Kevin Hillstrom: MineThatData: OMS: Optimizing Landing Pages

In old-school catalog marketing, it was hard to isolate the impact of any spread in a catalog. But in e-commerce, we have better metrics, metrics that help us understand the best ways to merchandise our website. We're grateful that we have talented web analytics experts and great software to guide us through the decision-making process.

Now we have the Online Marketing Simulation, the "OMS". And for many of us, we finally have a tool to understand the long-term impact of a shift in merchandising strategy.

In one dataset, I looked at thirteen merchandise divisions. Two of the merchandise divisions underwent significant online changes, one was featured prominently due to conversion rate improvements, one was de-emphasized because conversion was poor.

Given the results, I can simulate the five-year impact of this short-term decision.

In terms of annual sales, there was no difference between the old strategy and the newly optimized strategy, over a five year period of time. In essence, we emphasize one division, shifting business to that division. But long-term, customer spending habits are unchanged.

What did change was the distribution of sales by merchandise division, over time. After five years, here's how merchandise sales were altered:

Merchandise Division #1 = +4.2% (this division was emphasized due to optimization results).
Merchandise Division #2 = +2.4%.
Merchandise Division #3 = +0.9%.
Merchandise Division #4 = -2.6%.
Merchandise Division #5 = -6.0% (this division was de-emphasized due to optimization results).
Merchandise Division #6 = +1.2%.
Merchandise Division #7 = -0.3%.
Merchandise Division #8 = -1.5%.
Merchandise Division #9 = -2.4%.
Merchandise Division #10 = -1.8%.
Merchandise Division #11 = -3.1%.
Merchandise Division #12 = -2.1%.
Merchandise Division #13 = -2.7%.

By making simple changes to the merchandising of your landing pages, you unwittingly impact the long-term sales trajectory of many of your merchandising divisions. You subtly shift customer behavior.

The Web Analytics practitioner and Online Marketing expert can both benefit from the OMS environment. Imagine being able to sit with your Executive leadership team, helping them understand how decisions being made today impact the future of your business? We move beyond simple conversions and KPIs, instead gaining insight into what our business looks like in the future. Who wouldn't want to know what the future of our business looks like? Who wouldn't want to know how his/her own personal actions are influencing the future trajectory of the business?

And if you don't think this information would help your organization, why not use the comments section to describe the reasons why? There's nothing wrong with a dissenting point of view, it may make for a good discussion!

Labels: OMS, Online Marketing Simulation

2 Comments:

At 9:19 AM , grofid said...: Interesting insight. Three questions:

1) How does this reflect the dynamic marketing on the internet? Usually merchandise varies by season, and you might not stick with your decision all through the year. Is this factored into the Simulation?

2) If you do vary the merchandise on the web but do not vary in the catalog or store, the effects might be different. I understand that your forensic model treats different customers differently. Is the merchandise prognosis based on all customers or on web-customers only?

3) Oftenly customers do no longer start with the homepage or category page but follow deep links to products. So a "long-tail" strategy in search engines might counter the negative impact of the merchandising decisions on your homepage. Is this part of your "equation"?

Nonetheless, OMS sounds very promising. I hope to read more about this.

Best, Martin (Germany)
At 9:51 AM , Kevin said...: Thanks for the comments, Martin!

Merchandise does vary by season, and it is true that you wouldn't stick with your decision for the entire year. In this example, the company had a dozen or more merchandise divisions --- the landing pages began to focus on one of the merchandise divisions --- the merchandise within that division changed dynamically.

I agree with #2 that the effects might be different. In this case, the merchandise prognosis was only for online customers --- the methodology allows us to fold in telephone transactions or retail transactions if we wish, we don't have to specifically focus on just the online channel.

For #3, you can code long-tail searches as a separate channel, and then measure their impact vs. brand searches or other searches. The methodology gives you a good amount of flexibility.

Thanks,
Kevin