effect this all has. But I think as we use metrics to segment the long-tail or outlier of whatever to apply very, smart “new marketing” techniques to what I call Web 3.14159, the more accurate and the more integrated (across people, process, technology) we need our web analytics systems.

    Cardinality, how’s it handled at run-time? When you execute a query how many rows of detailed data are consulted? Then I ask:

    How many customers (or even sessions do I have).

    Do I believe?

    The truth is out there.

    🙂

This site uses Akismet to reduce spam. Learn how your comment data is processed.