One of the most common complaints about customer relationship management (CRM) systems is that individual reps don’t use them. This isn’t very surprising to me, as many CRM deployments are designed to give visibility to management or to streamline the ops person’s ability to forecast, rather than to add value to the individual rep. Organizations try to overcome this natural resistance by establishing KPIs to influence behavior. Unfortunately, they don’t always get the behavior they expect.
A client of mine had deployed CRM worldwide, including in its telemarketing centers in India. To its surprise, three months after going live, less than 20% of the number of leads they had projected had been entered into the system. When questioned, the reps explained they found it easier to keep track of leads via spreadsheets.
To change this behavior, management established the KPI ‘% of reps who log into CRM daily’ and tied employee pay to reaching a 100% target. They theorized that collective greed of the group would self-police the outliers into using the system. In fact, usage went to 100% but the lead problem didn’t go away. Clearly they were measuring the wrong thing.
Direct observations inside the call centers showed that many reps would dutifully log into CRM every morning but then continue to use spreadsheets. In response, management paid the CRM vendor to add a feature that automatically logged users out after 15 mins of inactivity and changed the KPI to ‘% of time that an individual rep is logged in’. To their horror, an enterprising call center employee wrote a script which automatically re-logged users in and installed it on every desktop. The new KPI hit 100% but lead management didn’t improve.
Not to be outdone, management changed the KPI to ‘# leads entered in CRM’, figuring this was what they were trying to achieve. After some tweaking to make sure reps weren’t entering fake leads to pump up their volume, nearly all leads used for demand generation were entered in the system. Unfortunately this uncovered another design flaw. While the leads existed in the CRM system, they weren’t managed by the system; that is, they didn’t follow a proscribed multi-step flow from identification (stage 1) through interest generation to qualification and submittal to sales (stage 5). Leads were stuck in stage 1.
As a result, management changed the KPI for a fourth time to ‘sales accepted leads’. Essentially, this tracked the number of leads sales (a separate group) deemed worthy of follow up by moving them from stage 5 to stage 6. As you might have guessed, leads did move but they jumped from stage 1 to stage 5 with no stops in between. Sales accepted leads gave visibility into campaign effectiveness but management still had no visibility into the demand generation pipeline – which was the reason they bought the CRM system in the first place.
Exasperated, the organization decided to ask for external advice and brought me in due to my previous work with call centers. On my first visit, I asked a series of simple questions designed to better understand what they were really trying to accomplish. As it turned out, they had three key questions in mind:
- How much does it cost us to generate a lead?
- How good are our telemarketing reps in converting a lead to a qualified opportunity?
- Where (and why) do leads get stuck during the qualification process?
The first two questions explained their gyrations around KPI setting before I showed up. Unless every lead was tracked in the system, they wouldn’t be able to calculate a true cost per lead. Separately, they reasoned the percentage of leads from stage 1 that made it to stage 5 would be a good approximation of rep effectiveness.
For the third question, we decided to measure the fraction of leads that exited each stage of the funnel. For example, movement in the early stages of the pipeline could be determined by ‘% of stage 1 leads that moved to stage 2’ and ‘% of stage 2 leads that moved to stage 3’. After some experimentation, we devised different target values for each stage – it was easier to go from identification to interest generation (1 to 2) than from detailed qualification to submittal to sales (4 to 5). As time went on, they were able to vary the targets by product and even by individual. Experienced reps got stuck in different places than newer ones.
When I returned about a year after the initial engagement, the company had experienced incredible results. Cost per lead had reduced by almost 30% and agent effectiveness had almost doubled. Even more interesting, they used rich language to describe the demand process, talking about the shape of the funnel and the viscosity of the lead flow. I was very impressed.
Over celebratory drinks that evening, the VP of Marketing Operations made a pithy toast:
We may never get the shape of our demand gen funnel to change from a martini glass to a champagne flute, but at least we finally know what’s in our glass.