Marketing Isn't Failing. Your Expectations Are.

Emily Goh · 2026-05-09

Marc Benioff said it at a scale most people don't say anything at.

"None of these agents work without data and context, or you just get all the hallucinations."

He was talking about AI. But swap "agents" for "marketers" and the sentence still holds.

None of this works without clean data and real context. And most marketing teams are running campaigns, building funnels, and defending pipeline numbers on data that is incomplete, inconsistent, and -- in most cases -- nobody has audited in over a year.

This is not a marketing performance problem. It is a foundation problem.

The target was not made up. It was just never grounded.

To be fair: nobody wakes up and invents a number to torture their marketing team with.

What actually happens is this.

The board sets a revenue goal. Finance builds a model. Sales works backward: if we close at 25%, we need $40M in pipeline. Marketing owns 60-70% of inbound. That is $24-28M in marketing-sourced pipeline. Marketing gets the spreadsheet.

What was not in the model:

• Last year's actual conversion rates -- which were off by 40% because attribution broke in March when someone reconfigured HubSpot
• The list quality -- which has 22.5% annual decay by default, so 30% of contacts from last year are already stale
• The blank campaign fields -- 47 opportunities in Q4 with no campaign source attached, which means they fell out of the attribution model entirely
• The channel mix -- which changed twice when the paid ads account got suspended in August

Marketing did not receive a goal. Marketing received the output of someone else's assumptions -- and was asked to build a funnel on top of it.

Reverse engineering a fictional target in real time

So marketing does what marketing does. They take the number and work backward.

We need $24M in pipeline. At a 5% MQL-to-opportunity conversion rate (last year's number -- probably wrong because attribution was broken), we need 480 opportunities. At a 15% lead-to-MQL rate, we need 3,200 leads. Across the campaigns we can run this quarter with a $60K budget.

Except:

• The 5% conversion rate was calculated before the SDR team churned three people in Q2
• The 15% lead-to-MQL rate was from a quarter where they ran a list import that inflated the count by 43%
• The $60K budget has to cover events, paid, content production, tools, and the conference that was just added to the calendar

Marketing runs the math. The math does not work. Marketing runs the math again with more optimistic assumptions. Presents the plan.

"Great. We'll need pipeline by Q2."

New logo acquisition takes a village. We keep asking one person to build it alone.

Getting a new customer who has never heard of you to become a paying customer requires four things working together:

Product -- something genuinely worth buying. A value proposition that is differentiated and provable.

Messaging -- language that connects the product's value to the buyer's specific pain. Not generic. The exact words that make a CFO stop scrolling.

Marketing -- the infrastructure to get those words in front of the right people, at the right time, in the right channel, enough times to create recognition and intent.

Sales -- the human layer that converts intent into a conversation and a conversation into a contract.

All four have to work. When new logo acquisition fails, it almost always fails because at least one of these four things was broken -- wrong product for the market, wrong message for the buyer, wrong channel or timing, or wrong follow-up.

But marketing gets the quarterly review.

And the data that would let marketing explain exactly where the conversion dropped -- which stage, which channel, which message, which segment -- is the data that is most likely to be incomplete, inconsistent, or missing entirely.

Campaign launched in a day: the brief arrived Tuesday

"Can we have something live by Friday? It's just an email and a landing page."

Let's count what "just an email and a landing page" actually requires:

1. A clear, differentiated message (requires product alignment, customer insight, competitive context)
2. A subject line that gets opened (requires testing or at minimum a validated hypothesis)
3. A validated list (requires checking for stale contacts, bounce history, unsubscribes)
4. A landing page with a clear CTA (requires design, copy, form, confirmation email)
5. UTM tracking so the campaign appears in attribution
6. Sign-off from legal, product, or leadership if any claims are being made
7. A follow-up sequence so the lead does not just sit in a queue

That is not a Friday morning deliverable. That is a two-week project -- or a two-day sprint that skips steps 2, 3, 5, and 6 and wonders why nothing converted.

"It's just an email and a landing page" is the most expensive sentence in B2B marketing.

The event confirmed Monday for Friday

Slack, Monday 9am:

"Hey -- we just locked in a booth at [Major Industry Conference] this Friday. We need collateral, outbound to attendees, a speaker session, and a dinner for 12. Marketing, can you own?"

Marketing owns it. Something gets printed. Someone books the dinner. Outreach goes out Wednesday. Three people reply. The booth looks fine from six feet away.

Monday: "How many leads did we get?"

Eleven badge scans. Two were competitors. One was a student.

"Events aren't producing pipeline for us."

The event did not fail. The timeline failed. And marketing absorbed the consequence.

80% inbound for SDRs. Then a sequence that goes nowhere.

Marketing delivers. 240 leads. 80% ICP match. Sales enrols them in this:

Day 1:

Subject: Quick question Hi [First Name], I wanted to reach out because I think we could be a great fit for your team. Are you the right person to speak with?

Day 4:

Subject: Re: Quick question Just following up on my note from last week.

Day 9:

Subject: Re: Re: Quick question Touching base one more time. We help companies like yours streamline their operations. Happy to share more. 15 minutes?

The lead does not reply. Marked as "no response."

QBR: "The leads from marketing aren't converting."

A three-touch, nine-day, zero-personalisation sequence on a lead who filled out a form about a specific problem is not a lead quality failure. It is a sequence quality failure.

The lead raised their hand. The follow-up gave them no reason to raise it again.

The attribution model changed three times. The data never did.

Every time the attribution model changes, the historical data does not update to match.

First-touch becomes last-touch. Last-touch becomes multi-touch. Multi-touch becomes a custom model nobody documented. And every time it changes, the "marketing-sourced pipeline" number changes -- not because more or fewer deals came from marketing, but because the definition changed.

Now at the board meeting:

"Marketing sourced $4.2M in pipeline last quarter." "But last quarter you said $6.1M." "That was under the old attribution model." "So which is right?" "...it depends on how you define 'sourced.'"

This is not a marketing credibility problem. It is a data infrastructure problem. But marketing absorbs the credibility loss either way.

What Benioff understood that most leadership teams haven't yet

When Benioff says AI cannot work without clean data and context, the implication is not subtle.

Every downstream system -- AI agents, sales sequences, pipeline forecasts, marketing attribution, campaign personalisation -- is only as good as the data underneath it.

Salesforce's own partner ecosystem has called dirty CRM data the number one barrier to Agentforce actually delivering results. The most sophisticated AI layer in enterprise software fails when the records underneath it have wrong owners, stale contacts, blank fields, and duplicate accounts.

This is exactly what happens to marketing.

Marketing cannot build a credible funnel on inaccurate conversion rates. Cannot defend pipeline on broken attribution. Cannot prove ROI when half the opportunities have no campaign source attached.

The problem is not the marketer. The problem is the foundation the marketer is building on.

What actually needs to change

Stop setting targets from the top down without auditing the data that should inform them.

Run the funnel math on last year's actual conversion rates -- not the aspirational ones. Validate the list before the campaign launches. Standardise the attribution model and leave it alone for at least four quarters. Document the campaign source for every opportunity.

Make the sequence quality as visible as the lead quality. If marketing can see that 240 leads went in and only 90 were worked, and only 15 of those got more than one touch, the lead quality conversation changes immediately.

When the foundation is clean, marketing can prove what is working. Defend what is not. Fix the right things.

When it is not clean, marketing absorbs everything that goes wrong -- even the things that were never theirs to fix.

For the full picture on how bad data breaks revenue operations on the last Tuesday of every month, read this.

Want to try it yourself? Gohvio is free to try -- gohvio.io