Benefit of exact match

when my broad match is very specific

[wheel]

The pro's here seem to push exact or phrase match for kewyords. Yet my take has always been to create lots of very specific keyphrases then broad match.

so I would use:
dancing blue widgets texas
and
dancing blue widget texas

both as broad match. In my mind, that ensures that I've covered somebody typing in
dancing blue widgets texas
and
dancing blue widgets antonio texas
(which the exact match would miss).

I don't believe switching to exact match would help my bid amounts or my positioning either.

So am I missing a reason as to why I should be using exact match instead of very specific phrases on broadmatch?

[skibum]

That can be a sound strategy wheel. If you get to specific and put everything on exact, its tough to get traffic. Add in a few negative keywords, free, homepage, stuff like that, to get rid of the obvious irrelevant possibilities and it would be like exact match but with just enough flexibility to capture some odd relevant combinations it would be nearly impossible to anticipate.

[howiejs]

good post. I am also struggling to see the benefits of listing all 3 per campagain - broad, exact, phrase

When I look at my stats - 1 or the 3 is always getting close to zero impressions . . .

[crankin]

The "fix" that Google suggests for campaigns that have tanked since the recent "improvements" is to go to an exact or phrase match. And yet, I consistently see my results crap out altogether whenever I change from -blue widgets- to "blue widgets" or [blue widgets].

I get MUCH better results with just plain blue widgets, so I really don't see any benefit in tighter matching via "blue widgets".

[eWhisper]

Exact match can be very useful for more generic terms when the negative keyword list for a broad or phrase match gets completely unwieldly.

Also, as many people start with a 1-2 word phrase before refining their search, often you can gather a decent conversion/CTR on very broad terms for people who don't refine their searches quickly.

If you were to use broad/phrase match with some of these generic terms, most likely you're CTR will plummet and your conversions will be terrible as your ad will show for some very untargeted searches (so will your exact ad, but at least the impressions are more controlled).

This type of bidding allows you to bid on some very broad terms will still trying to figure out visitor intent with regards to 1-2 word searches and bring in conversions/exposure for such broad terms.

A good example of this is [dental surgery]. Broadmatching this term, you'll be shown for people searching for research papers, pictures, and all sorts of total unrelated keywords.

Now, for broad match, I find it very useful when you start getting into 3-4 word combinations. You still need to put together excellent negative keyword lists, but the variations of possible searches is obviousally less for 4 word searches than 1 word searches. To continue the example, the broad matched term: pennsylvania dental surgery is much more targeted, but you don't want to try to enter every possible city in PA, so broadmatch is definately your friend here.

I'm a fan of broadmatch in many scenarios (not going to get into old vs new broadmatch - thats a different post), but you need to consider visitor intent, word variation, and usefulness of broadmatch vs exact when creating your keyword lists. Both have very good reasons for existing and being used.

[wheel]

Thanks eWhisper.

In fact, I do enter every city in Pennsylvania. And serve appropriate ads. Then I watch my logs for additional negative keywords. Most of my phrases are 3 or 4 words. I also keep a one or two word phrase section, but I tend to use this to pick up specific 3 and 4 word phrases that I've missed.

[eWhisper]

To continue the above example.

The possible searches are:
pa city dental surgery
city pa dental surgery
city dental surgery
pa dental surgery
dental surgery pa
dental surgery city
dental surgery city pa
dental surgery pa city
etc

Take 10 variations times 1000+ cities, leaves you with 10k+ keywords for just a few broad match searches.

The above keywords also only get a few impressions a month at best, so no single keyword will have enough impressions in a year to attain its own CTR (1000 impressions).

Now consider that when you first enable a new keyword, your CTR for the bidrank formula is the average of the existing advertisers for that KW, and you need your own impressions to raise your CTR to lower your CPC in the bid rank formula.

By broadmatching the keywords, you can have keyword combinations that can attain their own CTR. Thus, if you get a higher CTR than your competition, you can have a higher position and pay less for the exact same exposure.

Thats the reason for very small search terms I prefer broad match, if I can get a CTR higher than my competition, and thus lower my bids and achieve a higher position than someone who has 10k exact phrases and hasn't achieved their own CTR yet.

[howiejs]

eWhisper - thanks for the great reply.

So is it better to have one broad match
american widget
vs
lots of exact matches
"red american widget"
"cheap american widget"

[wheel]

Ewhisper's assuming that this gives him a better starting CTR - quite possibly the case. Unfortunately, you can't target your ads doing that. And IMO, targetting ads will lead to much higher CTR. An interesting point nonetheless.

Unless the idea is to start with broad match then move to more specific words?

[eWhisper]

Unless the idea is to start with broad match then move to more specific words?

If I see that a word is getting a lot of impressions, then yes, it'll be come its own match.

The above example is for words that get less than 25ish searches a month, and would literally take over a year to gather 1000 impressions.

For words with more impressions, it really depends on how various CTRs and competition works out. I've seen cases where it's more cost effective to keep broadmatch around when it could be broken into 10-100 phrase/exact matches due to how others bid - also, I've seen cases where it's better to break these words apart. Those type of decisions are case by case and I don't think you can make a general predictor that encompases every scenario.

[shaka1978]

From eWhisper:

'Now consider that when you first enable a new keyword, your CTR for the bidrank formula is the average of the existing advertisers for that KW, and you need your own impressions to raise your CTR to lower your CPC in the bid rank formula. '

But what if I bid on (eg)

'Ultra blue widgets factories'

And this is so specific that no other existing advertisers are bidding on this keyword. How does the CTR formula handle that for th e first 1000 impressions?

[edit_g]

And this is so specific that no other existing advertisers are bidding on this keyword. How does the CTR formula handle that for th e first 1000 impressions?

It decides if your ad stays on the right hand side of the search results or goes up into the premium slot.

[shaka1978]

Thanks edit_g, but I'm not quite with you.

If I'm bidding on:

'Ultra blue widgets factories'

I probably won't get the premium (or be first on the right hand side) as many people may be bidding on broad match on 'blue widgets', 'widget factory', 'widgets' etc...

So, what CTR would I be given for the first 1000 impressions if no other advertisers bid on 'Ultra blue widgets factories'.

Would this be a combination of the average CTR of everyone bidding on any keyword combinations within my key phrase?

i.e. an average of all the 'widget' bidders, and the 'widget factory' bidders etc... or some other calculation?

Thanks for your help,
Neil

[pmkpmk]

Is there any statistical data whether the premium slot really is premium when it comes to clicking?

I personally, as a user, often fail to recognize the premium position. I knew from co-workers (with variying internet-savyness) that they too often neglect the premium ad but rather click on the top-ad on the right side or even the second one on the right (soe #3 if a premium ad is present).

[shaka1978]

pmkpmk: google says side ads average 1-2% CTR, top 4-5% CTR. What you are your co-workers do isn't necessarily a representative sample of Internet users. I personally notice about a double in CTR if my ads are boosted to premium instead of just #1 on the sides.

Does anyone know the answer to my question though?

[AdWordsAdvisor]

So, what CTR would I be given for the first 1000 impressions if no other advertisers bid on 'Ultra blue widgets factories'.

shaka1978, two years plus into the AdWords program, it's unlikely that there are too many words left that someone isn't using. ;)

However, if you were to find one, I'm almost certain that the word would start out with a 1% CTR.

(BTW, if you search on the word and see no advertisers on it, this does not mean that there are none. Usually it means there are none who are budgeted to appear at that moment. So be sure to check multiple times.)

AWA

[pmkpmk]

AWA, I think this may be true for English words only. In my native tongue there's lots of words not covered yet - even when I tried checking it multiple times.

[shaka1978]

Thanks for clearing that up AWA. 1% CTR for the first 1000 impressions if no-one else bids on the exact keyphrase is good to know.

I think that even 2 years into the program it's not that difficult to find keyphrases that aren't bid on by other advertisers if you use keyword expansion tools (i.e. those that combine groups of keywords to give every possibility). E.g. combining:

group 1
blue
red
yellow

group 2
widget
widgets
wid get

group3
company
companies
factory

etc...

And it's not really possible to know if anyone else is bidding on the exact phrase as many people may be on broad match for (eg) ‘blue widgets’ if you are bidding on ‘blue widgets factory’. The results will show both advertisers for a search on ‘blue widgets factory’.

I use the above method to create keyword lists that run into thousands. The reason I do this is because I use the dynamic insertion tool as I like the title or the first line underneath to match the exact search phrase. This seems to improve CTR significantly.

Also, I know dynamic insertion only adds the keywords in your list, not the exact search phrase i.e. you bid on ‘widgets’, broad match. The user searches for ‘blue widgets factory’. The dynamic insertion tool only adds ‘widgets’, not ‘blue widget factory’ to your ad text. This doesn’t look so good. So it's important to cover every possible avenue using expansion tools to improve CTR.

I have noticed that most of my very specific phrases combined with dynamic keyword insertion get very good CTR despite low impressions (e.g. 10% CTR with only 50 impressions etc.) So, they're attaining a high CTR even though they are under 1000 impressions and their position is based on the average CTR (i.e. not benefiting from their own 10% CTR). This may be despite them being fairly low down the page if competitors’ broad match keywords are competing with them.

Therefore, there seems to be a trade-off between having a huge list of keywords with dynamic insertion (relevant ad text, low position as <1000 impressions) or a smaller list of less specific terms (less relevant ad text, high position as > 1000 impressions). My experiences so far have been that the first scenario gives a better CTR. I would be interested to hear everyone's views on this. (I am assuming that my ads will beat the average CTR in most cases and that the search phrases are within the character limits).

This may seem like analysing Google's system in far too much detail but if you're an affiliate who primarily concentrates on PPC issues like this have a lot of importance.

Thanks again for everyone’s time,

Neil