The numbers that should trigger an action, decided in advance

5 days ago
3 min read

Updated: 3 days ago

Introduction

Most funnel reviews end in discussion rather than decision. The numbers are presented, opinions are offered about what they mean, and the meeting closes with an agreement to keep an eye on things. Nothing was decided because nothing had been decided in advance about what would count as a problem.

A threshold fixes this. It is a sentence agreed before the data exists, saying what value of what number, for how long, causes what to happen. With thresholds in place, most reviews take four minutes and end with either nothing or a clear action.

The thresholds do not have to be sophisticated. They have to exist and they have to be written down.

1. The numbers that should trigger an action need a written threshold

The mechanism.

A value, a duration and a response, agreed beforehand. Without the duration, one bad month triggers a reorganisation. Without the response, the threshold changes nothing. All three parts fit in a single line.

2. Enquiry volume is the first threshold

The demand signal.

Two consecutive months below a stated floor means look at the top of the funnel. That is a clear enough instruction to act on. Set the floor from your own history, not from ambition. Use the worst normal month of the last two years.

3. Response time is the second

The process signal.

If the median reply time crosses a limit for two weeks running, something has broken in how work is handled. This is usually a capacity problem. It responds quickly once noticed. Check who is holding the enquiries before changing anything else.

4. Conversion rate needs a wide band

The rate signal.

Given how much it moves at small volumes, only a substantial and sustained shift should trigger anything. Quarterly, against the same quarter last year. Anything tighter produces false alarms. Five points is a reasonable band for most small businesses.

5. Quotes issued without wins

The qualification signal.

A rising quote count with a flat win count means effort is going into the wrong enquiries. This one is frequently visible before conversion rate moves. It deserves its own threshold. Ratio the two and watch the ratio rather than either count.

6. Cost per enquiry has a ceiling

The spending signal.

A figure above which a channel gets paused rather than discussed. Decide it when you are calm, not when the invoice arrives. It should be derived from revenue per enquiry. A channel costing more than it returns needs no further debate.

7. The dormant customer count

The retention signal.

When the number of customers past their repurchase interval exceeds a stated level, that becomes the month's priority. This is a trigger almost nobody sets. It is also one of the cheapest to act on. The response is usually a short list and an afternoon of calls.

8. Say what the response is

The other half of every threshold.

Investigate, pause, escalate, do nothing. A threshold without a named response is a discussion prompt. Name the person as well as the action.

9. Review the thresholds annually

The maintenance.

A floor set three years ago may now be below where a bad month lands. Adjust them deliberately once a year. Silent drift in thresholds is as damaging as drift in definitions.

Be careful about setting too many. Four or five thresholds get honoured; twelve get ignored within two quarters, and the whole approach is then abandoned along with them.

Conclusion

Write down, before the data arrives, what would make you act.

Give each threshold a value, a duration and a named response, set an enquiry-volume floor from your own history, put a limit on median response time, keep a wide band on conversion rate and review it quarterly against the same quarter last year, watch quotes rising without wins, agree a ceiling on cost per enquiry while you are calm, add a trigger on dormant customers, and revisit the whole set once a year.