Measuring the ROI of
What is ROI?
“The concept of an investment of some resource
yielding a benefit to the investor”
SLIDE #3 NOTES
There’s a perception that because digital is highly measurable, that this should be an easy question. It is and it
The challenge is the language; it’s a dangerous(ly easy) term to use without thinking about what it means
It doesn’t necessarily mean “conversions”, “sales”, or even “ecommerce revenue”
It’s multi-faceted, and completely bespoke. It differs by organisation, and by context. We’ll explore this.
There are challenges beyond definition; even if you can articulate what it is, how do you break out the ROI of
content marketing from everything else you do? How do you measure the ROI of a piece of content, vs. the
campaign you invested in to send traffic to it? Or vs. peripheral contributions, such as the colour of your logo, or
the physical brochures you sent out? Is content the asset, or the context? It’s where things come together.
Measuring it as a whole is impractical; you need to break it up into pieces.
Three places where we need to at or for ROI:
Activity or content which improves revenue
…which decreases cost
…or which builds brand – DANGER!
SLIDE #5 NOTES
• There’s a temptation to start talking about halo
effects, branding and intangible value.
• This is equally dangerous, because it takes us away
from data, from metrics, and from business cases
which rely on more than just good storytelling.
• Everything can be measured or approximated and
given a value; but it’s a conceptual challenge
ROI is about value
Image credit: epSos.de, on Flickr
SLIDE #7 NOTES
• Introducing another intangible concept… Value
• The question isn’t “what’s the ROI”; this masks a bigger business
question, which is always “which levers should I move in order to
make more money / spend time/resource more effectively / build
• That’s a better question, it’s the actual question being asked, and
it’s easier to answer.
• In most cases, value == money; but the same principles apply for
non-profits and charities; measure ‘equivalent economic value
SLIDE #9 NOTES
• Who’s ever looked at or used heatmaps or clickmaps?
• Who’s found them pretty but disappointingly un-actionable?
• …so people are clicking on stuff. Helpful. Not.
• This is useful!
• Typical content analysis focuses on the what.
• More advanced stuff, as our tools and cultures evolve, looks at the
why. None of that matters if you can’t answer the so what.
• How do we get to this?
There are some challenges to overcome
SLIDE #12 NOTES
Paid media (PPC, SEO, Affiliate marketing, etc.) and other mainstay digital activity goes after the tip of the
Prospects group. Grab people who are swayed, go after what works and where there’s volume.
This is brute force marketing.
Why are paid search landing pages radically different from organic search pages when the consumer generally
doesn’t know the difference? How often do you see paid search marketing on a content page? A blog post? A
We live in this ecosystem because people struggle to measure and prove the ROI of content
Challenges around differentiating the effectiveness of the traffic vs. the destination, and ownership of
budgets/performance + channel segmentation
PPC landing pages etc. don’t “work”, and they don’t tap into a different market; you’re not “removing distractions”
and refining targeting - you’re just refining your filter bubble (global warming, Facebook).
“TAKE OUR THING, TAKE OUR THING!”
Where does content sit in this? How do you either isolate or integrate it in a channel-centric marketing world?
The role of content
Risk of damage
Value & Performance
Links & citations
Challenges from living in a DM
But… we use data badly
SLIDE #16 NOTES
We use clickstream-level terminology, rather than business terminology. Conversion rates. Bounce
rates. Pageviews. Repeat visits…
Because it’s not as easy or joined up as we’ve been told, but we’ve all been in the industry for so
long that it’s embarrassing to admit that it’s not… That it needs as much – maybe more – thought
than the kind of complex econometrics models we see with established and mature brick-andmortar brands. But the tools and the industry tells us it’s easy – just put the code on the page.
Goals, context, and actual understanding are ‘nice to have’ extras. Data is worthless. More is not
better, or more useful. Better is better, and this is something we’re not very good at.
Limitations of the tools and processes means that:
Marketing activities are segmented by channel
We’re limited by visit-centric metrics (conversion rate) and optimisation.
We’re focused on last-click attribution
Who’s familiar with Google’s ZMOT principle? We need a model which aligns with this.
But… PPC works. Channel segmentation works. It all works, because that’s all you’re doing. How do
you pitch in multi-visit, ZMOT behaviour when all of your research says that it doesn’t happen?
image credit: Smart Insights and First10
SLIDE #18 NOTES
What’s the ROI of somebody reviewing an eBook? How do you get buy-in?
Unfortunately there’s no easy win – you have to do one and show that it works before you can get the big buy-in.
But the first one rarely works. So work your way through this list in order of most likely to demonstrate any kind of
tangible return. Win on *one* and scale it.
Which channel or segment of marketing does this belong to? Awkward!
Survey on conversion influences, reverse-engineer value
Apply at action-level, e.g., eBook = 1/100th of a £100 conversion, eBook = £1 per click. Or per read? > 1 min, scroll
to bottom; more complex with PDFs?
Move away from asset segmentation, and work on per visit goal value, cost per visit and value per visit
This is out there… This should be normal!
What drives change? The competitive landscape. Who goes first? Nobody. Mediocrity rules.
Be first to market and actually market. All this stuff we’re doing? It’s still just shouting into the wind.
Challenges from living in a DM
There are two routes
SLIDE #20 NOTES
You need to understand what’s actually happening.
What happens if somebody follows a link to a blog post from a Tweet, signs up for a
newsletter, comes in three weeks later from an email about a special offer, buys a product, and
three weeks later returns it?
You need to understand that behaviour, spot it, and use it to steer your direction – but you can’t
rely on it for tactical or practical decision-making. Join all this up, import/integrate downstream
data, but don’t aim to use it at the ground-level.
…what about service- or subscription-based models? Single value? Multiplied by duration? Lifetime
value? Propensity to renew as a percentage of average lifetime value? More and more out of sync.
Time delays, bad decisions.
Day-to-day… Work with approximates, and accept value inflation (multiple overlapping values).
You’re not trying to understand how much money the website makes – that doesn’t make sense
(differentiate site from brand, content, traffic, marketing, etc). You’re trying to understand which
combinations of activity generate value, and do more of that.
SLIDE #22 NOTES
“We’re not going to talk about conversion rates and numbers of sales – we’re going to talk about a
theoretical, abstracted notion of propensity to generate value”
Needs definitions; objectives process
This is much bigger than just content marketing
…and most organisations don’t have this kind of framework
Can’t educate them; but you can trap them. And there’s a process.
They get the concept of content marketing.
They read the blogs and listen to their peers.
But if you can’t make a business case, prove the numbers and make a compelling argument for
spending money on content rather than where it “works”, you won’t get buy in.
Use top-level business objectives
Be seen as thought leaders
Get people reading and engaging with our
Increase sales revenue
Double revenue by 2014
Increase customer lifetime value
Provide the best customer service
Create and curate great content that helps
users find the information they need
Achieve and maintain a high conversion rate
Improve marketing efficiency
Increase self-service actions on the website
SLIDE #24 NOTES
• Cut down summary of the process; link at the end.
• Get HIPPO business objectives. These will be poorly
defined, ambiguous, some will be radically
abstract, some will be overly-specific and metriccentric. Doesn’t matter.
• Translate these into things which digital does. Get
their sign off.
Build HIPPO opinion into explicitly defined metrics
Be seen as
Get people reading and
engaging with our blog
# triggers of time on page > 2
mins on a blog post (per
# average comments-per-post
# sales (per month)
Increase sales revenue
Increase customer lifetime
Provide great content and
help users find the
information they need
Department, channel, product
£ sales revenue (per month)
% of new consumers via the
website making a repeat
purchase within 3 months
% task completion rate (per
% satisfaction rate (per
Channel, source, campaign,
keyword, gender, geographic
region, entry page
Initiating page, Channel,
source, campaign, keyword
Maintain a high conversion
rate to sale
% visitors to
macroconversions + calls (per
Channel, source, campaign,
keyword, call type
Increase self-service actions
on the website
£ value of microconversions
source, campaign, keyword
SLIDE #26 NOTES
Add explicitly crafted KPIs. Don’t use stock. Don’t leave any room for interpretation.
Specify the unit of measurement, and the framework of reference (total, rolling, proportional, etc.)
Identify key segments to allay fear.
Get sign off at this point (they won’t challenge the KPIs)
Set sensible targets, thresholds and champagne moments; these won’t exist, and it’ll be a
Get sign off.
All those intangible, fluffy brand objectives the HIPPO committed to? They’ve got content-related
KPIs. And if they’re underperforming against targets, great. There’s your budget signed off.
This can take months, because nobody’s ever asked anybody to write down their business
1. Lower the threshold of success with
2. Gain incremental permissions and build a longterm relationship
3. Utilise nurture funnels and marketing automation
SLIDE #28 NOTES
This allows you to…
Micrconversions: small actions; a like/share, a comment, a newsletter signup, a
PDF download, a registration, a rating.
Permissions: sign up, give details, commit to next steps, reinforce
Nurture: Progress the user along a bespoke funnel. Entry and exit may happen at
any point. There is no one funnel.
Need these in place in order to be doing the kind of activity which drives content
ROI, and you need the kind of buy-in we’ve explored to get to this stage. This
should be your core processes; it’s a constant, evolving and iterative process
Picking metrics, and driving ROI
It’s all bespoke
(but here’s some food for thought)
Value per visit and cost per visit
Understand how much it costs to bring
them in, and what you get out
SLIDE #31 NOTES
• This is your ROI!
• Import production costs, marketing spends, even lighting
and heating if it makes sense against channels and
content. You can do this in GA.
• Look at dimensions which are the most profitable;
beyond keywords and pages, look at combinations and
segments and see what works. Do more of that; much
easier than brute forcing areas which under-perform
What’s the cost/value/savings of somebody
reading an FAQ and voting ‘this was useful’ vs.
phoning the call centre?
…or “this resolved my problem” in a help section
What would it cost to advertise to 1m on TV?
What’s the equivalent cost/value of the social reach
and syndication of your content?
…Or of 1,000 watching your 20 minute video end-toend?
Survey on the impact of key assets
“Was our white paper on *x+ influential in your
Reverse-engineer the value of the consumption of
this asset in terms of its propensity to be part of
the buying decision
SLIDE #36 NOTES
• Nature of purchase, scale of spend? Impulse or considered? B2B?
• GA metrics are all visit-centric. This isn’t great. Encourages bad behaviour.
• Use your own metrics.
• API, custom dashboards, etc. Perception that it’s expensive and/or
complex; it’s not if you’ve got explicit definitions and know what you need.
You should be able to design exactly what the output looks like. Outsource
• For GA, Analytics Canvas is a great entry-level tool, and also couples as a
good proof of concept visualiser
• Other tools; assess monthly unique visitors vs. visits, etc., and make a
Big Conversions are still important
Sales. Subscriptions. Revenue.
Don’t overlook these; and grab actual £
metrics whenever they’re available
SLIDE #38 NOTES
• GA does this automatically if you give it goal value; but it’s limited.
• Too big/high; need to account for the little things which add up.
• Conversion rate stops meaning ‘number of macroconversions’ and
starts meaning ‘% of traffic which we gained value from”. At this
level, it doesn’t matter what that action was; what matters is these
are the segments where you should be focusing your marketing
• Dimensions can be anything. Channel and keyword are obvious;
what about time of day? City? Recency and frequency? All of these
dimensions start to become useful and actionable with the context
of “value per” metrics.
Upload content production costs
Utilise page-level custom variables and
SLIDE #41 NOTES
Content production costs, GA automatically re-calculates
per-visit value, etc.
• Retainer costs.
• Call centre? Actual costs of time spent on phone
answering questions where you have content which
• Lighting and heating? Fewer staff, lower costs…
• This is technically easy. It’s conceptually challenging, and
it needs bravery from high up to implement.
SLIDE #43 NOTES
If the cost of producing good content is (for arguments sake) £1 per word…
Due to SEO, brand perception, nurture impact, etc… (reverse engineer better numbers, work it out
loosely, refine periodically)
Then a 100 word UGC response to a blog post is probably worth 1/4 of that (lower quality, less
editorial, but valuable in pure equivalent content) is worth £25.
Tweak the maths so that it makes sense for you, review and refine it periodically.
Event-based goals, use (variable?) value
Group, categorise and structure
Dealing with fractional stuff; GA doesn’t allow for fraction values… Use server-side analytics (UA) to
keep tally of tiny events (e.g., individual blog post reads) and fire events at integer increments
Haters There’s value here
Generate value from people who’ll never buy your product or services.
Passives and haters can generate value too; UGC example, or links/citations/reviews
• Think about content in terms of value generated
• Lower the threshold for the measurement of success, get
buy-in from commitment to the ideals, and get proof of
concept in order to scale
• Put review points in place to sense-check downstream and
big-picture data to steer understanding, decisions and value
• Conversion rate isn’t a useful metrics; unless you’re only
marketing to your core audience. Look at value per metrics