Context Gap: the Root Cause of Every Annoying Marketing Challenge

Wrangling data isn’t marketing. It’s just annoying.

But nearly half of marketers say they spend more time segmenting and preparing data than any other task.

Each time you hit copy/paste, use VLOOKUP, or cross-reference dashboards, you experience the context gap.

The context gap is the mismatch between the explicit data a software tool has access to and the additional information it would need in order to independently deliver accurate insights or actions.

When it starts to feel like the main purpose of your job is to get the half-truths out of system A and connect them to the partial picture form tool B, it’s time to make a change.

When did it become okay for humans, with all our creative power, to use our time performing manual, repetitive actions to make up for the shortcomings of systems we’re supposed to be depending on for help?

I believe help is finally on the horizon. Model Context Protocol (MCP) might just be the thing that finally frees marketers from the drudgery of data wrangling. Before I tell you about that, let’s talk about why we need it.

Where Does the Context Gap Come From?

First of all, it’s always been the defining challenge of marketing to find insights that lead to impactful action.

We’ve probably all used or heard this old chestnut:

Second, and more recently, the number of gaps and the pain they cause have been amplified by the sheer amount of tools in our martech stacks.

Like the above quote, I know you’re familiar with Scott Brinker’s Martech landscape graphic. Now, in 2025, it lists more than 15,000 solutions. Why so many?

Answer: there are innumerable, separate challenges facing marketing teams. (Remember that phrase - I emphasized it for a reason).

• Precise targeting is hard because privacy regulations keep shrinking addressable data.
• Acquiring customers profitably is hard because paid-media auctions continue to grow more crowded and costly.
• Standing out is hard because customers have access to more channels with more content and more ads.
• Coordinating cross-channel campaigns is hard because retail-media networks and other walled gardens split data and budgets.
• Retaining customers is hard because inflation makes them price-sensitive and restless.
• Measuring true impact is hard because signal loss saps multi-touch attribution.

…and so on.

Each of those challenges needs a SaaS solution, right? Right.

And since each SaaS vendor doesn’t seem to quite solve the entire problem, they must need competitors, right? Right.

But despite (and partially because of) the enormous number of solutions on the market, every marketer I speak to still more or less asks me the same question:

Marketers’ challenges remain separate and innumerable.

If I can be so bold as to speak on behalf of the entire marketing community, it’s clear in retrospect that investment in point solutions was never a path out of this quagmire. Maybe we thought by running at the ‘innumerable’ part of the problem, we’d eventually solve it all. Instead, we made the ‘separate’ part of the problem worse by multiples, and actually increased the number of challenges we faced, too.

It’s a classic gordian knot: we cannot puzzle it out piecemeal. We need to cut straight through it.

We need to reframe how we define our challenges, stop thinking about them as a plurality, and view them instead as all stemming from one root cause.

As for the allegorical fish who don’t know they’re swimming in water until it’s pointed out to them, it helps to give this root cause a name.

The context gap is the underlying cause of all your pains.

What Does the Context Gap Look Like in Marketing?

Attribution

Difficulty with attribution is the most common and painful manifestation of the context gap.

Google claims the click, TikTok claims the view‑through, and your email platform swears the nurture sealed the deal. Each tool only sees its own slice of the customer journey, so they’re actually incapable of agreeing on the truth.

The problem isn’t that each platform’s local truth stays local. And it’s not that these data cannot be combined. It’s that people have to put their hands on the keyboard to do it, rather than spending that time deciding what to do with the more unified picture of what happened in real time.

Internal reporting

Do you start your week by copying numbers into a deck? The moment those numbers leave the source, they begin to drift from the truth.

Or finance pulls revenue from the ERP, ops teams pull leads from a CRM snapshot, and by the time slides reach the C‑suite, no one can trace the lineage. The drudgery exists because context lives in silos, forcing humans to play courier.

Strategy development

Annual planning should be about bold moves, creative thinking, resource planning, naming roadblocks and coming up with contingencies.

Instead, much of the agenda goes to data wrangling, exporting last year’s results, stitching CSVs, and hunting for forgotten insights in old reports. Most of the creative energy goes to working to find agreement on what happened last year, not what should happen next year.

That scatter is the context gap in action.

Experimentation

Most teams expect their testing work to be manual and channel-specific. They treat the pain as “just part of the job,” not as a fixable system problem. Under the hood, though, it’s the context gap:

• Platforms keep their own local truth (test designs, lift calculations, confidence scores).
• Nothing natively stitches those truths together or feeds them forward.
• Marketers do the stitching themselves, often without realising that’s what they’re doing.

There are plenty of tools that can chip away at the manual work to launch tests across platforms, pull data, combine results, but ultimately marketers are still acting as the glue between them.

Spreadsheets

Spreadsheets are so useful they’re hard to hate, but they’re the universal middleware of marketing, and they only exist because the real contextual fabric between other tools is missing. Every ‘export to CSV’ button is a platform’s admission that it lacks a seamless way to contribute to context itself.

To clarify, there’s nothing inherently wrong with spreadsheets. But they represent the need for a person to manually intervene in data organization, and while that’s been necessary in the past, it’s not the best application for human talent.

Context Dividend: A New Lens for Leaders

Reframe each stubborn pain point as an embodiment of the context gap and the fog starts to clear.

You can begin to close the gap. When you work towards curing the underlying disease, rather than just treating its symptoms, you unlock what I call the context dividend: every metric—ROAS, MER, incremental lift—begins to compound in your favour.

When context becomes more readily available, you get more than your time back. You also get better decisions, better alignment, better lessons-learned. You can wonder less about what you’re missing.

Most importantly, you get to actually do marketing again.

How Do We Close the Gap?

I’m not asking you to imagine how you’ll work in a context-rich future just for fun. I’m telling you it is coming, and you need to prepare yourself, your team, and your strategy.

Last year, Anthropic released Model Context Protocol (MCP). If you haven’t read about it, that’s okay, in the same way it’s ok that you may not have read about HTTP when the first version was finalized in 1996.

HTTP is the protocol that standardized how communication (from client to server) happens on the web. It’s what makes the internet work.

MCP is already being called “HTTP for AI”. It’s a protocol that tells LLMs like ChatGPT, Claude, or Gemini how to understand data from other tools in your stack. In other words, it gives the LLM context.

That’s abstract, so here’s an example of how you’d analyze a cross-channel A/B test right now, compared to how you’d do it with an MCP-enabled LLM:

Right now, it’s incumbent on the human to collect and organize the data to establish enough context to interpret A/B test results across channels (each ad platform offers its own analysis of how the test performed within that channel).

Without MCP, an LLM like ChatGPT could help calculate lift, but would still require the human to clean the data.

With MCP, the LLM has a framework for understanding how each different platform structures its data, and can understand each of those differing structures in the context of the single prompt from the user.

Where We Go From Here: Four First Steps

I don’t buy into the “AI can take your job” fear. Instead, I see AI taking your work, while you get to finally do your job the way you always wanted.

That change will happen fast, but not overnight. You’ve got time to prepare. Here’s how:

1. Orient your problem-solving around the context gap and plan to seize opportunities for growth when the gap begins to shrink, as it is sure to do. Try to adopt a context gap lens when you face annoyances and see if the framework fits. It may not. Just get in the habit of asking, “is this frustrating because my systems lack context?”
    
2. Audit your workflows and tech stack for context gaps. Find the places where a person is manipulating data. Is there a theme? A single gap that causes friction in more than once place? You don’t need to fill the gap immediately - just focus on becoming aware of it.
    
3. Explore existing options to close the gap. As I mentioned, many platforms are already investing in adding support for MCP. Shopify, for example, offers a Storefront Agent built on MCP.
    
4. Reallocate team time away from wrangling towards experimentation. Even if the experiments they run focus on improving workflows themselves, this is time well-spent and will prime them to be prepared to leave data-wrangling tasks behind, or delegate them to AI more aggressively when ready.

What will you do when these annoying context gaps start to dissolve?

You’ll still have challenges, but they’d be the fun ones: figuring out how to generate personalized creative at scale, finding new messaging that unlocks repeat purchases, strategizing how to reach completely new audiences.