How to Align Sales and Marketing at a Startup

1. Blog
2. Ops, Teams & AI

By Saroj Jha · October 5, 2026 · 9 min read

Sales and marketing are aligned when both teams work from one written definition of a qualified deal, one set of deal stages built around how the buyer decides, and one short library of material that reps actually use. The four steps below show how, with one example company.

In my view alignment is a design problem, not a mood. The two teams usually disagree because nobody wrote down what counts as a real deal, what must be true for it to move, or which material belongs where.

The gap is common. In a Gartner survey of 412 senior marketing and sales leaders, the two functions "typically collaborate on only three out of 15 commercial activities", and "90% of marketing and sales executives report their functional priorities conflict with one another"^[1]. The research also points to structure, not slogans. A study of 337 EU-based companies suggests that "the most successful configurations are characterized by strong structural linkages between marketing and sales"^[3].

Running case: Example Co. Fictional

Example Co. is a fictional seed-stage B2B SaaS startup. Its software automates the month-end close for finance teams at companies with 50 to 500 employees. The controller is the champion and daily user, the CFO or VP Finance signs, and IT reviews security. The team is 8 people, including two founders and one account executive. Price: about $1,500 a month ($18,000 a year); a typical deal takes about 45 days; the goal is 30 new customers in 12 months. Marketing (one founder) counts every webinar sign-up as a lead, the account executive says most are not real, and deals stall once the CFO gets involved. All numbers are fictional.

Step 1: Write one definition of a qualified opportunity

Start with the word both teams fight about. Gartner found that "62% of respondents describe their sales and marketing functions as defining qualified leads differently", and that sales organizations that "align cross-functional KPIs are nearly 3 times more likely to exceed new customer acquisition targets"^[2]. These are survey associations, not proof of cause, but the fix is cheap: one sentence both teams sign. It names the fit (who), the problem (why now) and a buyer action (proof of intent), so two people reading the CRM record would agree. "Had a good call" is a feeling, not a definition.

Example Co., step 1 Fictional

The founders write: "A qualified opportunity is a company with 50 to 500 employees whose controller has described a close that takes longer than they want, and who has agreed to a second meeting that includes the CFO or VP Finance." Marketing stops reporting webinar sign-ups as leads and reports qualified opportunities instead. The illustrative target: about 120 qualified opportunities in 12 months, assuming one in four closes, to reach 30 customers.

What HubSpot's filings show

HubSpot's annual report describes marketing and sales as one model: "We believe that our marketing and sales model provides us with a competitive advantage, especially when targeting mid-market businesses, because we can attract and engage these businesses efficiently and at scale." Its marketing team "attracts over 57,000 new leads per month", "Inbound sources generated over 90% of our new customers and over 83% of new leads during 2015", and "The vast majority of revenue generated by our sales representatives originates with inbound leads produced by our marketing efforts"^[4]. Tracing lead source through to customers only works when both teams count the same thing. This is what the filing says, not an account of how HubSpot made its choices internally.

Do this

Write your qualified-opportunity sentence this week, put it in the CRM as a required field, and have both team leads sign it.

Step 2: Build the stages around how your buyer decides

Design the deal stages around buyer decisions, not your internal handoffs.

Keep it to four to six stages. Give each a verifiable exit criterion, such as "economic buyer confirmed", plus what the rep and the buyer each do to move forward. Then add a stall rule: a deal with no buyer action for a set number of days gets reviewed, not left to age.

If you sell an AI product

In my view, AI buyers add decisions your stages must show. They want to see the product work on their own data, and someone has to approve what data it sees. Make each one a stage with a buyer-based exit criterion, such as "pilot success measure agreed in writing". Example Co.'s working test connection is the same idea at a smaller scale. Then marketing and sales must agree what a pilot is and when a trial user counts as a qualified opportunity, or the old lead fight comes back under a new name.

Palantir's annual report for 2025 shows how central that test can be. It describes "our use of AIP bootcamps, which allow us to deliver real workflows on actual customer data in days," and says it often provides its platforms to potential customers "at no or low cost initially to them for evaluation purposes through short-term pilot deployments of our platforms, including at bootcamps"^[5]. This is what the filing says, not how the company decided internally.

One more rule, in my view: AI claims can change when the model changes. Keep one list of approved accuracy and data-use claims, owned by both teams, so the website and the rep's deck never disagree.

Example Co., step 2 Fictional

Five stages, built around the controller, the CFO and IT. Discovery ends when the controller confirms the problem and a target close date. Evaluation ends when the CFO has seen the demo and a test connection to the accounting system works. Security review ends when IT signs off the questionnaire. Business case ends when the CFO agrees the numbers and a dated plan to signature. Contract ends at signature. Stall rule: 14 days with no buyer action triggers a review. All fictional.

Do this

For each stage, write the exit criterion as something a buyer did, and add one stall rule to your CRM.

Step 3: Audit what sales has for each stage

Now marketing's job is clear: supply what each stage needs. I use ten needs across three stages. Discovery: a one-pager, a discovery guide, email templates. Evaluation: a battlecard, a case study, a demo script, an objection guide. Decision: a business case template, a security and legal pack, a mutual action plan (a shared, dated list of the steps both sides take to signature).

Check every asset three ways. Does it have an owner? Was it updated in the last six months? Did at least a third of reps use it in their last five deals? Those thresholds are my working rules, not research findings. Ask the reps; download counts are not use in a deal.

Write for the whole buying group. Gartner found that "Tailoring for buying group relevance positively impacts consensus by 20%", while content focused on individual-level relevance had "a 59% negative impact on buying group consensus"^[6]. And because "67% of B2B buyers state that they prefer a rep-free experience"^[7], the business case and security pack must make sense when the controller forwards them without a rep in the room.

Example Co., step 3 Fictional

Six of the ten needs have an asset, and only three pass all three checks. The discovery guide has no owner, the objection guide is eight months old, and the security pack exists but the account executive did not know about it. There is no battlecard, no case study (no customer has yet given written permission), no business case template and no mutual action plan. All fictional.

Do this

List every sales asset with its owner, last update and how many reps used it, then map it to the ten needs.

Step 4: Rank what to build by the deals you lost

Do not ask marketing for more content. Ask what would have saved the deals you lost. Code two quarters of closed-lost reasons, total the value lost under each, and map each reason to the assets that would have helped. Price points to the business case and the objection guide. A competitor loss points to the battlecard and a case study. No-decision points to the business case and a mutual action plan. Security points to the security pack.

Then fix before you build: an owner and a refresh cost hours, a new asset costs weeks. If an asset exists, is used, and you still lose for that reason, the problem is the message or the deal, not the library. A missing-feature loss is a product or positioning question.

Example Co., step 4 Fictional

Over two quarters (fictional), Example Co. lost 9 deals worth $162,000: 4 no-decision ($72,000), 2 security ($36,000), 2 price ($36,000) and 1 competitor ($18,000). The business case template sits behind $108,000 of losses, so it is the first build. The security pack and objection guide are fixes for this month: refresh the objection guide and walk the account executive through the security pack.

Do this

Run this ranking at a joint sales and marketing review every quarter. The list should change.

Free tool

When competitor losses point at the battlecard

The free Battlecard Builder builds a competitor battlecard reps can trust: proof for every claim, with sourced and dated facts.

What changes by stage?

• Pre-seed — The founder is both teams. Write the qualified-opportunity sentence anyway and log why deals were lost.
• Seed — With a first account executive, agree the stages and exit criteria before the hire starts, and fix the discovery and decision assets first.
• Series A–B — With several reps, adoption becomes the hard part. Run the three checks quarterly and give every asset a named owner.

What mistakes should I avoid?

• Measuring marketing by leads created instead of qualified opportunities and deals moved.
• Stages named after internal handoffs ("passed to sales") rather than buyer decisions.
• Building new assets before fixing stale ones. A stale battlecard in use can do more harm than a missing one.
• Publishing a case study without the customer's written permission.

Example Co. is a fictional company. Example Co. and all its numbers are illustrative.