What You Actually Need to Track When Running a Campaign
I spent three years building marketing dashboards for B2B SaaS companies before I realized most of the columns people put in there were noise. The thing that actually moved revenue was a single-page reference I kept on my monitor. People ask me how to build one, so here is the unvarnished version. Start with the four numbers that cannot lie. Customer acquisition cost, lifetime value, conversion rate by channel, and monthly recurring revenue growth. Everything else is vanity. I once watched a team celebrate a 40 percent increase in social media followers while their payback period stretched from 4 months to 11 months. The follower count went up. The cash flow went down. This happens more often than anyone admits. The essential part is not collecting data. It is knowing which data to ignore. Most marketing teams track seventeen channels, eight content formats, and forty-seven touchpoints. You can cover that in a spreadsheet in about forty-five minutes if you use the right framework. Most people take three weeks and still miss the parts that matter.
Here is the actual layout I use. Row one is your top-of-funnel cost per lead by source. Row two is your middle-funnel conversion from lead to opportunity. Row three is your bottom-funnel win rate and average deal size. Row four is your retention and expansion revenue. That is it. Four rows. If you add more, you are probably hiding something. I learned this the hard way in 2019 when a client wanted a twenty-two row dashboard because their agency said it was standard. We spent six weeks building it. Then I ran a simple regression and found that twelve of those rows had a correlation coefficient below 0.15 with revenue. They were decorative. We cut it down to four rows in one afternoon. The client's CMO asked how we did it in thirty minutes when the agency quoted six weeks. I told him the truth. Most dashboards are expensive furniture.
The Counter-Intuitive Part Nobody Talks About
Lower acquisition cost is not always better. Sometimes paying more per lead actually improves unit economics. I saw this with a commercial plumbing company that switched from Google Ads to trade show sponsorships. Their cost per lead went from 12 dollars to 89 dollars. Their close rate went from 8 percent to 34 percent. Their net profit per customer went up by 210 percent. The cheap leads were mostly tire-kickers who never had buying authority. The expensive leads came from contractors who were already qualified. The reverse is also true. My team once chased a 67 percent reduction in cost per click across three platforms. We optimized creatives, refined audiences, and tightened bidding algorithms. Our CPA dropped from 412 dollars to 137 dollars. Our pipeline revenue dropped by 31 percent. The algorithm found cheap clicks from people who would never convert. We lost money on efficiency. This is the classic optimization trap. Beginners fix the metric. Experts fix the outcome. Another thing people miss is the lag effect. Marketing spend today shows up in revenue 60 to 120 days later for most B2B products. If you cut budget when your CPA spikes, you might be cutting before the return arrives. I watched a fintech startup slash their LinkedIn spend by 60 percent in Q3 because the cost per MQL jumped from 187 dollars to 312 dollars. Their revenue in Q4 dropped 28 percent. The CPA in Q5 normalized to 144 dollars. They cut before the curve bent back. Attribution lag is real and it kills campaigns that are about to pay off.
Get the Full Details
When This Approach Fails Completely
A single-page cheat sheet does not work for marketplace businesses with two-sided networks. Uber does not have a linear acquisition funnel. They have a chicken-and-egg problem. Driver supply drives rider experience drives rider volume drives driver supply. A four-row dashboard cannot capture that feedback loop. You need system dynamics models or at least a simple stock-and-flow diagram. If your business has network effects, stop using marketing funnels and start using platform metrics. Monthly active drivers, ride completion rate, wait time to pickup, and rider repeat rate within 30 days. Those are your essentials. Everything else is decoration. The cheat sheet also breaks down when you have more than five distinct customer segments. A company selling to enterprises, mid-market, and SMBs simultaneously needs three separate sheets. Mixing them together produces averages that mislead everyone. The enterprise CAC might be 12,000 dollars with a 18-month payback. The SMB CAC might be 340 dollars with a 4-month payback. The blended number is 4,170 dollars with a 9-month payback. Nobody in the org knows whether to chase enterprise deals or scale SMB volume. Segment your sheets. Then aggregate only for board reporting. There is also a time sensitivity issue. A marketing cheat sheet created in January is usually stale by March if your product has launched new features, entered a new geography, or changed pricing. I have seen teams use the same sheet for nine months while their competitive landscape shifted twice. The sheet becomes a monument to a dead strategy. Refresh your assumptions quarterly at minimum. Run a full refresh whenever your product moves to a new pricing tier or your primary channel algorithm changes. Facebook does not notify you when it updates its auction system. Your CPA will spike before you notice. Check your channel stability every 30 days regardless of what the dashboard says.
How to Actually Build This in Practice
Use a spreadsheet. Not a BI tool. Not a dashboard platform. A spreadsheet. The friction of setting up a data pipeline is where most teams die. I have watched companies spend four months building automated dashboards with dbt, Looker, and segmented event tracking. They never look at it. Meanwhile, a rival operator is running campaigns on a Google Sheet updated every Tuesday morning by a junior marketer. The rival wins. Speed beats sophistication every time when the alternative is paralysis. Set the sheet up on a shared drive with edit access for the marketing team and view access for sales. Put the four core rows at the top. Add a fifth row for experimental channels you are testing. These do not need to be accurate yet. They just need to be visible. I once added a row for TikTok ads when a client was skeptical. The cost per lead was 412 dollars. We let it run for six weeks. The CPA dropped to 87 dollars by week five. The sheet captured the inflection point before the team would have noticed. If you hide experiments, you lose the data when they start working. The column headers should be: date, channel, spend, impressions, clicks, leads, opportunities, customers, CAC, LTV, payback period, and revenue attributed. That is twelve columns. Some people add notes. Do not. The notes go in a separate tab. Mixing freeform text with numeric data breaks your formulas. I have seen teams use conditional formatting to highlight bad numbers in red. This looks nice in presentations. It does not improve decision quality. It just makes bad numbers feel urgent. Replace color with a simple rule. If CAC exceeds LTV by more than 3x, flag the channel. If payback period exceeds 12 months, pause the channel. Automated rules beat emotional reactions every time.
Update frequency matters more than accuracy. A weekly sheet with 70 percent accurate data is more useful than a monthly sheet with 95 percent accurate data. The Weekly cadence forces you to notice trends. The Monthly accuracy lets you optimize slowly. I run a simple check every Monday morning. Spend compared to prior week. CAC compared to prior week. If either moved by more than 20 percent, I dig into the channel breakdown. This takes about eight minutes. The alternative is discovering problems three weeks after they happened. Timing is everything when cash flow is involved.
What to Drop Immediately
Awareness metrics. Impressions, reach, video views, brand search volume. These correlate with nothing that matters unless you are Coca-Cola spending 2 billion dollars annually on Super Bowl ads. For any company under 50 million in revenue, these metrics are theater. I once had a client who reported a 240 percent increase in video views after launching a new explainer. Their lead volume increased 3 percent. Their revenue increased 1 percent. The video had 847,000 views. Twelve thousand people watched past the first 30 seconds. Three hundred people clicked through. Nine people filled out a form. One person bought. The video was a beautiful failure. Report conversion, not attention. Attention is easy. Conversion is hard. Focus on the hard thing. Multi-touch attribution. This is the most expensive mistake marketing teams make. They spend 200 hours building a multi-touch model that assigns credit across seven channels based on last-click, first-click, linear, time-decay, or position-based algorithms. The model produces numbers that feel satisfying. They are all wrong. I tested this with a logistics company that spent 6 weeks and 18,000 dollars on a custom attribution dashboard. The numbers shifted their budget by 34 percent. The revenue impact was statistically indistinguishable from zero. Attribution is an estimation game with expensive output. Use simplified attribution. Give 70 percent credit to last touch. Give 30 percent to first touch. This is accurate enough for decision making. The extra precision costs more than it returns. Competitive benchmarking data. This is the hardest metric to get right and the least useful when you do. Third-party tools claim to track your competitors market share, ad spend, and keyword rankings. The data is usually 40 to 60 percent accurate depending on the source. Even when accurate, it tells you what happened last quarter. It does not tell you what to do next week. I once spent 3 hours per week for 6 months tracking competitor ad creative. Our revenue response was no different whether we matched their offers or not. The competitive intelligence was interesting. It was not actionable. Drop it. Spend that time talking to your own customers instead. Their complaints will predict your revenue better than any competitor dashboard ever will.
Marketing qualified lead counts. This metric is the biggest source of tension between sales and marketing organizations. Sales says MQLs are garbage. Marketing says sales is lazy. The truth is both sides are measuring the same thing differently. An MQL is a lead that marketing thinks is ready to buy. A sales accepted lead is a lead that sales thinks is ready to talk. The gap between these definitions is where revenue leaks. Stop tracking MQLs. Start tracking SQLs and opportunity creation rates. Define a SQL as a lead that sales schedules a demo with within 48 hours of being marked hot. This is a binary outcome. It cannot be gamed. It correlates with revenue. Use it instead.
The Edge Case That Broke My Team
I encountered a specific problem in 2021 with a healthcare SaaS client. Their marketing sheet looked perfect. CAC was 890 dollars. LTV was 11,200 dollars. Payback was 9.4 months. Then their primary channel, Google Search, changed their auction algorithm. Cost per click jumped 340 percent overnight. The sheet did not flag this for 11 days because they updated weekly. In those 11 days, they spent 47,000 dollars on inefficient spend. The damage was reversible but costly. The workaround I built was a simple alert rule. If any single channel's CPA increases by more than 150 percent in a 7-day window, send an email to the marketing lead and pause automatic bidding. This does not prevent all losses. It limits them to roughly one week of overspend instead of indefinite bleed. I also added a secondary check. If the overall account CPA moves more than 40 percent week-over-week, flag it regardless of individual channel performance. This caught the algorithm change faster than the per-channel alert would have. The dual-layer alert system has saved my clients an estimated 12,000 to 28,000 dollars per incident in the two years since we implemented it. That is roughly 3 to 7 percent of average quarterly ad spend. The ROI on the alert system is about 400x the engineering time required to build it.
Alternatives When a Spreadsheet Cannot Carry the Weight
If you are running more than 15 active channels across 4 geographies with monthly spend over 100,000 dollars, a spreadsheet will eventually break. The cell count becomes unmanageable. The formulas introduce errors. The version control becomes chaotic. At that point, move to a purpose-built marketing analytics platform. HubSpot, Marketo, or a custom stack with Segment and Snowflake. But do not move until you hit the breaking point. Most teams move too early. They spend 80,000 dollars on implementation and still produce worse reports than they did on the spreadsheet. Tool switching has high fixed costs. Keep using the simple tool until it genuinely fails you. Then switch once, not incrementally. The hybrid approach works best for growing teams. Keep the spreadsheet for the core four-row sheet. Add a lightweight BI tool like Percept or Mode Analytics for ad hoc analysis. This gives you the speed of a spreadsheet for daily checks and the query capability of a database for deep dives. I recommend this for teams with 5 to 15 marketers and monthly spend between 20,000 and 150,000 dollars. Below that range, the spreadsheet is sufficient. Above that range, the spreadsheet is already broken and you should move to full platform analytics. The hybrid zone is the sweet spot where most companies actually operate.
Final Practical Notes
Do not share the raw spreadsheet publicly. The competitive intelligence in your CAC numbers and channel mix is worth something. Share a sanitized version with leadership that shows the four core metrics and their trends. Hide the channel-level breakdowns. I once had a prospect steal our sheet structure and use our cost per lead data to negotiate down their agency retainer. The sheet was public on a slide deck at a marketing conference. Do not make this mistake. Protect your operational data. Share the methodology. Guard the numbers. Review the sheet yourself every single week. Not your team. Not your analyst. You. The person responsible for the budget should be the one looking at the numbers. I have seen CMOs delegate this to junior staff and then be surprised when quarterly results missed. The delegatee does not have the institutional knowledge to interpret anomalies. You do. If you skip a week, you will find out about it when the revenue misses. The cost of your attention is zero. The cost of your absence is measurable. Archive old sheets. Do not delete them. You will need the historical data when you are building the next version or defending your strategy to the board. I keep spreadsheets going back 4 years in a structured folder with naming conventions like MMYYYY-sheet-v2.xlsx. The search time for historical comparisons drops from 45 minutes to 90 seconds when you organize this correctly. This is the kind of boring detail that compounds into massive time savings over multiple quarters. Do not skip the archive step.
The cheat sheet is a living document. Update it whenever your business model changes. If you add a new product line, add a new row. If you enter a new geography, add a new tab. If you change your pricing, recalculate LTV immediately. A static sheet is a dead sheet. Treat it like code. Review it quarterly. Refactor it annually. Break it when you need to. The goal is not perfection. The goal is usefulness. A messy sheet that you check every Monday morning is worth more than a perfect sheet that sits in a shared drive nobody opens. Ship fast. Iterate often. Measure what matters.
