Lead Reactivation: What the Numbers Actually Support

Database reactivation is sold hard right now. The pitch is familiar: you are sitting on a list of old leads and lapsed customers, an AI agent will text all of them, and revenue appears from nothing. The supporting claim is almost always a version of "increasing retention by 5% increases profits by 25% to 95%."

That statistic has a real origin. It does not say what the pitch says it says.

Where the number comes from

The source is Frederick F. Reichheld and W. Earl Sasser Jr., "Zero Defections: Quality Comes to Services," Harvard Business Review, September to October 1990.

The original range was 25% to 85%, not 25% to 95%. The 95% upper bound appears to come from a much later HBR piece by Amy Gallo, "The Value of Keeping the Right Customers," published 29 October 2014, which restated the Bain research with that figure.

So the version most commonly quoted is a 2014 restatement of a 1990 chart, usually cited as though it were a recent finding.

What the 1990 chart actually measured

This is the part that matters. The chart in the original article carries an explicit footnote defining its method:

Calculated by comparing the net present values of the profit streams for the average customer life at current defection rates with the net present values of the profit streams for the average customer life at 5% lower defection rates.

Read that again. It is a net present value model of customer value, comparing two scenarios. It is not a measurement of company profit before and after an intervention. The vertical axis is percent increase in customer value.

The chart reported industry-specific results across nine categories: auto service chain, branch deposits, credit card, credit insurance, insurance brokerage, industrial distribution, industrial laundry, office building management, and software. The values ranged from 25% to 85% depending on the industry.

There is no single number. There is a spread across nine different service industries, produced by a model, in 1990.

The criticisms

The Ehrenberg-Bass Institute for Marketing Science, among others, has argued the loyalty literature of this era demonstrated association and modelled projection rather than causation. Retaining customers correlates with profitability partly because profitable customers are the ones who stay.

Two further objections apply directly to a reactivation campaign.

Not every retained customer is profitable. Indiscriminate retention spending can reduce margin. Some customers cost more to serve than they return, and some lapsed customers lapsed for good reasons.

The payoff depends on your economics, not the industry average. Margin, tenure, discount rate, and the cost of the retention effort all sit inside the calculation. A model calibrated to 1990 branch deposits tells you nothing about your margins.

None of this means reactivation does not work. It means the famous statistic is not evidence that it will work for you.

What is defensible

A dormant list has three genuine properties, independent of any statistic.

It is cheap to contact. There is no acquisition cost, and messaging costs are close to zero.

The contacts already know who you are, which removes the hardest part of cold outreach.

And critically: a reactivation campaign is one of the easiest marketing activities to test properly. You have a finite, static list. You can split it at random. That is a real experiment, available to a business with no analytics team.

That last property is the actual opportunity. Most marketing spend cannot be cleanly attributed. This can.

Run it as an experiment

Hold back a control group. Randomly exclude 10% to 20% of the list from the campaign. Whatever revenue arrives from the control group over the same window is revenue you would have received anyway. Without this, you will count customers who were coming back regardless, which is the single most common way reactivation results get overstated.

Define the window before you start. Decide that you are counting bookings within 30 days, then count bookings within 30 days. Extending the window after seeing results is how a flat campaign becomes a success story.

Segment before sending. Recency, past spend, and reason for lapse are all available in your own records. A customer who left after a complaint needs a different message to one who simply has not booked since last year, and a customer who cost you money should not be contacted at all.

Count net margin, not revenue. Reactivation offers usually carry a discount. A 30% discount to someone who would have returned at full price is a loss dressed as a win, and the control group is what reveals it.

Check your consent position. In Australia, commercial electronic messages require consent, sender identification, and a working unsubscribe under the Spam Act 2003. A list of leads from four years ago is exactly where consent gets shaky. Reactivating a list you have no right to message is an expensive way to find that out.

Where automation and AI fit

Reactivation is a reasonable use case for automation, because the work is repetitive, high volume, and low complexity per contact. Segmenting the list, sending sequenced messages, handling replies, and routing anyone who responds to a human are all appropriate to automate.

What should not be automated is the decision about who to contact and what a returning customer is worth. That is a margin question, and it is answered from your own numbers.

The honest position on reactivation: the mechanism is sound, the cost is low, the industry-standard statistic is a misquoted 1990 model, and the only number that matters is the one your own holdout group produces.

Sources

• Reichheld, F. F., and Sasser, W. E. Jr. "Zero Defections: Quality Comes to Services." Harvard Business Review, September to October 1990, p. 110.
• Gallo, A. "The Value of Keeping the Right Customers." Harvard Business Review, 29 October 2014.
• Ehrenberg-Bass Institute for Marketing Science, published commentary on loyalty myths.
• Spam Act 2003 (Cth).