How do I track print media receptiveness?
November 9, 2011 12:10 PM   Subscribe

Help me devise a sinister plan to track the effectiveness of a print media campaign.

My client spends tons of money on print media and wants to find out how many people are actually responding to it. Their suggestion is use a specific url in the print ads and track clicks.

Has anyone tried a similiar experiment? It is not a iron-clad test, so I am hoping that I can find something more concrete. google only brings me to things like adwords....

I'm not a media expert, just the web lacky so I'm grateful for any help from experts out there.
posted by OlivesAndTurkishCoffee to Computers & Internet (10 answers total) 2 users marked this as a favorite
Use a discount code or coupon rather than a url. The conversion from print media to web media is pretty atrocious, generally speaking, but a higher percentage of people will use a coupon.
posted by valkyryn at 12:14 PM on November 9, 2011

QR code rather than a specialized URL, if you are targeting people who will actually know what a QR code is. If your URL is I will never remember your special URL

of course the overlap between people-who-use-QR-codes and people-who-buy-fireworks is probably pretty small
posted by desjardins at 12:23 PM on November 9, 2011

Best answer: Yup, QR codes for the win. If you shorten your URLs via, it'll not only make the QR codes but provide detailed traffic stats.
posted by jbickers at 12:25 PM on November 9, 2011

A combination of dedicated URL, telephone number and email address is the traditional way to do this. Because people hack the trailing bits of URLs off, a discreet campaign domain, a shortened URL (dodgy, trust issues) or QR code works better.
posted by DarlingBri at 12:32 PM on November 9, 2011

QR codes, or even web addresses, are poor indicators of success unless you've got something concrete and useful there. If your ad is designed to increase brand awareness, and the address is just for a landing page, no one's going to take the time to type it in or scan it. You might get some interesting anecdata about geographic reach, but nothing anywhere near statistically relevant.

valkyryn's got the right idea- as abysmal as conversion rates are, let alone tracking ability, discount codes are fairly effective if you can handle them on the sales side.
posted by mkultra at 12:39 PM on November 9, 2011

QR codes are a terrible idea. The vast majority of your readers won't have any idea what a QR code is, and of those that do, very few will do anything with it. A QR code won't give you anything resembling a good measure of your audience.
posted by griseus at 12:42 PM on November 9, 2011 [1 favorite]

There are vendors out there that can spin up campaign-specific phone numbers. Customer dials the campaign-specific phone number, the call is forwarded invisibly to the business' normal phone, and that lead is tracked by the vendor for easy reporting later on.

Google "digital call advertising."
posted by Sauce Trough at 1:01 PM on November 9, 2011

put this on all of your ads: "ask for Bob and receive 10% off" and change the name for each ad you are looking to track, this way at the end of the month you can just add up the totals to find the most effective media. Just make sure to use odd names that aren't the same as any employee, every employee that answers the phone tracks the name and most importantly - is aware of the campaign so they don't hang up assuming it's a wrong number.
posted by any major dude at 1:15 PM on November 9, 2011

Best answer: This is what I do for a living. I work in digital marketing and about half of what we do is tracking the effectiveness of marketing tactics: online, search, print, etc. Attribution of conversions is, in my opinion, not very scientific.

We use all of the strategies listed above to directly attribute conversions: vanity URLs (un-encoded, bitly-ed, QR-ed, etc), invitation codes, etc. This provides direct attribution to a fraction of the conversions.

We also use referring URLs to attribute another fraction. If the referring URL is usable, we consider the conversion directly attributable.

Sometimes for print, we can get the distribution list of the print item and the sales data for the conversions and do a match-back to the self-entered information in the conversions. This attribution can be direct or fuzzy. There can be two John Smiths living at the same address. It happens a lot, actually.

Then we take the un-attributable conversions and try to assign them indirect attributions based on patterns of behavior. For example, a certain percentage of folks who come back to the site over and over again are considered bookmarks. Depending on the needs of the client, we may attribute these people to the original tactic that brought them to the site, or we may just attribute them as repeat visitors.

There are lots of other assumptions that give us indirect attributions. These depend on the tactics and marketing calendar. For example, if we see a spike in conversions while a single tactic is in-market, we indirectly attribute the incremental conversions (those that exceed our baseline) to that tactic. Over time, similar tactics will be assumed to have similar effects, and we presume to apply multivariate attribution even when multiple tactics are in-market.

Once we get to this fuzzy level, it's dangerous to assign specific conversions to specific tactics. Sometimes we do anyway, other times, we deliberately lose specific data and just report counts.

Finally, we've got a big pot of un-attributable conversions. We take the proportions gained in all the previous steps, apply some sanity checking (these particular conversion couldn't possibly have received the print item because they're in the wrong country, for example), and divvy up the remaining conversions into a big histogram.

At the end of all of this, we have approximate measures of the effectiveness of each marketing tactic. We also have a bunch of direct attributions that can be targeted much more precisely. Depending on the client, we may iteratively test and refine our models and go back in time, re-attributing conversions and refining our tactics.

There's a little bit of math and a whole lot of statistics involved in this, but it's fundamentally pretty simple as you can see. I'm in the online side of the business, not the analytics side, but I have to design the software with analytics in mind, so I do have a somewhat solid grounding in all of this.

If you're interested in the software design aspects of conversions and attribution, I know a little bit about that too.
posted by stubby phillips at 3:21 PM on November 9, 2011 [2 favorites]

It depends on what you are trying to measure- one ad buy versus the next, or different areas of distribution in the same campaign.
posted by gjc at 7:31 AM on November 10, 2011

« Older Get this out of my head!   |   How much will it cost me to read Metafilter on the... Newer »
This thread is closed to new comments.