Stage 05 · Follow-Up: Persistence Is a Probability, Not a Personality

Ariadne's fifth stage is blunt about its purpose: stop losing the leads you already paid for. This post is about the arithmetic of not giving up too early — and, equally, of not persisting into negative returns.

The geometry of attempts

Model each contact attempt as an independent Bernoulli trial with success probability $p$. The probability of having reached the person after $n$ attempts is

At $p = 0.25$:

Stopping at two attempts reaches fewer than half the people who were reachable. Going from two to six takes you from 43.8% to 82.2% — nearly doubling contact — on leads whose acquisition cost is already sunk.

Where persistence turns negative

Persistence is not monotonically good, and the same body of research that gives us the five-minute number says so. The 2007 lead-response study found that after roughly 20 hours, additional dials become net negative — each further attempt costs more in effort and goodwill than it returns Oldroyd & Elkington 2007. The companion survey-based analysis found each additional unproductive call attempt was associated with about a 5% decrease in qualification and roughly 1.5% in close rate Oldroyd 2007.

Both figures come from vendor-sponsored research, and I would not build a business on their exact magnitudes. The shape, though, matches what any experienced salesperson will tell you: there is an interval where persistence pays and a point past which it damages you.

So the design problem is not "follow up more." It is: how many attempts, over what window, through which channels, before disqualifying — written down once, executed identically every time.

The distribution nobody looks at

Vendor platform data covering over 55 million sales activities on 5.7 million inbound leads at 400+ companies reports that 57.1% of first call attempts occur more than a week after the lead arrives, and that only 0.1% of inbound leads are engaged in under five minutes InsideSales.com 2021.

The reason for that distribution is structural. A first attempt happens when someone remembers. A fourth attempt happens only when a system schedules it. Human memory is an excellent tool for the first touch and a terrible one for the fourth, because by then the person is doing billable work for a customer who already exists.

Writing the cadence down

A follow-up policy is a small artifact, and writing it removes an enormous amount of variance. Mine usually looks like this:

Define the window. Attempts concentrated in the first 24–48 hours, tapering after. This is where the survival curve is steep.

Alternate channels. Attempts through the same channel are more strongly correlated than attempts across channels. Phone, then email, then whichever channel they originally used.

Cap it, and disqualify explicitly. After $n$ attempts the record moves to a named state — not to silence. An un-disqualified lead is a permanent open loop and a permanent source of guilt.

Log every attempt as a record, not a memory. Otherwise the cadence is unfalsifiable: "we followed up" is not a measurable claim, and you can neither audit it nor improve it.

Automate the scheduling, not the conversation. The system should decide when someone is contacted and by whom. What they say remains human. That boundary — machines for timing, humans for judgement — is the subject of two later posts in this series, and it is the line I hold in every build.

References

1. Oldroyd, J. B., & Elkington, D. (2007). The Lead Response Management Study. Presented at MarketingSherpa B2B Demand Generation Summit, 16 October 2007. https://25649.fs1.hubspotusercontent-na2.net/hub/25649/file-13535879-pdf/docs/mit_study.pdfVendor-sponsored, not peer-reviewed; InsideSales.com platform data across six companies, 15,000+ leads and 100,000+ call attempts.
2. Oldroyd, J. B. (2007). Lead Response Management — new research summary (Kellogg Lead Response Management Survey). MarketingSherpa B2B Demand Generation Summit; survey of 495 companies, June–September 2007. https://content.marketingsherpa.com/heap/DG07SFSlides/LeadResponseManagementReport.pdfSurvey instrument, not peer-reviewed.
3. InsideSales.com (2021). Lead Response Study — response time matters. InsideSales.com vendor research. https://www.insidesales.com/response-time-matters/Vendor platform data: 55M+ activities, 5.7M leads, 400+ companies. Full report not independently retrievable.
4. Sabnis, G., Chatterjee, S. C., Grewal, R., & Lilien, G. L. (2013). The sales lead black hole: On sales reps' follow-up of marketing leads. Journal of Marketing, 77(1), 52–67. https://doi.org/10.1509/jm.10.0047