some of my keywords are coming in at about $3.50 lower than my flat rate bid....
should i manually adjust each of those keywords to have a lower CPC, closer to what google says im paying...
or leave them at a couple dollars higher than the paying CPC?
i think that made sense.
At any rate, Adwords will only let you pay exactly what you need to beat the guy below you, so having much higher bid than what you're actually paying only means there's a buffer (which means that if he raises his bid close to matching yours you will pay the full price). If you lower your bid closer to what G says you're paying you won't be paying less, but you may be outbid you if he raises his bid slightly (but your bids won't get jacked up if he raises his on the other hand).
So IMO your bid should always be what you're prepared to pay for a click. If the actual CPC is lower, then great - it means you'll hold on to your position even if your competitors raise their bids (to an extent, of course)
im concerned i would lose position by adjusting all of my CPC bids...
i know there are many factors in the overall QS, which i wasnt sure if adjust my bids was one of them, BASED on what i am bidding -vs- what i am paying...
ouch my head hurts now.
The ranking you talk about is calculated on each ad and keyword. So changing the CPC on a keyword affects only that keyword and whatever ad(s) is running and being served.
Your ranking is determined by your bid multiplied by your QS. Assuming 100% relevant keyword-to-ad-to-LP, your QS is effectively your CTR.
Let's say you are A1 (advertiser 1) with A2 below you with the following data (I'm making these numbers up as I go):
Advertiser Bid CTR
A1 $0.80 5.47
A2 $0.50 8.53
Your adrank is 80x5.47 = 437.6, his is 426.5. This means he only needs to get 9 points and a bit to pass you. He could achieve that by increasing his CTR to 8.76%. Or he could do it by increasing his bid to just 52 cents.
What you pay is calculated as his adrank times his bid divided by your adrank. That's $0.49 that you would pay using the above numbers. However, if you increase your adrank, either by bidding more or ad with higher CTR, your CPC would go down. You also make it more difficult for him to outrank you.
If you decreased it, your adrank changes, you end up in a lower position and you pay whatever adrank and bid of the new advertiser below you is. Same if someone below you increases their adrank.
And of course, bids change. CTRs especially change with each search. That probably makes you head hurt more now...
so adjusting the CPC manually for each KW, would NOT have an affect on my overall ranking?
If you mean in the case where your bid is way higher than the actual cost, then lowering the bid closer to the actual cpc will not *in theory* affect you overall rankings. But since bids and competitors constantly change, manually trying to match your bid with the actual cpcs soon become overwhelming and you need to constantly watch each bid to make sure someone haven't outbid you.
This is why it's way easier to simply decide what you're willing to pay per click at the most and bid for it - the auction system handles the rest and adjusts your price so you'll pay a lower cost if it's possible. Manually adjusting the bids is just not necessary, the system was built to do that for you so manually trying to match your bids with the observed cpc is redundant - your time is much better spent elsewhere.
Also, Google does deviate from this methodology for the #1 position in the auction. They charge an added premium for that spot so it is possible to pay more than you need to for the #1 position.
Google does deviate from this methodology for the #1 position in the auction. They charge an added premium for that spot
Do you have a source for this? I personally strongly doubt this, but I don't mind being proven wrong
[edited by: DiscoStu at 10:35 pm (utc) on Oct. 2, 2009]
Advertiser Bid CTR
A1 $0.80 5.47
A2 $0.50 8.53Your adrank is 80x5.47 = 437.6, his is 426.5. This means he only needs to get 9 points and a bit to pass you. He could achieve that by increasing his CTR to 8.76%. Or he could do it by increasing his bid to just 52 cents.
LucidSW, Adrank is calculated by Bid*QS, not Bid*CTR. CTR plays a large role in determining QS, together with other factors, but G has never specified exactly how CTR affects QS, so I don't believe there's any basis for the formula you're using
Like Stu, I don't believe that there's anything special about the #1 spot. No premiums added, so I'd like to see a source for that too myself.
Stu, I know the actual formula uses the QS and not the CTR. I used this as an example but the principle is the same. As you say, CTR plays the largest role in calculating QS, between 60 and 65% according to experts. The rest is keyword-ad-LP relevancy and a small percentage, about 10%, for other landing page issues, such as loading time. Assuming you have a perfect score in the relevancy portion (which everyone should have) and decent page loading time, what's left is CTR.
QS is a relative number calculated using historical CTR data of all advertisers for that keyword. It is also normalized by position. It very likely is a standard deviation calculation and the QS reflects how far away (how many deviations) from the mean your CTR is. It means it is entirely possible to have all advertisers fall within a QS range of say between 5 and 8 with nobody at 9 or 10.
Also, regarding paying a premium for the #1 spot, that was told to me by a former AW Rep of mine when I questioned if Google was as altruistic as their CPC model suggests. I know it's anecdotal, but I have reason to believe it's accurate based off of my experiences. I've run tests gradually lowering the bids of keywords with a strong #1 rank and have had success with achieving a lower CPC while maintaining the #1 position. I've also had numerous experiences where certain keywords jump in rank to the top spot during the weekend as other advertisers employ day parting. I almost always see an associated CPC lift in these cases which leads me back to the added premium for the #1 position. Obviously, there's a ton of factors at play and nothing can be proven but I'm inclined to believe what I was told based off of the aptitude of that particular Rep.
You know, watching it again for first time in a while, you're right. The formula he uses, he divides by your QS, not your adrank. He doesn't mention plus one penny but yes, using that formula A1 in my example would pay 78 or 79 cents.
Stu, I know the actual formula uses the QS and not the CTR. I used this as an example but the principle is the same.
Forgive me, sir, I didn't notice you said "all else being equal". I stand corrected
