Retention Management

When software was sold once and installed, there wasn't much left to do after the contract closed. Signature, installation, done. Whether the company used the product daily or forgot it existed, the money had already changed hands. The next visit came when a new version was ready to sell, usually two or three years later.

Annual subscriptions that get re-decided every year turned that around. The contract is signed once, but it expires twelve months later, and at that point the customer starts the decision from scratch. Continuing the contract is called renewal; walking away instead is churn.

The catch is that the decision isn't made on the renewal date. A sales rep calling a month before expiry would hear something like: "We consolidated onto another tool last quarter. Only a handful of people were still using yours." The contract was technically alive while the customer had made up their mind six months earlier. Pulling out a discount with four weeks left had nothing to work with.

Looking back, the signs were there before the ink dried. Fifty accounts were created in the first month and twelve people actually used them, those twelve never grew over the next half year, and when the internal champion left the company the remaining users lost their reason to log in. All of it was in the logs. Nobody was reading the logs. The renewal call is the last thirty minutes of a twelve-month story, and companies were preparing only for the thirty minutes.

Retention management is what you get when you stretch those thirty minutes across the whole year. Instead of starting two or three months before expiry, you begin collecting what this customer is getting out of the product from the day the contract starts, and when the numbers start shrinking you go looking for the reason then rather than later.

The scope isn't limited to defending what exists. Selling more seats to a customer who is using the product well, or moving them to a higher plan, falls to the same person. The two jobs sit together because they read the same evidence. Login counts and feature usage show whether an account is contracting or expanding, and the same record picks the moment to raise the subject of more seats. Holding a contract flat is a decent outcome; the better one is the same customer paying more.

The timing shift is the whole idea. A renewal conversation that starts in month eleven is a negotiation; the same conversation supported by eleven months of evidence is a summary. When Pagedeck's account managers open a renewal deck, most of it was written months earlier — which teams adopted the product, which workflows moved onto it, how many hours of manual work disappeared. None of that can be assembled in the last three weeks, because the events it describes either happened during the year or didn't.

A comparison people in the industry reach for is that landing a new customer costs some multiple of what it costs to keep an existing one. The exact multiple varies by company, and it matters less than the fact that the comparison is made at all: adding a head on the retention side goes on the same scale as adding one to sales.

Five terms sit under retention management, and they're easiest to follow in time order. What might happen next (management), the shrinking that shows up before anyone leaves (symptoms), the reason recorded after someone has left (analysis), and — separate from all three — the work of thickening the relationship itself (relationship).

Management has one term, and it deals with what hasn't happened yet.

1. Renewal forecast — taking the accounts that expire in the next quarter and judging in advance which ones will continue and which will leave. It isn't left to the account manager's instinct; it carries evidence. Logins over the last three months, whether the core features are in use, how long since anyone spoke to the decision maker, the count of unresolved tickets — each account is sorted into likely, uncertain, or at risk. The reason this exists is simple arithmetic on attention. One manager covering 60 accounts with 20 of them expiring this quarter cannot give all 20 the same care. The forecast decides where the hours go. It does a second job too: it feeds the company's revenue estimate for next quarter with a number that has reasoning behind it.

Symptoms has two terms, both describing the shrinking that happens before a full departure. They differ in what shrinks.

1. Downgrade — the same customer moving to a cheaper plan. A team on the enterprise tier dropping to the standard tier loses a few features but keeps the contract. It looks better than an outright loss, and usually it is, but it normally follows a drop in actual usage, which makes departure the following year more likely than not.
2. Contraction — the plan stays the same while the number of people or the scope of use goes down. A company that contracted for 80 seats renews for 50. If a downgrade is a drop in grade, contraction is a drop in volume, and both land in the same place: less money per month. They're counted separately to catch the state where the cancellation count says nothing happened while revenue falls. A hundred customers still being a hundred customers while total seats drop from 4,000 to 3,200 means a churn rate of zero percent and revenue down by a fifth.

Analysis has one term, and it deals with what already happened.

1. Churn reason — the recorded cause of a departure, sorted into categories decided in advance. Price, missing features, a change of contact, budget cuts, moved to a competitor: a handful of boxes, and the account manager picks one. Free text isn't used because the point is to count. Twenty prose write-ups are hard to draw a conclusion from, but twelve of twenty landing in "missing features" settles what gets built next quarter. One caution: the stated reason and the real reason often differ. A large share of "too expensive" means not that the price was high but that the customer never used it enough to justify any price.

Relationship has one term, and it behaves differently from the other four.

1. Customer advisory board — a group of ten or so major customers who meet quarterly to give input on where the product is going. It isn't a device for this quarter's renewal; it builds a relationship measured in years. A product someone helped shape is harder to walk away from, and the people who attend usually become the product's internal advocates at their own company. It also keeps the relationship intact when the day-to-day contact changes jobs. The cost is real, though. Gathering ten companies means aligning calendars, preparing material, and reporting back at the next meeting on which suggestions were acted on — and if nothing was, attendance collapses by the second session. It suits products with few customers and a high value per account.

The five form a loop. Churn reasons tell you which signals are worth watching, those signals become the raw material for the renewal forecast, downgrades and contraction appear first among the accounts the forecast flagged, and the outcome accumulates as new reasons. The advisory board sits outside the loop, holding the largest accounts far enough away that they rarely enter it. The growth side calls the same work churn prevention; the dividing line is how much one account pays. Thousands of small-plan teams get automated messages, while customers on annual contracts get a person on a call.

"Five of the twenty-two accounts expiring next quarter are at risk, and by revenue those five are 60 percent of the total."

Rank by money rather than by count to decide where to start.

"They renewed, but seats went from 80 to 50. Renewal rate says 100 percent and revenue says otherwise."

No cancellation is not by itself a reason to relax.

"Their contact changed last month. Before we bring up renewal we need to walk the new person through the product from the beginning."

The person making the decision is different now, so the relationship starts over.

At Pagedeck this comes up in the renewal review at the start of every month. Sofia puts up the accounts expiring within 90 days, ordered by value, with a rating of likely, uncertain, or at risk next to each one and a line of reasoning for the rating. The reasoning line became mandatory after three accounts rated likely left anyway last year. All three had a warm relationship with their account manager and a login count that had been falling for six months. A good impression of a person and evidence of a product being used were being treated as the same thing.

Mia reads the list from a different angle. She groups the at-risk accounts by which feature they stopped using. If they cluster on one feature, it isn't a set of individual situations but something the product has to fix, and the answer is a roadmap change rather than a round of phone calls. Grace asks the same question every month: are the accounts reported as saved last quarter still paying the same amount this quarter? Buying a year with a discount and getting a team to actually use the product look identical in the month of the renewal and separate twelve months later.

The most common mistake is treating retention management as something that happens in the weeks before the renewal date. The date is on the calendar and that's when the account manager starts moving, so the impression is understandable. But all those weeks can do is confirm a decision that was already made. Whether a customer continues depends on what they did with the product over the previous twelve months, and the first three months carry unusual weight, since that's when the number of real users either grows or doesn't. Push only at the end and the same year repeats: spot the risk three weeks out, buy another year with a discount, meet the same situation again next time.

The second is judging the whole effort by renewal rate alone. It counts departed customers, which makes it easy to calculate and easy to read out in a meeting, so teams lean on it. What it cannot see is money. A hundred customers all renewing while thirty of them cut seats gives a renewal rate of 100 percent and falling revenue. The reverse also happens: five customers leave, the remaining 95 add seats, and the account count shrinks while revenue grows. So retention gets read on two lines, count and value. On one line, contraction runs for a full quarter before anyone notices it.