Keyword Research: Exact vs. Phrase

[Hamilton]

Which do you generally go with when putting together a road map for clients to best manage expectations yet simultaneous excite them as to what is possible. Bare in mind I usually complete in the .co.za backwater do I expect the answer you go with could well be affected by the more competitive search realm you habitually compete in. That is to say I very often end up with a very small number exact searches for anything vaguely out of the ordinary but prefer to use exact and it is a more precise manner of managing client expectations from the get-go. Thoughts on this anyone?

[alfredthomson]

you should go for the phrase keywords. exact keyword will drive the quality user which will want that product by which he has reached your website while phrase is beneficial as it makes your keyword general. your results will be shown general to the users related to that keyword.

[morestar]

If I know what you're talking about, I would say I go with exact match at first and then dwindle down to phrase and long-tails after we've beat out the competition in the SERPs OR go for phrase if those are high volume search queries.

Basically whatever the high volume search queries are.

[crankydave]

Depends upon what your clients expectations are (what they hope to accomplish) and the type of site it is. You cannot create a roadmap without being very clear on what those things are.

Yes, I'm being vague but there's no way to give a proper answer without knowing those things. Different type sites, different client objectives require different approaches.

[williamc]

Different type sites, different client objectives require different approaches.

That about sums it up rather well, as far as what I would have said too.

[PoisonJam]

If you're talking purely about what you to use in a keyword research tool... I always use exact and make it an exhaustive list of exact match terms. The problem I see with phrase is this:

Exact Match
[keyword] = 1000 spm
[keyword one] = 500 spm
[keyword two] = 200 spm

Phrase:
"keyword" = 1700 spm
"keyword one" = 500 spm
"keyword two" = 200 spm

The figures for "keyword" are being artificially inflated because it includes searches for the longer tail of "keyword one" and "keyword two" as they match part of it. This could also happen with a completely irrelevant search term for another industry that's contains part of your key phrase that you may have overlooked.

To make exact match work though you really need to be exhaustive with your list if you don't want to miss any potentially great terms. I start with market/competitor research, asking the customer for as much info about their industry as possible and then plugging phrases from this into a keyword research tool on phrase or even board match and it comes up with potential synonyms, misspellings etc. I do this a few times adding in things as I learn. I then filter out the irrelevant stuff and then put it through as exact match only to get my final figures.

If they have any historical analytics or PPC data from the past remember to pay close attention, especially if they still have the same website with most of the same content as before

[office7]

The figures for "keyword" are being artificially inflated because it includes searches for the longer tail of "keyword one" and "keyword two" as they match part of it.

I don't think the term "artificially inflated" is the best choice of words. There is no inflation in the number of searches and the result is not artificial. I can't see a problem either. The last figures I saw, people don't use exact search as much as search without quotes. The search figures are what they are. If the "problem" is that the numbers are high, then you need to balance that against the supply for that keyword. How many sites are returned for that keyword search? Keyword research should take into account the demand and supply side of the equation.

[PoisonJam]

The last figures I saw, people don't use exact search as much as search without quotes.

My understanding is this:

Searcher types in keyword one is considered an exact search for keyword one. No square brackets or inverted comas necessary. It will also be included in phrase figures when the words are in that order with or without other words around it and in any broad figures with both those words in any order.
Searcher types in buy keyword one is a phrase match for keyword one. It will also trigger broad figures with those words in any order, of course. It won't meet exact terms.

I don't believe an exact search is dependant on people using [ and ] or quotes around their search terms - how many people apart from SEOs and very-tech-savvy people are even aware of it? This theory is also backed up by PPC results I've seen, which are obviously instantly measurable.

Or is this not what you meant? That line was a bit unclear to me.

When I say artificially inflated I mean it can be in terms of relevance. You just have to be careful that you're not thinking that all these searchers will be looking exactly for what you're trying to rank for, especially if these figures are being shown to clients at all in, say, a rankings report.

Looking at my first example the figures might suggest to someone that there were 24,000 searches, when in fact there were only 17,000. The phrase match option makes the market look more popular than it actually is to the untrained eye if you are looking at all the figures at once. One needs to be especially careful if you are presenting this data to clients in any way. Apologies, I'm really not the best at explaining what I mean without diagrams and such

[crankydave]

Exact phrase matches are searches within quotes. An exact phrase match search would be "keyword one". Without the quotes, the two words can appear anywhere on a page and can also be skewed further by anchor text in links pointing to the page.

Not many average surfers are aware of that, but that doesn't mean such searches don't have value for those who use it.

Using brackets... [keyword one] provides you with the same results at not using them at all.

ETA... Using Google as the SE

[PoisonJam]

Without the quotes, the two words can appear anywhere on a page and can also be skewed further by anchor text in links pointing to the page.

Aren't we talking about the number of search queries as reported by keyword research tools rather than results? Or maybe that was just me