AdWords conversion tracking -- am I missing out?

[Tonearm]

I track conversions and calculate ROI outside of AdWords by appending unique parameters to the landing page URL of my ads. I've never paid attention to Google's conversion tracking system because of this. Does it provide any interesting stuff that I would be unable to do via my method?

[RhinoFish]

Conversion Optimizer; Segmenting all sorts of reports to the Nth degree; Auto-tagging and auto-synching sales data freeing up your time; Better insights in Content Experiment A/B tests; and more...

what do you see as the advantages to external?

[Tonearm]

One advantage to calculating ROI externally is the calculation becomes more precise because it can be recalculated after the order has shipped to use the real shipping cost, it can be recalculated again if a customer returns something, etc. I think Google's requires you to know the final numbers on the order receipt page?

You mention freeing up time, but my external system requires no user intervention whatsoever. I actually don't use the AdWords reports or A/B tests at all. I know that's not the conventional way of running AdWords, and I surely miss out on $$$ because of it, but automation is a priority for me.

Does handing the keys to Google with the Conversion Optimizer really work out?

[RhinoFish]

If the PPC is built well, the CO is wonderful. If your PPC isn't good, it's like putting cruise control on a car with a flat tire that's firing on 3 cylinders.

The CO frees an enormous amount of time regarding keyword level bidding, and, in addition, the efficiencies of that bidding for results automation FAR outweigh your techniques. Yes, returns and adjustments are an issue for the feedback loop, but they are so minor compared to the gains it's silly.

Caveat: many disagree with me, and say the CO doesn't work for them. I assert the underlying PPC has issues, but there can also be business issues I suppose, like a high return rate, that also makes the CO less optimal. The CO needs time to "learn", suggest you try it on one campaign, see what comes, for you.

[Tonearm]

the efficiencies of that bidding for results automation FAR outweigh your techniques

What do you mean here? I would think both my technique and the CO go through a similar calculation and feedback loop. There is some lag time with mine, though, since there are too many keywords to process all of them each day.

[particleman]

I was in a similar situation as you. Google's conversion tracking is pretty easy to implement so I say do it. It allows you do all of these helpful features down the road if you want. I also tagged urls and kept my own data. This was important for me to validate conversions using direct data, not rely on google. In my experience the google data is pretty accurate, but on average I had a variation of around 1% comparing google's conversion data to mine.

[RhinoFish]

What do you mean here? I would think both my technique and the CO go through a similar calculation and feedback loop. There is some lag time with mine, though, since there are too many keywords to process all of them each day.

How many factors are you weighing BESIDES the keyword's bid/CPA/ROAS/conversion rate - these are really 1 thing measured, how many other factors are you looking at?

How many do you think G is looking at? :-)

You're missing out of what G knows about a visitor...

The CO is not a keyword bid optimizer.

[Tonearm]

Now this is getting interesting. Can you give me an example of what Google is doing with their CO that I can't?

[RhinoFish]

1) see list of factors here:
[google.com...]

2) note that they say the list linked above lists "some" of the factors they use.

[Tonearm]

I don't bid on the content network so the only stuff listed there that I can't control without CO are operating system, browser, search partner site (is this still true?), and particular broad-match query.

If Google does a good job splitting up broad-match queries, somebody should be able to define only a relatively small set of broad match 1-word or 2-word phrases and CO will take care of the rest. Does that work?

Letting Google determine my bids is also kinda scary. Do they keep your best interests in mind?

[RhinoFish]

Look at your Search Terms Report... they know what search came in, your tactic doesn't.

"If Google does a good job splitting up broad-match queries, somebody should be able to define only a relatively small set of broad match 1-word or 2-word phrases and CO will take care of the rest. Does that work?"
Broad match 1-word... hahhaa.
No, this tactic won't work.
DKI and other reasons means you want to have segmentation by keyword.
I understand what you're asking, but no, that wouldn't be wise.
CO or no CO, the whole effort is always to focus your money where it matters most, so you'd never skip your prime exact match terms.
And you're imagining that, given enough time, the CO would work things out well for broad keywords - but the amount of garbage coming in on broad match will weigh down the outcome so heavily, it won't work.
If you engine is broken but runs all sputtery, setting the cruise control to 50 mph... what happens? A: bad things.
if the whole ad group can't meet the CPA, it'll turn it all off (eventually).
Think of it this way - each keyword is an adjustable gas pedal. the more good gas pedals you add, the better the auto optimizer can figure out how far to press each one.
if you slap one big ass (broad match) gas pedal in there, it's not going to work well.
The CO cannot overcome sloppy, loose PPC.
But hey, i do like your optimism.

"Do they keep your best interests in mind?"
Of course not.
In Googleland, they primarily concern themselves with their users (the searchers) - they know if they do them right, G itself will do very well, and many advertisers will want to come and spend.
Not knocking G, but you and me, we're last in line at the G feeding trough.
What G also knows is this... you will spend more if you make more sales.
So no, your interests are not a prime mover here.
Again, love your optimism!
If their CO doesn't work for you, you'll spend less...
Why would they do that?

If "scared" is the right adjective, explore it, if you can't shake it, then don't use the CO, it's not for you, and nothing I say matters. Scared means you'll turn it off a few hours later - that serves no purpose at all.

I think I'm trying to hard to talk you into to trying it...
:-)

There are many of my peers and competitors who don't use it, so know this isn't a "must do". To me it is, but I'm not myopic. Many very good PPC folks never use it.

Have fun!

[Tonearm]

Thank you for taking time with me here RhinoFish. I greatly appreciate it and now I have a much clearer picture of the CO option.

[RhinoFish]

Have fun! I wish you great success no matter which path you take to reach it!

[p5gal5]

I think it's advantageous, not necessarily from an Adwords perspective, but for (as RhinoFish mentioned) Content Experiments. We've run many successful experiments - some with higher conversion rates, some with higher average transactions, no discernable difference in conversation/avt but a substantial one in bounce, etc etc etc.

My most recent test yielded no statistically significant difference in conversion rate (very slight decrease from the original), but an increase of 31.4% AVT and +32.4% value per visitor. Depending on your traffic mix, you can also break out data by new/returning users. I've found it to be invaluable, but YMMV. If your current solution does all this, it might be a moot point, but I've found it to be extremely helpful.

[RhinoFish]

During mini-trends (like certain post-xmas buying), we see the CO move with the trend, chasing volume when it is converting, while we're drunk on rummed up egg nog. Top that!

[p5gal5]

That's ok; I'm not a topper. Just a helper. :)

[Tonearm]

I think it's advantageous, not necessarily from an Adwords perspective, but for (as RhinoFish mentioned) Content Experiments. We've run many successful experiments - some with higher conversion rates, some with higher average transactions, no discernable difference in conversation/avt but a substantial one in bounce, etc etc etc.

My most recent test yielded no statistically significant difference in conversion rate (very slight decrease from the original), but an increase of 31.4% AVT and +32.4% value per visitor. Depending on your traffic mix, you can also break out data by new/returning users. I've found it to be invaluable, but YMMV. If your current solution does all this, it might be a moot point, but I've found it to be extremely helpful.

I have an internal A/B split testing setup for that and I agree it's really helpful. It's tied into profit which is updated after shipment based on shipping cost which I think gives it a leg up on the CO in this context.

During mini-trends (like certain post-xmas buying), we see the CO move with the trend, chasing volume when it is converting, while we're drunk on rummed up egg nog. Top that!

Mine does the same of course, but the CO is faster because there are too many keywords for my script to process all of them in one day so I think you've got me slightly topped in that aspect.