Measuring what a prospecting flow finds each week, and what the numbers mean
Four weekly numbers that show whether a lead-finding AI employee earns its keep: new finds, strong fits, reachable leads and repeats, and how to judge them.
Judge a prospecting flow each week on four numbers: how many genuinely new businesses it found, how many of those were strong fits, how many you could actually reach (a phone, a website or a named contact), and how many you skipped because they were wrong or already known. In xMap Flows the first two are on the employee's weekly report card and in the Monday email; all four can be counted from the Results page by asking in a sentence. Compare each week with the one before and with what the market can realistically produce.
This article is for sales managers and founders who have set an AI employee to find leads and now need to know whether it is worth keeping, changing or letting go. It explains each number, how to set a fair expectation from the size of the market, a made-up month of readings, and the measurement mistakes that make a good flow look bad.
The four numbers
| Number | What it tells you | Where to read it |
|---|---|---|
| New this week | Whether the market is producing anything for this job | Report card (found), Monday email, Results with the period set to 7 days |
| Strong fits | Whether the scoring separates good from average (a score of 75 or more) | Report card, Results filtered by score |
| Reachable | Whether a rep can act without more research | Ask Results: how many new this week have a phone or website |
| Skipped | How much noise reaches you | Your own tally, or a saved view of what you decided not to pursue |
Results in xMap Flows have no statuses, deal values or reminders, so measure what the flow finds here and keep sales outcomes in your own CRM.
Setting a fair expectation from the market
A flow cannot find more new businesses than open. Before judging it, look at how big the pool is and how fast it changes. xMap's places database lists 246 gyms within 10 km of the centre of Toronto, and of those 246 with an opening date, 14 opened in the twelve months before November 2026 (November 2026 figures). A flow watching for new gyms in central Toronto should therefore expect about one a month, not one a day. A week with nothing is normal there.
Bakeries move faster. Of the 366 bakeries nearest the centre of Toronto with an opening date, 28 opened in the same twelve months, according to the same database. That is roughly one every couple of weeks within that sample alone, so a bakery-supply flow covering the whole city can reasonably expect more.
A made-up month of readings
Say a coffee equipment supplier in Toronto runs a Sales scout that looks for new cafes and bakeries every morning and scores them. Over four weeks the manager writes down:
| Week | New | Strong fits | Reachable | Skipped |
|---|---|---|---|---|
| 1 | 11 | 3 | 8 | 4 |
| 2 | 6 | 3 | 5 | 1 |
| 3 | 14 | 2 | 9 | 9 |
| 4 | 5 | 2 | 4 | 0 |
A made-up example, not real figures.
Week 3 looks like the best week by volume but was the worst: nine of fourteen were skipped. Looking at them, most were bubble tea and juice bars that the search counted as cafes. The manager narrows the kinds of place, and week 4 is smaller and cleaner. Strong fits held at two or three a week throughout, which is the number the reps cared about. That steadiness says the employee is doing its job; the volume swings were the market and the search, not the scoring.
Reading the numbers in xMap Flows
- Open the Flows page. Each live employee's card shows this week's report card: found, strong fits and an estimate of hours saved, against last week.
- Read the Monday email, which carries a card per employee you own.
- On Results, ask in plain words, for example new results from the cafe scout week by week as a chart. Pin the chart to a dashboard so it is counted again each time you open it.
- Save the filters you check every week (this flow, last 7 days, score above 75) as a saved view, and pin it so its count shows on your Overview.
- If the numbers belong in a spreadsheet or a BI tool, make a live feed of the flow's results for Power BI, Excel or Looker Studio.
Once a week the weekly review asks for one decision per employee: keep it, raise its budget, change it or let it go. Have your four numbers ready when it comes.
Measurement mistakes
- Counting everything found, not what is new. Without Only new in the flow, the same places are found every day and the count means nothing.
- Judging a week in a slow market. If the market opens a dozen of your businesses a year, one empty week is not a failure. Look at four weeks together.
- Raising volume by lowering the bar. Widening the area or dropping the score filter fills the list and the skipped column together.
- No score at all. Without an xMap LLM block scoring each result against your customer, there are no strong fits to count, only a list.
- Changing three things in one week. Change the area, the kinds or the instructions one at a time, or you will not know which change moved the numbers.
The flow behind the numbers
A prospecting employee that measures well usually reads: Schedule every morning, Area of interest, Nearby places for the kinds you sell to, Only new, xMap LLM scoring against a short description of your best customers, Sort and limit to keep the top ten, and an Email. Its daily note in the Team inbox says what it found and the best few, and on Monday it reports the week.
Questions
Can xMap Flows tell me how many leads turned into customers?
No. Results have no statuses or deal values. Track outcomes in your CRM; use xMap Flows to measure what the employee finds and how well it is scored.
What counts as a strong fit?
A result an AI block scored 75 or more out of 100. The report card, the briefing and filters all use that line.
How long should I measure before changing the flow?
Read at least three or four weeks together, especially in markets where few new businesses open each month.
Are hours saved real figures?
No, they are an estimate based on runs that found something, and they are always labelled as an estimate.