Evaluating a Personalization Engine: Build vs. Buy in Retail eCommerce

In the fast-paced world of eCommerce, delivering a personalized shopping experience can be the difference between conversion and cart abandonment. A vendor pitches a plug-and-play AI-driven personalization engine that promises to improve product recommendations, boost conversion rates, and increase revenue per user.

The pitch is promising—but is it better to buy, or should the company build its own personalization engine in-house?

This is where Enterprise Architecture (EA) steps in to bring financial and architectural rigor to the decision-making process—by modeling each option using Net Present Value (NPV).


🛍️ Business Context

The eCommerce company currently uses basic product recommendation rules. They are exploring a more advanced AI-based personalization solution that can:

  • Use behavioral and transaction data
  • Dynamically update product rankings
  • Personalize search and landing experiences

The estimated revenue uplift from improved personalization is $400,000/month, once fully implemented.


📊 Two Options on the Table

Option 1: Buy Off-the-Shelf Solution

  • One-time integration cost: $1.2M
  • Annual subscription/license cost: $1.8M ($150K/month)
  • Time to deploy: 3 months
  • Revenue uplift starts in Month 4

Option 2: Build In-House

  • Development cost: $4M
  • Annual maintenance & infrastructure cost: $600K ($50K/month)
  • Time to build: 9 months
  • Revenue uplift starts in Month 10
  • Opportunity cost (delayed benefit): $400K/month during those 9 months

🧮 NPV Assumptions

  • Revenue benefit: $400K/month
  • Discount rate: 8% annually (~0.64% monthly)
  • Planning horizon: 5 years (60 months)
  • NPV calculated monthly

📦 Option 1: Buy Off-the-Shelf

Cash Flows:

  • Months 1–3: -$150K/month (no revenue yet)
  • Months 4–60: +$250K/month = ($400K revenue – $150K subscription)

NPV Summary:

  • Month 0: -$1.2M
  • Discounted Months 1–3: -$437,224
  • Discounted Months 4–60: +$8,555,842

➡️ Total NPV = $6,918,618


🧱 Option 2: Build In-House

Cash Flows:

  • Months 1–9: -$450K/month = ($50K infra + $400K opportunity cost)
  • Months 10–60: +$350K/month = ($400K revenue – $50K run)

NPV Summary:

  • Month 0: -$4M (build cost)
  • Discounted Months 1–9: -$3.56M
  • Discounted Months 10–60: +$8.82M

➡️ Total NPV = $1.26M


🧾 Final NPV Comparison

Option5-Year NPV
Option 1 – Buy$6.92M
Option 2 – Build$1.26M

🚀 Conclusion: Speed Wins in Retail

The off-the-shelf personalization engine delivers nearly 5.7x more NPV than the in-house build, thanks to:

  • Rapid deployment (3 months vs. 9 months)
  • Earlier revenue realization
  • Avoidance of large upfront opportunity cost

Even though building in-house offers lower run cost long-term, the delay in benefit realization and heavy development investment result in significantly lower NPV.


💡 Key Takeaways for eCommerce Leaders:

  • Personalization is a revenue driver—delay has a real cost.
  • Use NPV to compare build-vs-buy beyond just IT costs.
  • Enterprise Architecture can quantify the tradeoffs between speed, spend, and strategic alignment.
Posted in ,