From Four Disconnected Systems to One Governed Revenue Record

Company profile

Background

Halyard had grown by adding tools as needs arose. The CRM was configured by the first sales hire three years earlier and extended by whoever needed a field. Marketing automation, support ticketing and renewal tracking each lived in a separate system, with no integration between them. The product platform, where customers created accounts and invited their own users, held the cleanest records the company owned, and none of it reached the CRM.

Leadership planned to raise a larger round within twelve months. Investors would ask for a retention curve, pipeline coverage and a cohort view of expansion. Producing any one of them took most of a working week.

The challenge

Every Monday the forecast was assembled by hand from four exports. The numbers rarely agreed on the first pass, and the weekly pipeline review began by deciding which number to use.

An audit of the CRM, marketing platform, integrations and data found three root causes. Sales and marketing each defined a qualified lead differently, so the monthly handoff report never matched. Stage definitions existed and were not enforced, and lead fields were completed four different ways. No rule assigned ownership of a lead after first touch, which accounted for the double-owned accounts. A customer with three subsidiaries appeared as seven company records, each with its own owner.

The audit also identified an asset the company was not using. The product database held current organization and user records tied to real contracts. Matched to the CRM, it could resolve most of the duplicate-company problem and show for the first time which accounts were using the product.

Approach

1. Agree the definitions before configuring the system. The leadership team asked for the CRM rebuild to begin immediately. The recommendation was to sequence agreement first. A rebuilt CRM enforces definitions, and Halyard did not yet have shared ones. Configured first, the system would encode one function’s view and the others would contest it for a quarter. Sales and marketing met in a series of working sessions and settled two questions: what counts as a qualified lead, and what happens in the first hour after it is created. The first session reached a deadlock over how much product usage indicated buying intent. It was resolved by defining two states, a marketing-owned lead that becomes sales-owned when it meets a documented set of signals, with an acceptance step and a service level on the first response. The agreement was written on a single page in plain language before any field was changed. The visible result was a five-week period with no system change. The benefit was that the later configuration met no resistance, because the rules were the teams’ own.

2. Rebuild the CRM around the agreement. The rebuild took about four months, in layers, so the sales team could continue to work in the system. Pipelines were restructured with entry and exit criteria that the system enforced. A deal could not advance past the third stage without a named buyer, a stated project and a next step. Lead scoring was built from the agreed signals, and automatic routing replaced the shared inbox. Double-owned accounts were resolved by territory logic drafted from the account data and reviewed account by account with the sales managers, who reassigned about sixty accounts by hand before approving it. Duplicate and orphaned records were cleaned under written ownership rules, published inside the CRM.

3. Connect the product to the CRM in phases. This was the largest piece of work and ran over many months, because the data was live. Customer organizations were matched to company records first. About two-thirds matched on domain and name, and the remainder were reviewed individually. Product users were synchronized to contacts next. In this phase an early merge pass combined two legitimate subsidiaries of one customer, because the match rule treated a shared domain as proof of identity. The error was identified within a day when an expansion opportunity disappeared from a rep’s pipeline, and it was reversed from backup the same afternoon. The rule was revised to treat parent and subsidiary as a hierarchy, and every later merge batch went through sample review before running at scale. Enrichment was brought under governance last, one segment at a time, so that no field had two sources.

4. One reporting layer. Pipeline by stage and segment, marketing attribution from first touch to closed deal, and renewal dates with product health signals were placed in one view. Each rep received a view of their own territory.

5. Transfer ownership. Governance rules and field standards were published inside the tools the team already used. Sales operations, marketing operations and customer success each named an owner for their part of the stack, with an escalation path. Training was organized by workflow, such as what a rep does when a lead arrives and what a customer success manager does ninety days before a renewal. The handoff from sales to customer success was designed with the people who would run it and tested on twelve live accounts, which exposed two gaps in the handoff form before it was embedded.

Results

The weekly forecast ceased to be assembled. Leadership and reps worked from the same pipeline, the same lead definition and the same renewal calendar, and the forecast review moved from reconciling numbers to deciding on them. Marketing could show pipeline by program for the first time, leading to budget decisions based on contribution and not on activity. The unified record gave sales and marketing visibility into which customers were using the product, leading to a go-to-market engine the team operated on its own with no outside support. The retention curve, pipeline coverage ratio and expansion cohort requested for the next round were each produced in an afternoon.

About this study

Illustrative: modeled scenario, not a client engagement. Halyard Software is invented. The starting conditions are modeled on patterns common to Series B software companies. The scale of the final record, more than 10,000 contacts and 8,000 companies tied to over 5,700 customer organizations, is the scale at which this work has been done.