Cost Estimator

Before investing in promotions, marketplaces ask how much will it cost to add and support native promotions?


For advanced estimates, please email info@promoted.ai.

Background

Building and maintaining a promotion system is expensive. Google, Facebook and Amazon spend ~$1B/year on employees to maintain and grow their promotion systems.

Why so expensive?

  • Incremental costs grow. Each incremental change to business logic and features/surfaces adds future requirements. If a feature costs $10k in version 1, the total long-term cost can easily be >$100k. Most implementations have debt that increases the incremental costs.

  • Hard to understand. Most promotion systems are poorly designed and difficult to understand. Promoters cannot figure out how achieve their goals. Instead of fixing the auction, marketplaces usually try to patch fix the problem using support teams and internal tools.

  • Gameable. Most marketplace implementations are easily gameable. Some users will keep gaming the system. Marketplaces have to continuously invest in patching their gameable systems. Not addressing the issues will hurt users.

  • Team-size. These problems lead to increased team sizes. As teams grow, so does overhead (communication, knowledge ramp up, employee churn, hiring delays). The work becomes less fun. Team effectiveness drops. Pay increases to retain employees.

Story

Here is a common story for successful promotion systems.

  • Year 1. A couple engineers crank out a quick solution.

  • Year 2. The scale and debt are larger than expected. Quality improvements are implemented. Teams fix compliance and privacy support. Promoters are angry at lack of tools. Engineering team grows to >10 to keep up.

  • Year 3. Revenue growth slows. Team realizes their auction has issues. Promoters do not understand how to achieve their goals. Promoters start getting lower return on investment. Marketplaces add incremental features to grow revenue. They hire a larger sales team to convince more promoters onto the platform. The team grows to >50 people. More expensive and experienced employees are hired to lead.

Cutting Costs

Teams try many ways to reduce costs. Most fail.

  • Increasing tech debt. Tech debt in the wrong spots can be devastating. Trying to change the fundamentals of an auction requires redoing most of the system. Example: if you change what a bid means, how do you update promoter bids? Previous data cannot be relied upon.

  • Cutting sales and support. This rarely works out well. Sales and support bridge the gap between the ideal promotion system and the current one. If sales and support cannot scale, the marketplace risks ruining relationships with big promoters.

  • Cutting corners. Poorly implemented improvements will annoy promoters and risk lower lifetime value. It is better to keep a smaller set of higher quality functionality.

  • Reducing risk / Making conclusions too early. Teams want to validate the business model with very few resources. They will implement a v1 system, let it grow and make a conclusion about the future of the promotion system. You might need many versions of a promotion system to make conclusions about it.

  • Cutting sales and support. This rarely works out well. Sales and support bridge the gap between the ideal product and the current one.

  • Keeping low value features. As systems grow, low value features should be pruned to reduce future costs. Systems often keep features around too long and pay for the extra debt.

  • Cutting low value features. New promotion systems often cut useful features that do not directly drive revenue but are part of critical user journeys. They do not realize how important the features are until they have consumed many years of time. E.g. timezone-support for metrics. Supporting change history.

  • Avoiding self-cannibalization. Teams will say we already have a solution for X. They will reduce costs related to new business lines that might cannibalize new business lines.