A Plain-English Way to Tell What's Working
8 min read
Marketing measurement has a reputation for dashboards, acronyms, and consultants. For a small business, most of that is noise. You need answers to exactly three questions: which marketing brought people in, what did it cost you (mostly in time), and what should you do more of next month. That's it — and AI turns the raw scraps you already have into those answers.
Collect scraps, not dashboards
You don't need tracking software to start. You need a running note — the same discipline as your reality list — where you jot marketing outcomes as they happen. The email tool said 38% opened the spring email and 9 people replied. Four social posts went up; the oven-saga one got real comments, the tip posts got crickets. Two callers this week said "I found the furnace-noise page on Google." A customer mentioned the one-pager their neighbor gave them.
Thirty seconds per scrap, whenever one appears. If you do use platform stats, copy the two or three numbers that seem meaningful into the note — don't study the dashboard, harvest it.
The monthly review prompt
At the start of your monthly planning session (previous lesson), spend the first ten minutes looking backward:
Here are my marketing notes and results for [month]: [paste]
And here was my calendar for the month: [paste]
Answer plainly:
1. What clearly worked, and what's the likely reason?
2. What flopped or went unnoticed?
3. What can't we conclude yet because there's not enough signal?
4. Give me 3 concrete "do more / do less / try" changes for next
month's calendar.
Be blunt. Do not find silver linings in things that flopped, and
do not invent numbers I didn't give you.Question 3 is the one a cheerleading tool would never ask. One month of scraps is thin evidence, and an honest "too early to tell" protects you from lurching strategy-to-strategy over noise. Patterns you can trust take two or three months to emerge — which is fine, because your planning session recurs anyway.
Judge effort against return
The subtle version of "what works" is what works per hour. If the monthly web page took three of your twelve monthly marketing hours and produced two phone calls that became jobs, that's likely your best channel even if the social posts get more visible engagement. Tell the AI roughly what each item cost you in time and ask it to weigh results against effort — comfortable channels have a way of surviving on comfort rather than results, and this question flushes them out.
One caution as you get comfortable: resist the urge to optimize into paralysis. The point of measuring is to make next month's calendar slightly smarter, not to achieve certainty. Marketing for a small business is a repeated bet with modest stakes; a rough answer now beats a precise answer never.
That closes the loop. Voice document feeds prompts, reality list feeds channels, channels feed the calendar, results feed the review, and the review feeds next month's calendar. Every piece is a prompt you now own — and the whole system runs on about 90 minutes a week plus one monthly hour.
Try it now
Start the results note today, and backfill what you remember from the last month — even three scraps are enough for a first review. Run the review prompt at your next planning session. The first one will mostly say "not enough signal yet." The third one will quietly tell you where your marketing time actually pays.