55+ Google Analytics Custom Alerts – The Check Engine Light For Your Data
Last month, Phil wrote a blog about five features of Google Analytics that you probably are not using, but there obviously are more than five, and here’s another one that you probably aren’t using, but you should. In fact, I think this is probably one of the KEY things you should do when setting up an account and a new website, and it’s my bet that the vast majority of people don’t do it at all.
Custom Alerts
Why use Custom Alerts? Because you don’t check your analytics every day. OK well some of you data geeks that read this blog do, and I do, but most people who are just regular people, they don’t. That guy who is wearing 20 hats, he doesn’t have time to go over his data every morning for an hour or two. You could argue he should, but maybe he doesn’t. Custom Alerts can let him know, in general, if there is something that needs his or her attention. Is something significant, or at least possibly significant happening? Should someone take a look? it’s like a check oil or check engine light on your car. If the light doesn’t come on you probably (I hope) change the oil on your car regularly, but if something is going wrong with your oil, then you sure are glad that light is coming on to let you know to check it.
That’s what custom alerts do. They just give you a text or an email saying “hey buddy, something is going down, you need to check it out.” And that’s often enough to get a marketer in FRONT of a problem, rather than get months down the road and realize that all the data on the site is corrupt, or it wasn’t tracking data for a month, or the ecommerce numbers tanked. But it’s not just BAD things. It’s letting you know when your site is getting a spike in social media referrals, or a boost in non-branded organic search traffic.
Now, like my like my recent 20 Ways to Use Google Analytics Custom Variables post, this isn’t a “how to” post covering exactly how to set up specific custom alerts. There’s plenty of tutorials around the web. This is about suggesting specific alerts for you to use on your site. My recommendation is to set up basically all of these, with the possible exception of the ecommerce ones if you don’t participate in ecommerce. While intelligence alerts are great, are you using them? Are you checking your data constantly, daily, or weekly, for problems, or weird spikes or drops? Custom Alerts can let you know when things change as soon as it possibly can by email or even text. Some things you don’t want to wait for a week or two to find out that “oops, we haven’t been tracking data for the last 2 weeks.”.
One other note before I start listing them, I mention various percentages like 10 or 20% for increases or decreases. This is a ballpark, and for each site it will be different. The key is to tailor your site so that these alerts don’t fire for your normal traffic fluctuations, but only send out their alerts when something interesting might be happening. Set them up with a certain percentage, and if they seem to fire all the time, then try bumping the numbers till they make sense. I’ll lay out the general area of the alerts, what it’s covering, and then the basic configuration for each alert.
So let’s jump into it.
1. No Data (daily)
If we could set this as an hourly alert we would. This will let you know if you had no visits on a particular day, a great way to let you know that something is wrong with the tracking on your site (or maybe your site itself is down).
all traffic -> Visits -> is less than -> 1
2 through 13. Site Usage
Get a general idea for big changes in your site compared to the previous week, either up or down. Spikes of visits or visitors could indicate new referral traffic, or even some sort of media mention, whereas a traffic drop could indicate some other sorts of problems, such as maybe a LACK of referral traffic, or a problem with your PPC ads. Bounce Rates increasing could indicate a content problem on your site, and so on. The idea from these alerts is that your’e seeing some sort of general and significant change to the use of the site, which might warrant a closer look at what’s going on.
Visits
all traffic -> Visits -> % decreases by more than -> 10% -> same day in the previous week
all traffic -> Visits -> % increases by more than -> 10% -> same day in the previous week
Visitors
all traffic -> Visitors -> % decreases by more than -> 10% -> same day in the previous week
all traffic -> Visitors -> % increases by more than -> 10% -> same day in the previous week
Pageviews
all traffic -> Pageviews -> % decreases by more than -> 10% -> same day in the previous week
all traffic -> Pageviews -> % increases by more than -> 10% -> same day in the previous week
Bounce Rate
all traffic -> Bounce Rate -> % decreases by more than -> 10% -> same day in the previous week
all traffic -> Bounce Rate -> % increases by more than -> 10% -> same day in the previous week
Avg. Visit Duration
all traffic -> Avg. Visit Duration -> % decreases by more than -> 20% -> same day in the previous week
all traffic -> Avg. Visit Duration -> % increases by more than -> 20% -> same day in the previous week
% New Visits
all traffic -> % New Visits -> % decreases by more than -> 20% -> same day in the previous week
all traffic -> % New Visits -> % increases by more than -> 20% -> same day in the previous week
14 through 17 Goals
In reality these two should be duplicated across all your goals. Track big swings in your goal conversions, decreases and increases. If suddenly a goal conversion drops sharply, there could be reasons for it. Maybe a weird traffic spike of users who mostly bounce from your site is affecting the rate, maybe there is a technical problem with the form. But Goals are what you in some ways care most about on your site, so if you aren’t converting them, you’re going to want to know. I’m alerting here on a daily, as well as weekly basis to help even it out.
Goal 1 – Conversion Rate
all traffic -> Goal 1 – Conversion Rate -> % decreases by more than -> 20% -> same day in the previous week
all traffic -> Goal 1 – Conversion Rate -> % increases by more than -> 20% -> same day in the previous week
Goal 1 – Conversion Rate Decrease (weekly)
all traffic -> Goal 1 – Conversion Rate -> % decreases by more than -> 20% -> previous week
all traffic -> Goal 1 – Conversion Rate -> % increases by more than -> 20% -> previous week
18 through 25 Ecommerce
These are pretty much the simplest of ecommerce alerts. Increases and decreases of revenue, transactions, average value, and the overall conversion rate. The smaller the amount of ecommerce you do, the higher your percentage is probably going to need to be, but if you process a good number of transactions, then you’ll want to be alerted with a significant change in your ecommerce, either to take advantage of some sort of positive consumer movement, or to stem the tide from a hit to your transactions.
Revenue
all traffic -> Revenue -> % decreases by more than -> 10% -> previous week
all traffic -> Revenue -> % increases by more than -> 10% -> previous week
Transactions
all traffic -> Transactions -> % decreases by more than -> 10% -> previous week
all traffic -> Transactions -> % increases by more than -> 10% -> previous week
Average Value
all traffic -> Average Value -> % decreases by more than -> 10% -> previous week
all traffic -> Average Value -> % increases by more than -> 10% -> previous week
Ecommerce Conversion Rate
all traffic -> Ecommerce Conversion Rate -> % decreases by more than -> 10% -> previous week
all traffic -> Ecommerce Conversion Rate -> % increases by more than -> 10% -> previous week
26 through 30. Content
Just doing a few things here. Are your visits using site search decreasing, or their effectiveness decreasing? Are the exits of people from a failed search increasing? If you have a lot of people using your site search then you’re going to want to know when suddenly it’s not working as well as it should, and why. I’m tracking the total events here, but specific event tracking and increases and decreases would also be useful if your’e using them. Also watching for an increase in Average Page Load Time, just in case you have some server speed slow downs, it could be a red flag for you to watch out for.
Visits with Search Decrease (week)
all traffic -> Visits with Search -> % decreases by more than -> 10% -> previous week
Time After Search Decrease (week)
all traffic -> Time After Search -> % decreases by more than -> 10% -> previous week
Search Exits Increase (week)
all traffic -> Search Exits -> % increases by more than -> 10% -> previous week
Total Events (month)
all traffic -> Total Events -> % decreases by more than -> 10% -> previous month
Avg. Page Load Time
all traffic -> Avg. Page Load Time -> % increases by more than -> 10% -> previous month
31 through 42. Clicks (Adwords)
If you’re running adwords you’re going to want to know fluctuations in what’s going on. Are impressions changing, clicks, cost, etc? Also importantly are your (not set) visits increasing indicating an issue with your ads? The fluctuation here can be higher, but again, set it to make sense for your usage. However with so many businesses now focusing on Paid Ads, you want to be aware if there’s a problem.
Impressions
all traffic -> Impressions -> % increases by more than -> 20% -> previous week
all traffic -> Impressions -> % decreases by more than -> 20% -> previous week
Clicks
all traffic -> Clicks -> % increases by more than -> 20% -> previous week
all traffic -> Clicks -> % decreases by more than -> 20% -> previous week
Cost
all traffic -> Cost -> % increases by more than -> 20% -> previous week
all traffic -> Cost -> % decreases by more than -> 20% -> previous week
CTR
all traffic -> CTR -> % increases by more than -> 20% -> previous week
all traffic -> CTR -> % decreases by more than -> 20% -> previous week
CPC
all traffic -> CPC -> % increases by more than -> 20% -> previous week
all traffic -> CPC -> % decreases by more than -> 20% -> previous week
Keyword (not set)
keyword -> matches exactly -> (not set) -> Visits -> % increases by more than -> 5% -> previous week
43 Social Media Spike (daily)
Well we’re only counting this as one alert here, but it could be many. you could track the source traffic JUST from Facebook, or twitter, or your blog. Maybe you want to know if your Facebook Referrals are increasing dramatically so you can jump on whatever is happening as quickly as possible. Here I’m setting up a rough regular expression matching a bunch of social sources, and creating a custom advanced segment with them. Then we apply that segment, and watch for a spike of social visits.
Social Sources Custom Advanced Segment with Source matching regex of
facebook|twitter|t\.co|blogger|stumbleupon|naver|wordpress|hootsuite|youtube|linkedin|
answers\.yahoo|delicious|myspace|reddit|livejournal|flickr|typepad|vkontakte|netvibes|
ning|orkut|tumblr|diigo|fc2|nowpublic|buzznet|livemocha|weebly|ameba|gooblog|habbo|
okwave|oshiete\.goo|deviantART|instapaper|scribd|sharethis|xanga|answerbag|brizzly|
cyworld|multiply|ow\.ly|squidoo|studivz|wikia|wiserearth|associatedcontent|blogsome|
groups\.google|sportsnaviplus|taringa!|tuenti|wikio|babygaga|bebo|bitly|bloglines|care2|
circleofmoms|cocolog|couchsurfing|digg|douban|gather|hatena|ibibo|is\.gd|mixi|paper\.li|
photobucket|plurk|renren|skyrock|smallworlds|sonico|ustream|vimeo|weeworld|wikianswers|
XING|you?ku
Now those aren’t all, and you might not want to include some. Add or remove the social sources you want to that regular expression.
Applies to Social Sources Segment -> Visits -> greater than -> 250
44 through 55. Traffic
Lastly we’re going to watch our general traffic, both for internal site tracking errors such as self referrals indicating pages on your site that don’t have the tracking code installed, to changes in your organic traffic, or your non-branded organic traffic, or your BRANDED organic traffic. You might want to know if the traffic to your site with people looking for your company name has spiked. Maybe there is something going on in the news you need to get on top of with PR. Also a check for Non-Domain traffic indicating people maybe stealing your site content. A variety of alerts to indicate issues with your traffic sources, good or bad.
Self Referrals Occuring (daily)
source -> contains -> yourcompanywebsite.com -> Visits -> is greater than -> 1
Organic Traffic change
Medium -> matches exactly -> organic -> Visits -> % increases by more than -> 10% -> same day in the previous week
Medium -> matches exactly -> organic -> Visits -> % decreases by more than -> 10% -> same day in the previous week
Non-Branded Organic Traffic
exclude keyword matching regex -> your company name|companyname
include medium containing organic
Applies to Nonbranded Organic segment -> Visits -> % increases by more than -> 10% -> same day in the previous week
Applies to Nonbranded Organic segment -> Visits -> % decreases by more than -> 10% -> same day in the previous week
Branded Organic Traffic
include keyword matching regex -> your company name|companyname
include medium containing organic
Applies to Nonbranded Organic segment -> Visits -> % increases by more than -> 10% -> same day in the previous week
Applies to Nonbranded Organic segment -> Visits -> % decreases by more than -> 10% -> same day in the previous week
Non-Domain traffic (daily)
Nondomain Custom Advanced Segment with Hostname matching regex of
(not set)|domainname.com|example.com|etc
Applies to Nondomain Segment -> Visits -> greater than -> 40
Direct Traffic change
Source -> matches exactly -> (direct) -> Visits -> % increases by more than -> 10% -> same day in the previous week
Source -> matches exactly -> (direct) -> Visits -> % decreases by more than -> 10% -> same day in the previous week
Referral Traffic change
Medium -> matches exactly -> referral -> Visits -> % increases by more than -> 10% -> same day in the previous week
Medium -> matches exactly -> referral -> Visits -> % decreases by more than -> 10% -> same day in the previous week
—
So That’s 55+ Ways to Use Google Analytics Custom Alerts. Don’t let those limit you though. Think about the things you look for, or would look for, or might expect, that aren’t covered here and create an alert for that as well. Then even if you forget to check out your analytics on a daily basis, Google Analytics itself will still give you a poke if something interesting is happening.
About Sayf Sharif
Sayf Sharif is a Web Analyst, and expert in Usability and UX, who has worked with businesses large and small to maximize their online presence since the beginning of the Web, winning numerous awards along the way. Sayf has studied human tool use from the stone age (he went to graduate school for Archaeology) to the information age (he started programing on his father’s TRS-80), and is always interested in what goals people wish to accomplish using their tools, and how successful that experience was.
Traffic
Analysis
Action
Subscribe to this blog's feed
Hi Sayf! Thanks for some great ideas. For some time now I’ve been thinking how to set up an alert for checkout abandonment. Any ideas? Maybe with events?
Hmm. Maybe you could set a visitor level custom variable when someone adds something to their cart, or has something in their cart.
Then create an advanced segment that only has people who have a value in that custom variable, so you’re only looking at people with items in their shopping cart.
Then have a Goal on your checkout completion.
Then if you create an alert that is just looking at that segment, and alerting when that goal decreases or increases, you have an idea on just checkout abandonment, rather than overall conversion rate on the site.
Thanks Sayf, great post. It’s worth looking at your standard deviation for the aforementioned metrics/segments to help determine what % change to choose. You’ll end up tweaking the rules to death if you don’t start there.
Although, simple WoW and MoM % metrics can miss the mark if there is a sustained % change. Nevertheless, a great start for most marketers unable to dig in regularly. Thanks for sharing.
John, spot on and said better than I did. You absolutely should go by your own standard deviation to start, otherwise you’re just going to get hit by tons of alerts every day that then just become noise.
Also I agree that WoW type metrics can miss trends and sustained change, but HOPEFULLY people are checking in and not completely forgetting things. A weekly, or even monthly visit to the actual analytics will hopefully indicate general sustain trends to people looking at their data, while these are more to help indicate either good or bad things, that may need immediate attention in the data.
This is a fantastic list. We “SEO’s” tend to focus in on organic traffic and often loose site of what is happening with the other segments. Setting alters for direct and referral traffic is a quick way to monitor them.
Also, watching conversion rates is huge when dealing with large e-commerce sites. In addition to weekly updates I like to set them for monthly since weekly can get a little to granular at times depending on seasonality and other factors. Great staring points though.
Thanks TJ. Yeah depending on the site you want to adjust the comparison. Each site needs to be tweaked once they’re set up so that the alerts fit each sites individual needs.
nice list here…well done Sayf! will pass a few of these along!
Jim
Thanks Jim, glad you liked it.
Really nice list here, best one so far online, keyword I used to find this site “analytics Custom Alerts”
Just got sent some script which I think is useful to tack on.
First adding some javascript to the page…
window.onerror = function(message, file, line) {
var sFormattedMessage = ‘[' + file + ' (' + line + ')] ‘ + message;
_gaq.push(['_trackEvent', 'Errors', 'Browser', sFormattedMessage, null, true]);
}
Then setting up an alert to track the javascript errors on the site. So you can get a custom alert from javascript errors. There will probably always be a bit of noise, but it might help you discover when something goes wonky and catch it sooner rather than later.
Great post, thanks!
The Social Media referral Regex was very helpful and comprehensive. I ran into some issues with t.co though. I’d suggest specifying a line start and/or line end (i.e. ^t\.co$) to avoid including all websites that end with a “T” and use a TLD that begins with “co” — somewebsitet.com, somewebsitet.co.uk, etc.
All in all, very helpful!
Great pickup Jon! Absolutely, the more specific the possible there, the better.
Hi Sayf, Thanks for a great article. I’ve been trying to set up an alert something like ‘When any single .htm page has a daily increase of 50% pageviews’ but I cant get it to work. Any ideas? Is this possible?
Andrew,
The only thing you can do is to create a custom advanced segment, and then base your alerts off of that. Unfortunately when you filter a segment to show only say visits to single.htm it is going to show all the page hits to all pages during the sessions where someone hit single.htm.
You also can’t just have some generic one that looks at specific pages.
However, if your’e interested in a traffic change to a specific page, as long as it’s maybe a secondary or tertiary page, if you were to create a custom segment where you included visits to a page containing that page name, then you would be isolating that traffic down.
You could do this on your important pages, but in general this isn’t going to pick up on smaller fluctuations or even bigger ones across sites with thousands of products for instance.
In that case you’re probably better off setting the level of change to notify you in the alert to be finer, and then looking at the intelligence it gives you on those alerts, which will sometimes call out a specific page if it’s the cause of a change in pageviews or visits.
I feel you though, I’ve thought about how great something like this would be more than once.
Hi Sayf,
Thanks for your reply, and the info. I’m working on a large (>20,000 pages) gov site. The content is mainly information, rather than transactional and I’d love to know when a certain page or section has a huge increase in traffic, and the cause in the community for that increase. e.g. tv news report on kindergarten causes spike in education pages or new years eve causes increased traffic in noise complaints.
I’ll try your advice, thanks again!
Andrew, I definitely then recommend using custom variables to tag the specific pages for their sections. You could then base alerts off the sections.
That wouldn’t be perfect, because it would show all page hits during those sessions, not just to that one section, but if the changes are large enough they might trigger the alerts depending on our threshhold, and then give an indication of top pages causing the alert to fire.
Unfortunately there’s no perfect solution for what you want to do (there should be but there isn’t, so no you aren’t crazy).
Great article Sayf.
I started adding the alerts today and I’m looking forwards to seeing the results.
Just one thing: you’ve repeated the same paragraph for Non-Branded Organic Traffic and Branded Organic Traffic. I guess the only thing to be changed is “include keyword matching …” instead of “exclude keyword matching …”.
Thank you for sharing this.
Shams
Shams,
good catch! That’s right.