When people talk about reporting on data, they usually envision a scenario where the analytics user traverse the application from one report to the other. Which might or might not be OK. But I am sure we agree on this just being one way of communicating the information in the data we’ve collected. We can communicate this information in anything from reports, dashboards, scheduled emails, documents to presentation slides – and no one channel is per definition better than the other. The channel is simply something which is chosen as a best possible medium to support the goals you must have in distributing this information to begin with.

That said; I would like to advocate for another communication channel, the Metric Alert Email, because it tends to help focus on metrics that matter, be on time, insightful and for the most part almost actionable as is.

Instead of pitching a web analytics tool feature, I’ve create a setup where we use Web Analytics Metric Alerts to illuminate a set of organic Search Phrase Opportunities. There is a whole industry dedicated to keyword research and insight, which I think is just splendid. Alerts is not to replace any of that, this is more to show you how a simple thing like alerts can create a lot of value inside your organization; which is interesting, if you don’t necessarily have a full-time analyst work your data day and night.

So HOW do I use Web Analytics Metric Alerts to create Search Phrase Opportunity Insight ?

1.

Go find your most popular search phrases report, which in most tools, means a report that shows a sorted list of all the search phrases that delivered traffic to your website. The list is usually sorted by visits, which is also the case in the below screen shot. This is, as I am sure you’ve heard repeated a hundred times, a traditional long tail distribution. Meaning that we have a few keywords bringing us a lot of traffic – and an abundance of keywords which drives little traffic to our site, individually, but in aggregate makes up a large traffic influx.

I don’t necessarily think that this report, holds much insight by itself without further analysis. This is, being very blunt, more a report that we do for your entertainment.

2.

Let’s assume, just for a second, that you are in retail and that you are driving online revenue. If that’s not the case, then the following examples and conclusions are still very much valid, but you would of course have to substitute the Revenue metric, with another success metric of your choice. A different metric that for some people could be a proxy for off-line revenue.

The first thing I suggest we do, is customizing the above most popular search phrases report, by appending it with the following metrics:

- visit to sale conversion rate
- product views
- avg. order value
- revenue

But not only that, I suggest you add in a channel dimension as well (such as the individual search engine). The setup and customization task look like the below screen shot as you create the custom report:

3.

See, now we are talking! (looking at the custom report result below) This is exciting, ..and provides almost instant insight. Just go look at the report below, there is an abundance of immediate insights – and that, without doing any other magic. Beyond my point and goal of taking you towards a metric alert search phrase opportunity, I suggest that you save this report as it should be the absolute minimum information that goes along with search phrases. As you can tell – I LOVE – this custom report. Moving along though.

I’ve highlighted the third most popular search phrase, as it truthfully indicates that we might indeed get a lot of visits, but NO revenue. What does this mean? it means that something is likely to be very wrong and that we as retailers screwed up. You need to figure out why we drive this traffic, what the landing page is, why we don’t convert and so on (but thats not the task for this exercise). We might not sell this item, we might be out of stock, we might have an error on the landing page, but something is very likely to be wrong – and now we know. And as you go down the list you can see that there are other potential ERROR’s and thus revenue opportunities.

4.

The screen shot above is just fine, but in includes the process of us (the analyst) having to go hunt for the data in our row based data. So I suggest that we apply a simple set of filters.

- show only those search phrases that drove traffic, but created $0 revenue
- show only search phrases that drove above 100 visits*

* the reason for this filter; is that we will have a long tail of very unique search phrases that we get a few visits from and no revenue, which is acceptable. Search phrases like that, are some we probably have to analyze in bulk, but that’s for a later debate.

The result of applying these filters (and how they look) is presented below. Which clearly shows a long list of search phrases that I need to dig into. You could surely, and probably should, bookmark this report and re-visit it once in a while as you do your weekly analysis schedule.

5.

Finally; to the actual suggestion. Alerts! I suggest that you apply an alert on Product Views (you could do this on visits as well). In YWA this works simply by right-clicking the data point in the column you would like to monitor. The reason for me using Product Views is that there can be a very valid set of search phrases that you know for sure are non-converters. I do not want to be alerted about those search phrases, as that is just data noise.

Elaborating a bit, looking at the above screen shot, you see a term for this hardware tools web site that says “ jigsaw puzzles”. This is obviously a set of visitors looking for something completely different and this retailer just happens to rank well for the phrase jigsaw – and thus get some ranking for some of these playful terms as well. As we can also see, people do not go look at the product pages, which is OK and expected (This attitude might be flawed in some instances, dependent on your tracking code setup and other issues, as such, you are more than welcome to set this up on the visit metric).

When I use the Product Views metric as the alerting metric to monitor, I am essentially just applying a simple filter, about NOT sending alerts out on search phrases with less that 10 product views. But as I am sure you thought through, I could simply have applied this as a filter in my above reporting filtering.

What you can see in my Edit Alert box below, is that I have used the following monitoring method; to Monitor each and every item individually, as every search phrase is important to me (*I added a small appendix section below describing what metric alert type to use). I’ve set it up to check the last full 30 days, once a month, which is of course something you will change to match your analysis routine and traffic volume – and finally set the Alert up so that it triggers on Product Views above 10 (*so I don’t get my jigsaw puzzles alerts).

I added my email and a small description (don’t discard naming and description as something you shouldn’t think about, because then you would be wrong – but that’s for another post as well).

**Just as a side note, you need to bookmark the above report in YWA for alerts to function on it. I assume it is pretty much the same in other tools.

6.

There you go! The conclusion and result of our endeavors. Once a month a list of search phrases, which you MUST look into, is delivered to your inbox. This is highly actionable and I would find it hard to believe that you would just accept a list of 50 something search phrases sending a lot of traffic to your site and generating exactly $0 revenue for you. I would personally throw up in my keyboard every month on the 1st at 13:00 EST. Does that make me bulimic – or just plain nerdy!? :-)

A small disclaimer; all of the above screen-shots are from Yahoo! Web Analytics, but you have the same metric available in e.g. Omniture Site Catalyst and you can go do the same thing there.

AND if you want to know a bit more about how you customize report and work with alerts, I have a good section about this in my Data Driven Insights with Yahoo! Web Analytics Book.

Cheers :-)
/ Dennis Mortensen (@dennismortensen)

Appendix: What Metric Alert type to use:

Average – You can monitor the average of the metric for the specified column. Suppose you allow marketing to do any campaign they want, as long as the overall visit-to-sale conversion from campaigns is above a certain level. Setting up an alert that monitors the average will look at a total for all your campaigns and you will not be alerted if a single campaign performs badly, as long as the overall conversion rate is where it is supposed to be.

Each Item Individually – Choose this option if you want to ensure that every metric in that column is measured against the triggers. Suppose you want to monitor the visit-to-sale conversion rate on all referring visits from several campaigns. You might have hundreds of marketing campaigns running at any given point in time and they are always changing. Setting up an alert is a way to ensure that you are instantly warned about campaigns going below a given negative threshold, even as marketing is setting up new campaigns and deleting old ones.

One Unique Item -  This is the most specific of all the alerts: it lets you monitor the metrics that you clicked against the triggers on that one item only. Suppose you engage in a narrow SEO activity, which is intended to drive up your visits from AOL Search. If the promised output and closure of the project is expected to occur at a visit number above 2,500, you can create the alert shown in Figure 7.2 and wait for an email to arrive. Your situation will determine which option you choose. In general, I have observed that the Average alerts tend to go to managers, who might not even log into the tool but rely solely on scheduled emails and alerts. Each Item Individually is somewhat the same as setting up 20 alerts on each unique item, and thus those two options tend to go together and are mostly used on an operational level—for example, if you are a Search Engine Marketing (SEM) or SEO manager or if you in charge of specific website optimization tasks.