|SEO: Closed-loop ROI tracking|
How to track SEO ROI in a completely closed-loop fashion.
| 3:31 pm on Jul 11, 2002 (gmt 0)|
I am wondering what specific techniques, if any, are being employed by SEO professionals on this board to track SEO ROI in a *completely* closed-loop fashion. I do SEO work professionally, and my current method for measuring campaign ROI involves analyzing log files to determine whether, and what amount of traffic is coming in on the keywords that I optimized the site for, and correlating this data roughly with independent sales data (from the client). This way I can say, for example, "search referals increased by 20% and so did sales in July."
But, now, since many of my clients have become accustomed to this from doing PPC campaigns on Overture, using tracking URLs, I am being asked to track ROI much more tightly, which would necesitate me tracking all the way from the search engine referal to the conversion. In other words, my clients want me to tell them, for instance, Google delivered 1000 uniques in July for some keyword, say, "ailrines pilot" and that of those uniques, 50 converted to purchasing a product (i.e. made it to a conversion landing page).
I find that it would be difficult to provide such data. And, as far as I can see WebTrends (which is what I use) doesn't faciliate this.
What can I do to execute this kind of closed-loop tracking?
Anything you can say to help me out would be great.
Thanks in advance,
[edited by: frostlorraine at 3:50 pm (utc) on July 11, 2002]
| 3:44 pm on Jul 11, 2002 (gmt 0)|
If you can change your client's pages, I would add a cookie or a session ID, which would follow users from their entry page (including referrer and search phrase) to the conversion page. At that point, the cookie info could then be saved with the sale data. Then you just have to compare that to your webtrends stats.
<added> and btw, welcome to WMW (finally got to write that :) )
| 3:53 pm on Jul 11, 2002 (gmt 0)|
That is certainly a feasible solution. Do you know of any off the shelf tools that can be employed on hosted basis that would allow me to achieve this effect. Or, do you believe that a custom solution needs to be built (which shouldn;t be impossible)?
| 4:00 pm on Jul 11, 2002 (gmt 0)|
I don't know of any tool doing that (would like to, though, so if anyone knows such a tool...).
I would go for custom.
| 8:41 am on Jul 12, 2002 (gmt 0)|
We do exactly what was mentioned and it's really quite simple. Not only do I provide my client a monthly report showing his 'ranking' in major search enignes (Web Position Ranking or Top Dog), but we review Top 20 Search Keywords by Server and THEN...we track the referer in all email correspondence from the site.
There's not one single email address on the site(s). Instead we point everything to contact forms w/ basic information (name, city/state, email, comments, etc.). When the email is submitted, we have it send 2 emails: one to the client w/ all the info and another to me w/ basic info (city/state, comments, and REFERRER (with keywords).
Here's how it's important. A client does home remodeling in Atlanta, Ga. We kick butt (let's say) for "basement remodeling" in SE's. But we find out through tracking the referrer that actually 70% of his contacts searched for "atlanta remodeling" which doesn't come up our Top 10 keywords because the "broader" or "national" keywords get more "traffic" but less qualified leads.
This helps tremendously when it comes to paid sponserships, for example, in Yahoo!. Now I'm not paying for "home remodeling" (which would drive traffic) but "atlanta home remodeling" that generates real contacts.
Also, many of our clients have dedicated 800 or local numbers so we can track phone leads from the site. It's sooo cheap to get a dedicated toll free number that we encourage most of commercial clients to do so.
Hope this helps!
| 4:36 pm on Jul 12, 2002 (gmt 0)|
You are doing exactly what we want to do, and for exactly the same reason we want to do it. We need to know which terms we *really* need to optimize for. You start with a theory, so to speak, when initially selecting keywords to optimize for, but need to validate this theory through data. Otherwise, you're simply expending energies on optimization that produces no or little value. So, you and I certainly agree on the need and importance. That said, can you let me in on *how* you went about actually implementing the tracking? And, do you know of any *off-the-shelf* solutions to enable this?
| 9:09 pm on Jul 12, 2002 (gmt 0)|
There are a bunch of different ways to do what you're talking about. I'm *sure* there is software/freeware/open source code you can use but we have never researched it. The way *we* do it on *some* of our sites is this (there are much more complex tracking methods as well that can tell you w/ a Session ID writing to a cookie how many times a user has visited a site before contact/sale):
THIS ASP CODE SETS THE COOKIE. IT MUST GO ABOVE EVERY OTHER LINE OF CODE ON YOUR PAGE - ABOVE THE <HEAD> TAG. Change "SITENAME" to your site name (something totally unique).
if request.cookies("SITENAME")="" then
vRef = Request.ServerVariables("HTTP_REFERER")
response.cookies("SITENAME")("Referer") = vRef
response.cookies("SITENAME").expires = now + 720
ON YOUR EMAIL FORM OR SHOPPING CART FORM YOU NEED TO CAPTURE THE REFERRER FROM THE COOKIE AND PUT IT IN A HIDDEN FIELD:
<input type="hidden" name="Referer" value="<%=Server.HTMLEncode(request.cookies("SITENAME")("Referer"))%>">
WHEN USING AN ASP SENDMAIL FORM LIKE WE USE, YOU SIMPLY REQUEST THE 'REFERRER' AS YOU WOULD ANY OTHER FIELD FROM THE FORM.
vReferer = request.form("Referer")
WE HAVE THE FORM SEND TWO SEPARATE EMAILS, ONE W/ ALL INFO *EXCEPT* THE REFERRER TO THE CLIENT AND ONE TO US W/O PERSONAL INFO AND *WITH* THE REFERRER.
I hope this helps!
| 2:48 am on Jul 14, 2002 (gmt 0)|
This is very helpful indeed. Thank you so much!