Social Media ROI and the Last Cookie Conversation

by Shannon Paul on November 12, 2009

[tweetmeme source=”@shannonpaul”]
Cookie Monster on the Google HomepageThe first push back regarding social media is the dreaded ROI question, right?

There’s also been a lot of talk about how conversation and dialogue can’t be measured in terms of revenue too. Maybe that’s true. Or, maybe we’re measuring the wrong things or we’re only capturing part of the picture.

Since I come from a communications background I never learned much about web analytics until I took it upon myself to learn. It becomes much easier to have the ROI conversation when your piece fits into the overall sales or cost savings equation. However, understanding how ROI is measured in other online marketing disciplines can definitely come in handy.

Mind you, I don’t have all the answers and I know each situation is unique, but I think this fact is worth highlighting. Most conversion analytics credit the final click, with credit for a conversion or sale. This means that the last ad, or affiliate link or referring site gets credit for the sale or conversion. Each one of these types of click installs a cookie in the browser that determines which referrer gets credit. Most links shared in social networks do not come with cookies, but Google Analytics will still pick up the traffic from the referring site.

Why is the last cookie question important for measuring social media ROI?

I have been thinking about this for awhile since having a conversation with Angel Djambazov on the way to Blog World Expo (he and I were on the same flight from Seattle to Las Vegas for the conference). He explained to me how the final cookie, or final click, practice can often skew data. Then a recent post from Rob Birgfeld inspired me to share this information with you here.

Rob’s post on the “myth of the last click” cites a study by the Microsoft Advertising Institute that states, “users interact with an average of 2.2 other ads from the same brand over two days before the conversion.” This study includes a sales funnel of 14 days. Rob then asks what if we were to extend the sales funnel to 90 days?

“…let’s expand the conversion funnel to 90 days, in which case the user is exposed to an average of 18.5 ads over a three-month period… [the] operating theory is that all of these interactions play a role in the final conversion, and they should not be discounted when quantifying ROI. According to Strong, 94% of touch points in today’s “last click model” are thrown away and not given any credit for a sale.”

Several of the touchpoints cited in the above quote could definitely include interaction in social networks and social media channels.

The answer isn’t who gets monetary “credit” for the sale, but rather, doing what it takes to fund, maintain and foster the truly valuable channels that bring customers and that final click into perfect alignment. Maybe one way to examine the data is to look for correlations between social media mentions and actual click-throughs from other online marketing channels.

The idea that clicks trump views and impressions may have been overplayed in social media to give the appearance of substance and business prowess. However, the problem with the final click metric is that it will often discount valuable channels like those in social networks if they do not generate a money-shot style click upon the first impression.

Based on this information provided in this study, my advice to anyone searching for the ROI of social media is:

  1. Question how online conversion is being calculated and whether it’s a final click metric. If so, recommend an alternative view to accompany final click metrics based on engagement and look for correlations between the data.
  2. Don’t be so quick to abandon outreach efforts that appear to be underperforming based on a final click model.
  3. Track engagement on your site from referring sites. Drill down time on site and number of page views by each referring blog, community or social network.
  4. Examine the number of return visits generated by social networks.
  5. Examine the bounce rate from visits referred by social networks.
  6. Look for trends between mentions, sentiment and conversion that attempt to identify the real time it takes most customers to complete the sales cycle.

Social media has the potential to generate a valuable first impression, but it may not ever generate a final click. Aligning outreach efforts in social networks with other types of online marketing and SEO efforts can give a much more holistic view of what networks and channels are adding the most value to your entire sales funnel.

By the way, one of the best presentations on determining the ROI of social media is by Olivier Blanchard, aka @thebrandbuilder on Twitter. Check it out here on SlideShare.

What other types of analytics do you use to determine the true ROI of social media beyond fluff and without giving too much weight to the final click and worrying about who gets the last cookie?

Image of Google’s homage to Cookie Monster and Sesame Street posted by COG LOG LAB.

Bookmark and Share

If you enjoyed this post, make sure to subscribe for regular updates!

InstapaperLinkedInSphinnPosterousMixxStumbleUponGoogle ReaderNewsVinePrintFriendlyTumblrSave and Share
Cancel reply

Leave a Comment

{ 12 comments… read them below or add one }

November 13, 2009 David J Carr 1

We’ve found that you can actually use successful Journey Analysis case studies from eCommerce or digital direct to sell the benefits and ROI of social activity to more traditional “hard metric” clients.

Journey analysis can allow subsequent sales to be attributed up to 10 clicks earlier on in the customer journey which enables you to understand search activity value.

Conventional wisdom says you don’t spend vast sums on generic or lateral research search terms – they never convert effectively. However, that’s only if you just look at the last click. Journey Analysis can identify specific non-brand or product research terms that feature further back in the customer journey, and it turns out they’re often highly (and measurably) influential. Basing keyword investment on this insight, one project delivered sales up 122%, spend up just 42%. Over a marked and measurable period of time.

Essentially, help people when they’re researching and evaluating, before they’re ready to purchase, and you get a positive outcome when they do get closer to buy.

This kind of “harder” traditional ROI proof within such a widely accepted channel like search and eCommerce has enabled us to demonstrate an insight that is highly applicable to social media.

In turn it has helped to get senior client buy in – it not only makes sense but has numbers to back it up – so that we can develop social activity and campaigns focused around stimulating research activity in the Exposure and Engagement categories of online relationships documented here:

Anyway, hope that is of use. Thanks for your post, really good food for thought.

David J Carr


November 14, 2009 David Vellante 2

ROI is a mathematical calculation.

It’s: Benefit divided by cost of benefit.

That’s not to say there’s not other value in social media as you’ve pointed out in this excellent post. However practitioners of social media should remember that when you’re having conversations with CEO’s and CFO’s and they bring up ROI, you’d better remember this equation– because that’s how they think.

Here’s some additional info on this topic that might be useful.


November 14, 2009 Shannon Paul 3

I don’t think anyone has trouble with the benefit minus cost of benefit piece… But rather, deciding what a benefit IS and weighting benefits in a meaningful way so that the data actually gives a realistic interpretation of any benefit isn’t always so easily done.

I agree that we shouldn’t over complicate the issue, but oversimplification doesn’t help either. Make sense?


November 14, 2009 Shelby Thayer 4

Great post. I wholeheartedly agree. We use Omniture and it’s very easy to change to first referrer in it.

You can hack Google Analytics to do the same thing. Check out Brian Clifton’s blog (Measuring Success with Google Analytics) and specifically his hacks and downloads page. It’s great stuff.

Also, I’m glad you mentioned Olivier Blanchard. One thing I’ve been playing with lately are the activities and outcomes timelines.

Olivier has so many great posts about social media measurement and just social media in general.

Not for true ROI, but a couple other metrics I look at if the goal of the social media campaign is brand awareness, are trends in branded keyword referrals and direct traffic.

Thanks for the great post.


November 14, 2009 roger ewing 5

This is a good piece, Paula. Technical enough to be interesting, simple enough to keep my attention. As a result of this post, I am going to spend more time exploring the last cookie.
Best wishes


November 14, 2009 Angel Djambazov 6

Hi Shannon,

Great job laying out a well developed argument! It should give your readers in charge of social media initiatives the ability to quantify the benefits of these efforts in comparison to other channels.

I think it is important to understand the difference between ROI, as David Vellente laid out, which gives some room for a branding/engagement argument, and ROAS (Return on Ad Spend) which by definition has a tighter set of parameters. If the advertiser’s (whether they are involved in a social media campaign or not) main focus for success is ROAS then most elements of branding/engagement are irrelevant. The reason is this metric for success is specifically looking for a direct dollar value generated as compared to the actual budget being spent.

This usually implies a need to know which channel ultimately closes the sale. It is this need to know that makes advertisers hyper-focused on the final click or last cookie in.

If an advertiser is primarily concerned about ROAS and they are using the last cookie in method of determining who ultimately closes the sale, then there is one other step I would recommend. It goes in conjunction with steps 1 and 2 that you laid out and deals with cannibalization.

Cross-channel cannibalization is when the marketing costs/efforts of one marketing channel are not considered because a different marketing channel is being given credit for them. This impacts both budget allocation and proper allocation of costs.

For example, I see an initiative in a social media setting like Facebook, and let’s say I interact with the conversation but don’t buy. Later, I see an ad for the same company in my Gmail account, delivered to me via their efforts in Google’s Content Network. I am then prompted to buy, in part based up on my initial engagement socially.
In this example the social media channel lets me know about the company/ad/promotion, even though at the time I didn’t buy. Later, it was an ad delivered by the search channel that enabled the final sale.

Most advertisers mistakenly credit only the search channel for the purposes of measuring ROAS but they also only attribute COSTS to that channel as well. When really (x cost from social media) + (x cost from search) = sale (converted the user into a customer). Later companies who don’t measure correctly will only give credit for and thus additional budget to search.

Savvy advertisers, when their marketing initiatives become complicated enough to have multiple channels, track both the “first cookie” and the “last cookie” in. They then measure what marketing channels brought in the “first cookie” or the first time a potential new customer came to the site, and compare that to how often a conversion/sale was made. Then they see how often the channel that initially brought the customer also closed the sale. It may very well be that the social media channel is always creating the first engagement and the affiliate channel (through coupons, for example) is always closing the deal.

In that case proper measurements in terms of cost should be allocated, credit given and budget set aside so both channels can continue to provide value. It also allows for better tactical decisions to be made since you then have a better understanding of how your marketing channels interact together and with your customers.


PS Glad to know my hyper-babbling about marketing was helpful. Look forward, as always, to seeing you again on the conference circuit.


November 15, 2009 Shannon Paul 7


Thank you so much for sharing even more of your insight here. I am forever impressed with your sensitivity to all of the complexity involved in any kind of integrated marketing approach. I love your hyper-babbling.

Thanks again for sharing what you know and letting me pick your brain!


November 14, 2009 Larry Bruce 8


From the article above and comments it would seem to suggest social media as brand support. I am struggling with question:

Is social media more prone to brand awareness or brand validation?

What would your and / or the groups thought on this question be?

Larry (@pcmguy)


November 15, 2009 Shannon Paul 9


I’m of the opinion that social media can both promote brand awareness and help with brand validation. It all depends on the strategy and how you’re aligning your presence with those you’re trying to reach… it also depends on how much built-in awareness there already is with your particular brand. The benefit/challenge with social media is that there is a lot of flexibility here depending on what the goals are.

I think in this instance, it might be helpful to have an example or scenario to better answer the question since there are so many variables.

That’s a great question though. Thanks for posing it here. :)


November 16, 2009 Rob Birgfeld 10

Really great post. I think it’s clear to see this is an area that interests many of us– especially those of us looking to justify social media strategies. In my eyes, it will continue to be difficult to deliver ROI for social media activities if there is no true link between the sale and the interaction. As we know, an overwhelming amount of people who read blogs, twitter updates, reviews, etc. do not interact in the space. They are the lurkers. If they truly do represent 90% or more of our audience(s), then the timeline approach detailed by Olivier Blanchard may be our best tool right now. While we’ll still struggle to fight for clean, direct connections to a sale/action, the good news is we have far more data on our side than those still advertising in traditional (print, television, radio) media.

Once again– awesome take. I think this is a discussion that will continue for some time.


November 16, 2009 Jeremy Epstein 11

Solid post, as usual.

The analytics are critical, but I think there’s also a cultural mindset here.

The glory goes to the guy who carries the ball into the end zone. The average fan gives him the credit.

The true fan, the “student of the game” says, “wow, there were a lot of critical steps that led to that score. There was some great blocking along the way. Some things we could have done better. Let’s look at the tape and figure out how we can make the offense even MORE effective.”

There’s a cultural element to this as well. Not easy to change or establish, but it is there.


November 19, 2009 Scott Severson 12

Hi Shannon-

I would guess that today the vast majority of online marketers measure ROI either on the last click, or they just give Google the credit.

While ROAS is a move in the right direction, I believe it’s only a matter of time before the measurement tools and user sophistication reach a point where we can truly measure how all of our media works together in concert to achieve a marketing goal.



{ 5 trackbacks }

  • 4 videos infallibles para convencer a su jefe de invertir en medios electrónicos y redes sociales | Viralogía November 17, 2009
  • Netmedia Blog : We Know IT – 4 videos infallibles para convencer a su jefe de invertir en medios electrónicos y redes sociales November 19, 2009
  • Play to play: Wie sich professionelle Social Media bezahlt macht November 27, 2009
  • PR & SM ROI or Public Relations & Social Media Return on Investment « Jules Zunich::Z Group PR December 4, 2009
  • Measuring ROI in Social Media Campaigns « Avi Joseph March 7, 2010

Previous post:

Next post: