This post on TechCrunch by a fellow who runs a daily deal aggregator really got under my skin. It wasn’t so much that he was trumpting the daily deal model as much as he was providing completely unrealistic numbers.

If you take a peek in the Facebook comments, you’ll see a nice, long response by yours truly. Yes, I’ve finally decided to weigh in publicly on the rage that is Daily Deals (Groupon, Living Social, etc).

I’ve reposted my response below for posterity, but let me be clear: there’s a place for daily deals but they are not a panacea. They can be one of the most expensive ways to acquire new customers and retailers need to be careful.

Here are my three (3) rules for evaluating deals:

1. Always make a decision on the numbers – There are some things you might do just for the fun of it. We usually refer to this as “doing it the brand”. However, deals where you are putting real money on the table need to be evaluated using metrics and realistic expecations.
2. Always cut your expectations in half – If it turns out you’re right, then you can be pleasantly surprised.
3. Always be blunt when someone tries to make you violate #1 or #2 – You can’t blame the partner if you didn’t stick to your guns.

If it looks too good to be true it probably is, but that doesn’t mean there’s not something to it.

Now that I’ve said all the tough guy stuff, let me also say that I’m not exactly shy about business. I’m always up for trying new ideas and testing new channels. I mean, ThinkGeek grew by 55% last year. Don’t think I’m being cocky here either. A lot of people worked very hard to make this possible. What I’m saying is that we certainly didn’t accomplish results like that by running away from new business.

So, here are my three (3) rules for testing new sources:

1. Limit Your Exposure – A test is only a test if you can control the size of the deal.
2. Avoid One Way Streets – Can you back out of the deal if something goes wrong or shut down the test early? if not, you’re probably going to regret it.
3. Measure Everything – If you can’t track the test, you can’t say it worked. If you don’t measure the results, you’ll be n the dark about true impact of the rollout. Measure everything you can.

Maybe this isn’t a perfect formula but it’s working pretty well.

Got an opinion? Drop way down to the comments and tell me what you think.

Appendix: The Daily Deal Reply

Here’s what I had to say:

Sorry but your numbers are unrealistic for most if not all retail businesses.

The deal hunters of the world are interested in only one thing: deals. They won’t pay more, and they’re more likely to take advantage of low margin products (to maximize the value of the deal). There’s nothing wrong with that of course, but retailers really need to think this through using realistic data.

Here’s a analysis of your post and model:

Overage: The example in your post assumes a 50-100% overage rate. In the retail world, that simply does not happen. Ever. Your Excel model is a little more realistic, but 14% is too generous for forecasting. In selling to a discount-minded audience, the best you can realistically expect is 5%.

COGS: In your Excel model, you assume 40%. This may be true of service business but not true in retail. Even most apparel businesses don’t get close to that number. In addition, discount-minded customers rarely buy items where you’ve got great margin. They look to maximize the deal by buying items on sale or items with razor thin margins where they can’t get a deal anywhere except by using that coupon.

Commission: Your Excel model assumes a 40% commission. That’s not happening with any deal network. 50% is the floor and it goes up from there.

Using your model, if I change overage, COGS, and commission to realistic numbers the merchant loses $3 per customer.

But I can’t really stop there…

In your model you assume three (3) repeat visits for new customers. You also assume that the “conversion-to-repeat” will be static. In other words, there’s no entropy over time of the original 20% who return. This is unrealistic. First, for forecasting purposes you shouldn’t assume more than one (1) repeat visit. If you can’t make money from an acquisition after the second sale, you’ve got a bad channel. Second, over time, the number of customers who return from the initial acquisition is going to degrade. Not only that, but it’s going to degrade very sharply. You’re going to go from 20% on second purchase to 5% to 4% and then probably level off after that at 2% of the original batch of customers… if you’re lucky.

I’ll meet you half way though and only change repeat visits to 1.5 and ignore the entropy factor. Now, I’ve lost $5.90 per customer.

You also recommend that businesses provide a second incentive to a discount-minded customer. You’ve brought in a deal hunter to your shop and converted them to a customer and then you reinforce the discount mindset by offering a second deal? I’m sorry, but this is really bad advice. There’s nothing wrong with bringing in deal hunters, but retailers need to understand what that does to their margins. They also need to understand that it is very, very difficult to convert a deal hunter to a full margin customer, and by very difficult I mean impossible.

So, my advice to retailers is to be very careful about the offering and very conservative on the forecasting. If you can, limit (i.e. eliminate) your exposure on low margin items or stacking discounts. Deals are great, just be smart about it.

To help, I’ve uploaded this version of the model using the more realistic assumptions to Google Docs and made it public to the web. Interested parties can download it here.

•