Abstract frameworks are useful up to a point. What's often more useful is a concrete story — what does the journey from reactive to predictive actually look like for a real PI firm? What changed, when, and what surprised them along the way?
What follows is a composite narrative based on patterns common across PI firms making this transition. The firm is fictional, but the challenges, decisions, and outcomes reflect what this progression typically looks like in practice.
The Starting Point: Level 1 — A Good Firm with a Data Problem
The firm is a plaintiff-side personal injury practice with 18 attorneys, running $220,000 per month in marketing across seven vendors. They've been in business for 12 years and are well-regarded in their market. Their marketing director, who joined two years ago, is competent and analytical. She wants better data. She just doesn't have it yet.
The reporting process when she arrived: each vendor sends a monthly report in their own format. She pulls the numbers into a master spreadsheet. She adds the signed case counts from the case management system. She calculates cost per lead and, when she has time, cost per case at the portfolio level. The whole process takes her about 12 hours per month.
What she doesn't have: vendor-level cost per case. Conversion rate trends. Any way to tell which vendors are improving or declining. Settlement attribution. And no early-warning capability — problems surface when the month closes, not when they begin.
This is Level 1. Not bad management — absent infrastructure.
Month 1-2: Building Level 2 — Getting the Data in One Place
The first step was not technology. It was process. She built a more structured spreadsheet — one with consistent vendor metrics, a rolling three-month view, and a standardized monthly data entry process. She also started pulling intake conversion data: for each vendor, how many leads became signed cases each month?
This took about three months to get right. The intake team tracked case sources inconsistently, so there was cleanup work before the data was reliable. The managing partner was skeptical of the time investment. But by month three, she had something she didn't have before: cost per signed case by vendor, for three months running.
The first result from this data was immediately useful and immediately uncomfortable. Vendor F — a vendor the firm had worked with for four years and considered reliable — had the highest cost per case of any vendor in the portfolio. Not by a little. By 40%. The cost per lead was competitive, but the conversion rate was poor enough that the effective cost per case was $3,800 vs. a portfolio average of $2,700.
Nobody had known this before because nobody had been tracking cost per case by vendor. The conversation with Vendor F was difficult. But the data made it possible.
Vendor F Cost Per Case
$3,800
Highest in portfolio
Portfolio Average
$2,700
Across all vendors
Conversion Rate Range
7%–22%
Wide variation by vendor
What surprised her at Level 2:How much the conversion rate varied by vendor. She had assumed vendors sending similar quality leads would convert similarly. They didn't. The range across her vendors was 7% to 22% — and the vendors with the highest cost per lead were not always the worst performers on cost per case.
Month 4-8: Approaching Level 3 — Connecting the Systems
The spreadsheet process was better than what she had before, but it was still eating 10 to 12 hours per month of data assembly. More importantly, it was always backward-looking. She could see last month. She couldn't see this week.
The transition to Level 3 required connecting three systems that had never been connected: the ad platforms and vendor portals (marketing spend data), the intake CRM (lead and conversion data), and the case management system (signed case data). The firm used LeadDocket for intake, which had a native integration that simplified the data connection significantly.
Getting to connected data took about three months — one month of integration work and two months of data validation and process adjustment. The intake team needed to consistently tag lead sources in a way that matched the marketing data. Small inconsistencies in how sources were labeled created gaps. Fixing those was necessary before the connected data was reliable.
By month seven, they had a working dashboard showing real-time pacing, vendor-level cost per case, and conversion trends. Monthly reporting time dropped from 12 hours to about 90 minutes — and most of that 90 minutes was reviewing data, not assembling it.
Level 1–2
- 12 hours/month on data assembly
- Backward-looking monthly reports
- Problems found at month-end
- Manual vendor portal data pulls
Level 3
- 90 minutes/month reviewing data
- Real-time pacing dashboards
- Alerts fire within days of changes
- Automated data integration
What surprised her at Level 3, first month:An alert fired in the second week of month eight. Vendor B's conversion rate had dropped from 14% to 8% over three weeks. She wouldn't have seen this for another two and a half weeks in the old process. She contacted Vendor B, who confirmed they had changed their lead qualification criteria — without notifying the firm. The targeting was corrected within a week. Under the old system, she would have paid for a full month of underperforming leads before knowing anything was wrong.
That single alert paid for the infrastructure investment. The marketing director would say later that it also changed how she thought about the role — from reporting what happened to managing what was happening.
Month 9-18: Operationalizing Level 3 — Building the Rhythm
Having connected data is not the same as operating from connected data. The months after reaching Level 3 were about building the rhythm — the daily checks, the weekly vendor reviews, the monthly portfolio analysis — so that the data was consistently informing decisions rather than being available but ignored.
A few things changed during this period:
The monthly partner meeting changed format. Instead of the marketing director presenting a summary of what happened last month, the meeting became a 30-minute discussion of what was changing and what decisions needed to be made. The managing partner could see the vendor rankings before the meeting. The conversation was about interpretation and action, not information transfer.
The intake team became part of the marketing conversation.The intake manager had visibility into which sources were converting well and which were producing high rejection rates. For the first time, intake could tell marketing that leads from a specific vendor were failing at intake for a specific reason — not a vague quality complaint but a data-backed observation. Marketing could bring that back to the vendor with specifics.
Vendor allocation became dynamic. In the Level 1 and Level 2 periods, vendor allocations were adjusted quarterly at best — and usually only when something went obviously wrong. At Level 3, with real-time pacing and monthly vendor grading, the marketing director was making small allocation adjustments monthly based on performance trends. The portfolio was being continuously optimized rather than periodically reviewed.
What surprised her during this period: The compound effect. Small monthly optimizations — 10% shift from an underperforming vendor to a better-performing one, a budget increase to the vendor with the improving trend — accumulated meaningfully over 12 months. By month 18, the blended cost per case across the portfolio was 22% lower than it had been at month six. Not because of any single dramatic change, but because monthly optimization compounds.
Month 19-24: The Beginnings of Level 4 — When History Starts Talking
Level 4 arrives slowly, not as a sudden capability jump. It emerges when the historical dataset is rich enough to support pattern recognition.
At month 19, the marketing director noticed that Vendor A's inquiry-to-contact rate — the rate at which leads from Vendor A were successfully contacted by intake — had been declining for six weeks. The conversion rate hadn't moved yet. But historically, when that leading indicator dropped in this pattern, conversion followed within four to six weeks.
She flagged the issue with Vendor A before the conversion rate moved. They identified a lead quality shift in one geographic sub-market. The issue was addressed. The conversion rate never declined. Under the previous operating model, she would have seen the conversion rate drop and investigated retroactively.
This is the defining characteristic of Level 4: the ability to act on signals before they become problems. It requires historical data — you can only recognize the early pattern if you've seen it before — but once you have that history, the decision quality improves substantially.
The other Level 4 capability that emerged during this period was settlement attribution. Cases signed in the first year of connected data were beginning to settle. For the first time, the firm could connect a settled case to its original marketing source — and calculate the actual revenue generated per dollar spent on each vendor over a full cycle.
The results were not what anyone expected. The vendor rankings by cost per case shifted materially when settlement value was factored in. One vendor that looked expensive on cost per case consistently produced higher settlement values — likely because of case severity mix. Another vendor that looked efficient on cost per case was producing cases that settled at the low end of the range. Settlement-adjusted ROI looked different from cost-per-case alone.
What surprised her at Level 4: How much the settlement data changed the vendor conversation. Every previous evaluation had been based on cost and conversion. Adding settlement value by source introduced a dimension nobody had previously been able to measure. The portfolio allocation that looked optimal on cost per case was not optimal on settlement-adjusted ROI.
Where They Are Now: What the Journey Cost and What It Returned
Twenty-four months in, the firm has materially different marketing operations than it did at the start.
The data assembly that used to take 12 hours per month now takes 90 minutes. That's roughly 125 hours per year returned to the marketing director for actual analysis and decision-making.
The blended cost per case across the portfolio is 22% lower than it was at month six of Level 3. On a $220,000 monthly budget producing approximately 55 cases per month, a 22% improvement in cost per case means the firm is getting roughly 12 additional cases per month for the same spend — or the same case volume for $48,000 less per month.
The managing partner now asks fewer questions about marketing ROI — not because he's less engaged, but because the data is visible to him in the same system and the monthly review covers it routinely.
The journey was not linear or smooth. There were months of data cleanup, integration challenges, and process adjustments. The intake team needed training on consistent source tagging. There were moments when the data produced unexpected results that required investigation before the team trusted the numbers.
None of that was unusual. It is the honest reality of building connected data infrastructure in a firm that was not designed around it. The question is not whether there will be friction — there will be. The question is whether the outcome justifies the effort. For this firm, the answer was clearly yes — and the most meaningful impact was not the cost savings but the confidence. Decisions that used to feel like educated guesses now feel like grounded choices.
Related guide: See our complete guide to revenue intelligence for PI firms — the four layers, the maturity model, and what RI replaces in your current stack.