Lightning Bolts & Stats

Have you ever flown through a thunderstorm? I’m writing this email on board the flight to New York (in a massive Airbus). Last night a bolt of lightning nearly hit the plane – it did make a load bang – scared the sh*t out of everyone, especially Charlene! Fpr more information on the possibilities of this near-death encounter we faced – read this article I found on USA Today Luckily, we escaped unharmed. The plane was grounded at Dakar, Senegal for an extra hour, until the storm cleared.

The chance of dying in a plane crash is something like 81,000 to 1 – a car accident causing your death is more likely at 90 to 1 – but you try rationalising these odds while sitting 32,000 feet in the air! Why aren’t we more rational way up high? Perhaps it’s because most of us are control freaks in our own little way. When sitting in the drivers seat of a car, we feel that we have total control over our destiny, so the statistics don’t apply to us. How short sighted – even I don’t blink at driving 160km/h but shudder at the slightest bit of turbulence (which, incidently, has a 750,000 to 1 chance of downing an aircraft of this size).

What’s my point? Well, for those of your who manage PPC campaign, you know that you’re working with thousands of keywords, each with a unique conversion ratio. So in the first few days, you might get 1 click across 1000 words at $1/click but not convert a single customer. What happends? Well, firstly, you don’t have enough data on a keyword level to determine which keywords are good converting and which ones aren’t. Secondly, you know that it’s statistically impossible to get a 100% conversion rate across 1000 keywords with only 1 click per keyword. Therefore, you should know that you do not have a statistically valid sample size to make any decision regarding the campaign.

So, rationally, the expected outcome of any assessment of the campaign is : PANIC!!! Even though you know that your expectation should be zero, you just spent maybe $1000 without any ROI – most offline marketers wouldn’t even blink – but the savvy online campaign managers want ROI more than anything else. However, this knee jerk reaction is actually more damaging then good (unless your total campaign budget is $1000!). The safest way to launch a campaign with a limited budget is to obviously bid lower amounts for lower traffic volumes, until you have enough data to evaluate a particular keyword or group of keywords, then increasing prices will lead to more volume, hopefully with a directly proportional increase in ROI. You need 2 budgets, 1 for gathering data (test budget) , and the other for the full campaign. Always be prepared to burn the test budget – no burn, no bucks.

Going into PPC with a new product normally means treading on existing campaigns which are established and the existing players know what works and will defend their positions if an incumbent starts taking traffic away. All you are trying to do is probe the system and establish which keywords you want, after that, you can lay on an assault. Remember that you are coming from behind, literally and playing catch-up. The existing players know about the keyword’s traffic volumes, their site’s conversion rates, life value of customer at a keyword level, dayparting, etc etc. You need to gather this information before trying to score the goals.

Vinny’s 3 Steps to PPC Success:

1. Gather information from keywords using a reasonable test budget. The test budget should usually be 20% – 30% of your expected monthly PPC spend.
2. Eliminate poor performing keywords (roughly more than 200 clicks and no conversion – site dependent) and negative match where appropriate, to avoid burning cash.
3. Relax, be patient and remember to follow the numbers, stay within budget, and don’t panic – it’s a number game – apply your strategy and stick to it, provided you are within your limits.

I did also manage to catch ” Walking Tall” on the plane, starring “The Rock” – not as bad as I expected – I’d give it a 6.5 out of 10.

If you enjoyed this post, make sure you subscribe to my RSS feed!