How we Generated $250M of Lifetime Traffic Value (LTTV) in 2021
The strategy, economics and operations of an emerging inbound engine
Last year we added over $50M of annual organic traffic value to ibm.com. If you consider that many of these pages will rank for a long time, let’s say five years (3 and 5 year models are normal caps for SaaS LTV models), you get to a quarter billion of incremental lifetime traffic value (LTTV) we earned last year. That’s not a perfectly scientific model (doesn’t include churn/decay, cost of any optimizations or discount rates), but that’s not really the point. The point is that it’s a pretty big number.
The reason I like to use LTTV to talk about impact and return is because organic traffic behaves a lot like MRR/LTV in a SaaS business. You might pay $5k to acquire a customer. They might only bill $1k in the year you acquired them. But they also might stay for 10 years and increase their spend every year they are with you.
That reminds me a lot of organic. Organic traffic has something like CapEx cost model (large up front cost in building a piece of content) and an MRR/LTV return model (returns that can happen over many years given strong urls can rank for over a decade). For those reasons, the ROI model is less ROAS and more LTV:CAC.
Despite what the chart says, this is one of those overnight success stories that has been work in progress for several years. It all just seemed to come together last year (or at least the numbers did).
As a result, it felt like a good time to talk about the program, what we learned and what lessons other aspiring serperstars might take from what we’ve done.
What I won’t cover as part of this are the obvious things that every program requires to be successful and that this one did too. It required leadership support. It required a great team. It required investment. We were fortunate to have all of those.
When I thought about what we could share that might be interesting or novel, about what I could share, I think it’s really about the strategy, the economics and the program. If you want the latest and greatest around technical SEO, how the algorithm is changing, etc probably every other SEO blog and website on the internet will be better than this one. But if you want to learn about how to build a a pretty big inbound program, then this post, and maybe even this substack, will (hopefully) have something for you.
Success begins with (un)spectacular failure
IBM’s marketing programs historically have been oriented around paid media and the combination of events/field marketing/ABM.
Different efforts had been made to build a real inbound engine for probably the last ten years or so. Once such (failed) effort was led by my team in our cloud business unit probably five years ago or so.
Our team was young, analytically minded and had no real-world experience in SEO. We were still determined to make a go of it. We built something we called the Inbound Workbook.
It was a fancy excel document that someone on my team created. The document and the underlying thought process got a lot of (well, some) attention and interest within IBM.
We were making waves as being ahead of the curve and bringing more modern methods and tools to IBM.
And as a result of the Inbound Workbook, we shipped zero urls. Zero.
I’m sharing the story not because I want to contribute to the failure is good bumper sticker, but because we did learn some really important lessons.
The first was that IBM actually had really great internal SEOs and we didn’t consult with any of them. They could have helped us. Ego prevented us from making the call and it’s why I have complicated feelings about the bumper sticker around failure. Is failure ok when it’s due partly/largely to ego? You could argue it is good if you walk away with the gift of humility, but I still don’t know. I still often feel like we failed in ways we shouldn’t have.
We knew that we wanted to be on page 1 for things related to “mobile” but didn’t understand any of the nuance around which pages, with which intents, should target which clusters of keywords. When we did build a page, what content actually needed to go on that page? When the page existed, how did it need to get promoted and linked?
These are all questions that actually talking to our SEO team would have helped answer.
But those gaps around the (essential) details of execution weren’t our only problem. We had major gaps in the quality of our operational thinking.
Our plan was to take the workbook to each of our teams, explain to them how to use it and then, magically, they were just going to start planning and executing their SEO programs and outcomes would fall from the sky.
Here are some questions we didn’t answer…
who is the person doing the keyword research?
who prioritizes that backlog of content?
who actually writes the content?
given the team doesn’t have infinite bandwidth, what tradeoffs are we recommending?
does our executive team support those tradeoffs?
what’s the financial model that supports those tradeoffs?
how does any of this fit into our broader demand gen model?
I know it’s shocking we didn’t accomplish much (anything). But that experience helped me to understand the difference between high-level ideas and actual strategy. A real strategy for building an inbound program couldn’t be an excel file we just handed to other teams.
Strategic thinking needs to include operational thinking
One of my favorite business books ever is A.G. Lafley’s (former CEO of P&G) How Strategy Really Works. In that book, he positions strategy as a series of five distinct cascading choices that a business must make.
These choices don’t stop at high-level choices and capital allocation, but go down into the capabilities and management systems of the organizations, details that are often considered execution.
The reason I think it’s important to consider these as part of strategy is that you might allocate your capital entirely differently depending what the lower-level operational model looks like. The expense and return model could change dramatically.
For example, simple question. Do we expect people already employed by IBM to create all the content we need for SEO? If not, who is doing it? Are we hiring new full-time employees? Freelancers? Agency?
What’s the cost model of each of those choices? How do each of those choices impact processes and management systems? If an agency is building the content, who reviews it?
If we wanted to build a successful program, all these questions demanded answers. Our answers looked something like this…
What is our winning aspiration - We want to be the biggest unbranded inbound program in all B2B tech
Where will we play - We will compete broadly across the scope of IBM’s business (200+ topics/categories) and deeply within select categories (~10-20).
How will we win - IBM’s brand and the domain’s backlink profile are natural competitive advantages. We will use the topic cluster model as the foundation for the broad and deep approach.
What capabilities must be in place - Centralized team of SEOs, editors and writers. SEOs and editors should be in-house while writers can be a blend of IBM and 3rd party.
What management systems are required - Systems that support topic identification and prioritization, content production and reporting/performance
To understand why we made some of the choices we did, it helps to understand how IBM was structured internally at the time (it’s very different now). IBM was in something we called the diamond team model (it’s similar to squad model, two pizza team, etc) and our structure looked something like this…
We had many marketing divisional heads (~10) and each of those divisions had a handful of diamond teams. Each of these diamond teams was semi-autonomous in their operations. There are a lot of problems with this approach but at a large company it’s at least (mostly) obvious how a team would come together to market a product or service.
The problem with this model is that you now have something like 40 distinct diamond teams. If you want them all to do something consistent (eg approach search differently), it’s hard to do that. The model is about the autonomy of the individual team so short of an unflinching mandate from our SVP, you aren’t going to get every team to do something.
And let’s say you want to improve SEO across all IBM, short of that mandate, you actually need to convince 40 distinct teams to change their behavior. Convincing people to change their behavior is very difficult because it starts with the team having to acknowledge they are not making all the right choices today. So you have some major ego and territory issues.
In our original model, we didn’t solve for any of this. We were just going to give the teams the Inbound Workbook, teach them how to use it and then we would wake up some time in the future and be good at SEO.
That was obviously a fantasy.
During our second attempt at building an inbound program, we centralized much of the execution to avoid all of the operational complexity associated with trying to change the behavior of 40 teams. Because that was our strategy, it meant we needed certain capabilities within that central team (SEOs, editors, a pool of writers). We (successfully) argued for budget. We changed people’s jobs. We did everything we had to do to ensure that the high-level aspirations were connected to the strategic choices we were making 3, 4 or even 5 rungs down strategy cascade.
I think one of the traps that’s easy to fall into is to present something like the Inbound Workbook and say hey, this is a tool to help you organize yourself around this obvious digital marketing discipline. And then no one does anything and you blame them for not understanding the value of SEO and failing to organize around it.
It’s not them that’s trying to build an SEO program, it’s you. And if you are the one driving then it’s up to you to think through all the choices that need to be made, to involve your management team throughout the process and to ultimately own the strategy and the outcome.
The TIPI framework as our methods anchor
At the end of 2021, we were on page 1 for 80% of the primary keywords that we were targeting and had live urls for. We have a good batting average because we have methods for all the madness.
Every system has constraints and so does SEO. In SEO, intent is probably the biggest constraint. You want to be on page 1 for some high-volume search term with your product page but guess what, it has informational intent. If you want to be on page 1 for that term, you need to accept the constraint and build an informational piece of content.
The more we understood the fundamentals of SEO, the more we learned how to translate those low level primitives into higher levels of abstraction and strategy. The model we are playing around with now (which is the subject of a longer, now published post) is the TIPI framework.
If you understand something as basic as topics and intent, you have the foundation for designing entire digital experiences.
Our approach goes like this…
Topics - identify the core topics associated with your business using concentric circles of relevance going from 1:1 category match up through anything of interest to my audience
Information needs - pick out the common query modifiers associated with those topics (eg best, free, what is, solutions, etc) and then catalog the user intent associated with each of those information needs
Page types - for each intent type, design a corresponding page type (this is the one that we’re still exploring as we venture deeper into headless content/content types and things like object oriented UX)
Information architecture - cluster query types and page types into identifiable stages of a buyer journey and then prioritize linking within stages of the journey via menus and nav, contextual linking, in page objects, etc.
The TIPI framework gives us structure to how we think about the markets we’re in, how we discover the common information needs that exist across those markets and how we design web infrastructure and experiences that can consistently deliver value to clients and to IBM within that context.
It is foundational to both how we plan our content backlog and how we design content models, page types and linking patterns within our CMS. It’s a system for both relevance and engagement.
We have a long way to go to get to our fully realized vision (including developing a fully realized vision), but it’s the tool we’re using along the way.
Building the future required owning the past
Early in our second attempt we faced some failure to launch challenges. I originally thought the problem was optics, but it was actually about having the wrong mentality about being a builder when you aren’t starting from scratch.
A few years prior the team had built a somewhat limited SEO program in our cloud business and then for reasons I don’t recall, or wasn’t privy to, the program was closed down. There were a few pages from that time that had secured some pretty good traffic that had persisted despite the lack of focus.
We made the decision to build the content for our new program in the same directory as that older content which made sense from a web architecture perspective, but made reporting either harder or better, depending on how you looked at it.
As we tried to build momentum around the new program we were reporting on the directory and it just was not going well.
We started this new program in Q4 of the prior year and so we are a solid ten months into the program and it seems like we are flat. What had all this strategy and funding and people actually delivered? Nothing?
“Well actually” we would say and then produce a chart like this one…
The green line was the old pages and the blue was the newer pages. The new content was legitimately doing well and we were nearing the point where the new program was going to easily offset/surpass any losses from the legacy program and produce top line growth.
But this was an uncomfortable time. We were almost a year in and showing net flat numbers. It seemed like if we had done nothing at all the business outcomes would be the same.
We did two things as a result. We kept building the new content, but we also devoted a group of people to optimizing a chunk of the legacy content in the directory and fix a handful of technical issues, like redirects that had expired.
The outcome is not only did we accelerate the blue line, but we stabilized the green line and returned it to (modest) growth. And the combination of those two things is strong top line growth.
At first I was annoyed with the performance of this older content. It was obscuring our success and casting doubt on our work.
But the longer I thought about, the more I realized that it was right and good for the past to be dragging on us like this. Because that’s real life.
My job wasn’t to own this project. It was to own inbound for this business. And that meant managing things I inherited and things I needed to build. If I wasn’t a good steward of the world that came before me, it was only accurate for that to drag on the top line numbers. We shouldn’t have been able to claim victory just because we had a new thing growing if we also allowed older, successful pages and content to wither and die under our watch.
Sometimes the right approach is to let something go but that should be a conscious strategic choice, not an issue with “not invented here” syndrome.
As our goal later became to deliver strong growth to all of ibm.com, we took on much more of a portfolio management mindset. Growth means making sure we can hold or gain position on key existing terms across the domain while adding new topics and query types into the mix.
We were focused on building a movement, not just a program
This wasn’t just about ranking for some keywords for us. It was the foundation of a different way to do marketing. It was about education. It was about organic channels. It was about different economic models. It was about pulling IBM in a different direction in terms of how we engaged with our customers and prospects.
So for us it wasn’t just about building a program, it was about building a movement.
To do that, we had layers of communication and community interaction that helped educate people about what we were doing, why, and the results we were seeing along the way.
We adopted a three-level comms plan that helped build momentum around the program.
Level 1 (the close collaborators) - this was a community of around 50-75 folks that we worked closely with and whose work was impacted directly by ours. They wanted to feature links and content on your pages. They wanted topics related to their business to get into our backlog. We had a slack channel for this group (#inbound-empire) and for a solid year or so I would post almost every morning remarking on whether yesterday was a new record (it usually was), new keywords we got on page/position 1 for, other exciting evidence of progress like copyright requests or backlinks from extremely popular 3rd parties (like Seth Godin). It was a very informal channel of mostly quips and screenshots. But it was there to build and sustain interest and momentum. And make the people closest to us most familiar with our strategy and progress so that they might be effective advocates for us.
Level 2 (leadership) - every month we put together a couple summary slides that showed up and to the right progress. What really made a difference was when we started including 50-100 keywords on the chart as examples of where we were on page 1. The actual words mattered a lot, particularly with stakeholders outside marketing. Business unit leaders and product teams wanted to see that we were on page 1 for the key term(s) in their category. The keyword (and them subsequently doing a search to double check) was what made it real for many folks.
Level 3 (the masses) - the precursor to this blog was an internal site called the Cookie Pool. Cookie Pool had a blog, a “guides” section for in-depth and structured learning around a topic area and a speaker series called MarTalk. All of it was about sharing what we were learning and doing as we were doing it. It was a reasonably “technical” site as well, focusing on more advanced topics. The goal was to both project sophistication and to genuinely nurture a more curious, more expert community of marketing practitioners. Several of the more popular posts got thousands of views and are still regularly read years later. We would often post content in Slack channels that had thousands of people. We were very proactive at taking our message out into the world.
Each of these audiences required a different strategy and different content. Taken together, the impact of those three strategies is that we have a program that now has broad, robust support and where the narrative about where we are and how we’re doing is consistently very different than it was a couple years ago, regardless of whether me or anyone from my team is in the room or not.
We still have much further we can go on this journey, but we are doing reasonably well bringing the broader marketing function at IBM along with us.
Nothing more important than delivering
For all the fancy talk about strategy and failure and communications, nothing mattered more than simply delivering.
When our leadership could go to our CEO and tell him that it wasn’t pockets of success here or there, but that ibm.com top line is up dramatically, that was a game changer. It bought instant credibility in a way that no deck on theory ever could.
And with that credibility came more budget and more responsibility and more willingness to bet on the next thing, because now we have this track record of success. When we say we need something, the default is now to believe us.
And beyond that, delivering successfully has also just been great for culture. Everyone wants to be part of a winning team. IBM has transformed itself many times over the years and it’s not always easy or fun. But it’s a lot easier and a lot more fun when you start to see evidence of that transformation, when you start winning at things you didn’t used to.
And when you deliver against one thing successfully, and take aim at the next hill, the team now believes you can take this one too. And the next one. And the next.
That’s powerful and a real momentum builder.
If we have done anything, it’s create belief across the team in our ability to compete and deliver value at the highest level. And that’s exciting. That’s the work we all want to be doing. And it’s the right expectations to have for ourselves and for one another.
More to come…
This is still only the beginning, both for this program and for this blog. If there’s anything you found particularly interesting, or things you’d like to hear more about, or even things you thought were spectacularly boring, would love to hear from you so I can get better at this as I go.
Also feel free to follow me on Twitter if you want to watch me embarrass myself as I try to learn how to use that platform in a useful/interesting way.
Biggest takeaway for me (that I've seen as well) was reporting on a cohort of content as if it were a campaign. SEOs and content folks struggle with this, but it's KEY to getting the internal momentum we need for more buy-in and investment. Loved the "here's what we should have shown" graph.
Damn it, Nice Case Study!