The key to successful experimentation is managing the balance of testing vs. implementing your “best” plan. You want to ensure that the process of experimentation does not have a noticeable negative impact on results and financial contribution. You are testing high-risk, high-return strategies but the net impact should be neutral. First of all, the experimentation is typically delivered to only a small portion of the target audience. Secondly, there is a good probability that the experiments generate just as much positive lift and negative shortfalls to net even. Finally, the experimentation should not have much of an incremental cost with the exception of losing some efficiency, such as purchasing market-specific media without national media discounts.

Each company will need to find its own balance of testing but one split to consider is allowing 20% of your budget for testing slight variations of low-risk and about 5% of the budget for experimentation of high-risk, high-potential strategic alternatives. That leaves 75% of your budget dedicated to the market plan that consists of the best-known strategies.

Budget Spread
Core Marketing Plan 70% – 75%
Low-Risk Variation Testing 15% – 20%
High-Risk, High-Potential Experimentation

5% – 10%

If the 5% experimentation portion of your budget under-performs by 20% and all other marketing delivers as expected, the worst case scenario is that you delivered 1% below your total objectives (20% * 5%). If that same budget outperforms the target by just 20%, there is an immediate payback of 1% on delivered results but then the next version of the core marketing plan can increase close to 20%. In that scenario, experimentation that generates one winner out of every 20 experiments that fail will break even.

So the upside potential makes this well worthwhile. Over the long term, the learning leads to new strategies implemented for the majority of marketing to deliver better outcomes, which can more than compensate for any lost sales from experimentation. The learning process also leads future experimentation in the right direction, building on the knowledge of which strategies are most effective.