Measuring The Unmeasurable: Facebook, Twitter, Loyalty

Have you ever wondered if all those hours you team spends "engaging" customers on Facebook or Twitter pay off?  Or have you ever wondered what the ROI of a loyalty program might be?

There are methods for quantifying the impact of these activities.

Here's one of the easier ones.

Freeze your database as of a certain date, say August 1 of 2010.  Freeze customer attributes on that day ... historical spend, recency, membership in a loyalty program, that kind of thing.

Let's use a loyalty program as an example.  Create a simple segmentation strategy, then measure average spend per customer from 8/1/2010 to 7/31/2011:

				Next
	Historical	Loyalty		12-Month
Recency	Spend	Member		Spend	Increase
0-3 Month	Low	Yes		$63.00	6.8%
		No		$59.00
	Medium	Yes		$106.00	7.1%
		No		$99.00
	High	Yes		$152.00	4.1%
		No		$146.00
4-6 Month	Low	Yes		$32.00	6.7%
		No		$30.00
	Medium	Yes		$41.00	7.9%
		No		$38.00
	High	Yes		$49.00	4.3%
		No		$47.00
7-12 Month	Low	Yes		$19.00	5.6%
		No		$18.00
	Medium	Yes		$23.00	4.5%
		No		$22.00
	High	Yes		$27.00	-5.3%
		No		$28.50
0-3 Month		Yes		$107.00	5.6%
		No		$101.33
4-6 Month		Yes		$40.67	6.1%
		No		$38.33
7-12 Month		Yes		$23.00	0.7%
		No		$22.83
	Low	Yes		$38.00	6.5%
		No		$35.67
	Medium	Yes		$56.67	6.9%
		No		$53.00
	High	Yes		$76.00	2.9%
		No		$73.83
		Yes		$56.89	5.0%
		No		$54.17

Overall, those in the loyalty program spent 5% more, $2.72 per customer, than other customers.

Now, I realize that I oversimplified an awful lot, I had to in order to make the point.  You get to see how high-dollar customers were not impacted as much (on a percentage basis) as low-dollar customers, suggesting that the loyalty program helped low-dollar customers more.  In addition, recent customers were impacted more than non-recent customers.

If you're a statistician, use a GLM-style model to estimate if the differences are statistically significant.

If you're in Finance, you can run a quick profit and loss statement.  Say you have 100,000 customers in this program.  Say that 35% of demand flows-through to profit. Say that your loyalty program costs $125,000 a year to administer, between program costs and perks and discounts.

• Profit = 100,000 * $2.72 * 0.35 - $125,000 = ($29,800).
• You lost $29,800 on the program, or $0.30 per customer.

Again, most analytically minded folks and CFOs will find many ways to poke holes in this methodology ... I'm stating the methodology this way for illustrative purposes.  Be creative, enhance the methodology, and come up with your own ideas.

This methodology does a reasonable job of "measuring the unmeasurable" ... you can't execute a test where loyalty members, Facebook Fans, or Twitter Followers are not allowed to engage with you.

Ok, time for your thoughts.  What methods have you used to measure issues like the issue outlined in this post?

Loyalty Programs

Loyalty programs are popular, aren't they?

Too bad that they are often ineffective.

The key to a good loyalty program isn't the loyalty program.  Rather, it is the inherent customer behavior that is tied to the loyalty program.

Loyalty programs have a chance of working under the following conditions:

• An annual repurchase rate of 60% or greater.
• Customer places five or more orders per year.

In these situations, the customer has a need that can be met by a loyalty program.  Maybe the customer wants to feel special.  Maybe I want to sit in Economy Plus on United.  Maybe the customer wants to save 5% or 10% on a purchase.  Maybe I want to pay six cents less on a box of Mac 'n Cheese over the course of one hundred boxes over the course of a year.

In any of the above situations, the benefit is amplified by purchase frequency.

Too often, I witness loyalty programs tied to infrequent customer behavior, as if the reason the customer doesn't buy more often is because there isn't a loyalty program.  Wrong.  If a customer has a 35% annual repurchase rate, and only purchases 1.3 times per year, there's very little incentive for the customer to buy more --- the customer simply doesn't have a need to buy more often.

Loyalty programs have the best chance of working when the customer is already pre-disposed to buy numerous times per year.