Are They Lying: $_SERVER["HTTP_REFERER"] Question

[Good Man]

Hi There

A client of mine has some banner ads placed on the net. Instead of relying
on the sites the ads are posted on tell us how many clicks they are
getting, we wanted to track the number ourselves.

I settled on using $_SERVER['HTTP_REFERRER'] to determine what referred the
user to our page. Sure enough, when testing, it all works (ie: i click the
ad on 'sample.com', which takes us to my page, and sure enough in our
tracking database i see 'sample.com' has had the visit count increased by
one.) I am looking for the string "sample.com" to make sure that the count
goes up whether or not the person visited sample.com via
http://www.sample.com or http://sample.com/

My question is: is $_SERVER['HTTP_REFERRER'] reliable? Essentially, a
place my clients have advertised with have said '80 people clicked your ad
this month' while our own stats show that only 15 people clicked it.

Is my method of tracking good/reliable enough to call them liars, or is my
method not too dependable?

[Daniel Tryba]

Good Man <he***@letsgo.com> wrote:

My question is: is $_SERVER['HTTP_REFERRER'] reliable?

Short: No

Slightly longer: with the referer you depend on the client and
all involved proxies to supply information.

[Yttrium]

"Good Man" <he***@letsgo.com> a écrit dans le message de news:
Xn***********************@216.196.97.131...

My question is: is $_SERVER['HTTP_REFERRER'] reliable?

Hi,

Of course not..
You can change it easily,( ie : in firefox, exists a plugin which you can
choose to active or not http_referrer..)
That depends on the browser..

( Sorry for my english, caus' Im French :-( )
--
http://www.danstesyeux.com

[Dave]

Good Man (he***@letsgo.com) decided we needed to hear...
<snip>

My question is: is $_SERVER['HTTP_REFERRER'] reliable? Essentially, a
place my clients have advertised with have said '80 people clicked your ad
this month' while our own stats show that only 15 people clicked it.

Is my method of tracking good/reliable enough to call them liars, or is my
method not too dependable?

Not really a PHP question, but anyways no, HTTP_REFERRER is not that
reliable. It *should* be filled in, but e.g. some ISP caches (such as
AOLs) block it and some personal firewalls (Norton is one IIRC) also
block it - you can configure Norton to send it to certain sites but
its off by default.
--
Dave <da**@REMOVEbundook.com>
(Remove REMOVE for email address)

[ECRIA Public Mail Buffer]

Let me clarify -

Sure, the $_SERVER["HTTP_REFERRER"] can easily be changed - but don't even
try to tell me that 80% of users modify their referrer because they can -
that's just plain nonsense...

On a similar note, if 80% of web surfers are using a proxy, then I am Mickey
Mouse!

So... someone is probably taking you for a ride.

ECRIA
http://www.ecria.com

[ECRIA Public Mail Buffer]

They lie.

ECRIA
http://www.ecria.com

[Ivan Omelchenko 608308824]

ECRIA Public Mail Buffer ÐÉÛÅÔ:

Let me clarify -

Sure, the $_SERVER["HTTP_REFERRER"] can easily be changed - but don't even
try to tell me that 80% of users modify their referrer because they can -
that's just plain nonsense...

On a similar note, if 80% of web surfers are using a proxy, then I am Mickey
Mouse!

Hi, Mickey! :)
that for sure - in my country at least 80% using a proxy server

Living in Ukraine.

[ECRIA Public Mail Buffer]

Of course, the acid test is to take the $_SERVER["HTTP_REFERRER"] out of the
equation.

Create a page that is not indexed or linked from any other sites and then
change the target of your ad to point to this page. You would probably want
to put a separate hit tracking script on this page followed by a simple
header redirect to your main page. Then, since ONLY clicks from the ad site
will result in hits on this special page, it will be very easy to determine
if their claim is exaggerated or not.

Please post your findings if you decide to do this...

ECRIA
http://www.ecria.com

[Philip Olson]

BTW it's HTTP_REFERER not HTTP_REFERRER :-)

[Good Man]

"ECRIA Public Mail Buffer" <ng***********@ecria.com> wrote in
news:d8**********@murdoch.acc.Virginia.EDU:

Of course, the acid test is to take the $_SERVER["HTTP_REFERRER"] out
of the equation.

Create a page that is not indexed or linked from any other sites and
then change the target of your ad to point to this page. You would
probably want to put a separate hit tracking script on this page
followed by a simple header redirect to your main page. Then, since
ONLY clicks from the ad site will result in hits on this special page,
it will be very easy to determine if their claim is exaggerated or
not.

Please post your findings if you decide to do this...

ECRIA
http://www.ecria.com

i'm upgrading/redoing their website this month, and indeed this is what I
will do for all of their banner advertisements. good idea, i'll let you
know what happens in the future...

[ECRIA Public Mail Buffer]

Rats, you're right. This always gets me...

dictionary.com says:

1 entry found for referer.
<World-Wide Web> A misspelling of "referrer" which somehow made it into the
HTTP standard. A given web page's referer (sic) is the URL of whatever web
page contains the link that the user followed to the current page. Most
browsers pass this information as part of a request.
(1998-10-19)

ECRIA
http://www.ecria.com

"Philip Olson" <ph*********@gmail.com> wrote in message
news:11*********************@g49g2000cwa.googlegro ups.com...

BTW it's HTTP_REFERER not HTTP_REFERRER :-)

[Malcolm Dew-Jones]

Good Man (he***@letsgo.com) wrote:
: Hi There

: A client of mine has some banner ads placed on the net. Instead of relying
: on the sites the ads are posted on tell us how many clicks they are
: getting, we wanted to track the number ourselves.

: I settled on using $_SERVER['HTTP_REFERRER'] to determine what referred the
: user to our page. Sure enough, when testing, it all works (ie: i click the
: ad on 'sample.com', which takes us to my page, and sure enough in our
: tracking database i see 'sample.com' has had the visit count increased by
: one.) I am looking for the string "sample.com" to make sure that the count
: goes up whether or not the person visited sample.com via
: http://www.sample.com or http://sample.com/

: My question is: is $_SERVER['HTTP_REFERRER'] reliable? Essentially, a
: place my clients have advertised with have said '80 people clicked your ad
: this month' while our own stats show that only 15 people clicked it.

: Is my method of tracking good/reliable enough to call them liars, or is my
: method not too dependable?

In addition to what others have said, if the pages are cached anywhere
along the route (gateways and/or proxies) then you won't see all the
traffic.
--

This space not for rent.

[Bernie Cosell]

Good Man <he***@letsgo.com> wrote:

} I settled on using $_SERVER['HTTP_REFERRER'] to determine what referred the
} user to our page. ...
}
} My question is: is $_SERVER['HTTP_REFERRER'] reliable? Essentially, a
} place my clients have advertised with have said '80 people clicked your ad
} this month' while our own stats show that only 15 people clicked it.

The short answer, as others have commented, is "no". But the longer answer
is that by and large it *IS* reliable: not very many folk know how to, or
have the facilities to, *change* the REFERER -- there are lots of
filters/firewalls/plugins, etc, that'll _block_ REFERER. so I'd say that
to first order, if you *GET* a REFERER, then it is _most_likely_ correct
[especially if you look for the specific REFERERs that you are expecting]
but you'll be left with the problem of how to sort out hits that come in
with no REFERER.

/Bernie\

--
Bernie Cosell Fantasy Farm Fibers
be****@fantasyfarm.com Pearisburg, VA
--> Too many people, too few sheep <--

[Samuel (lets dot monroe at gmail dot com)]

A good amount of spanish connections are behind a certain ISP's massive
proxy server too.

It's a long story.

[John Dunlop]

Good Man wrote:

A client of mine has some banner ads placed on the net. Instead of relying
on the sites the ads are posted on tell us how many clicks they are
getting, we wanted to track the number ourselves.

'Clicks', a marketroid's term, isn't it?

--
Jock