AI and growth: stop automating only to cut costs

Ask ten executives, "What is AI for in your company?" Nine will answer with some variation of efficiency — save time, cut costs, lighten the team, process faster. It is useful, measurable, reassuring. It also misses half the value.

In an article published in June 2026, three Harvard Business Review researchers make a simple case: companies concentrate their AI investment on cost reduction, when the greater potential lies in growth — selling more, creating new offers, serving better. This article explains why that shift matters, and how to start making it concretely.

Why efficiency alone eventually plateaus

Automating a repetitive task saves time. That is real, and an excellent starting point. But efficiency follows the law of diminishing returns: once your heaviest processes are automated, each new optimisation yields a little less than the last. You cannot cut costs indefinitely.

Growth works the other way. Better sales targeting, a faster customer response, a new offer made possible by AI — each opens a space that did not exist before. It is the difference between squeezing a lemon and planting a lemon tree.

Three concrete ways to use AI for growth

1. Personalise at scale

Advanced personalisation used to be reserved for large companies with data teams. Generative AI puts it within reach of a very small business: tailored recommendations, emails calibrated to real customer behaviour, content adapted per segment. The same effort produces an offer perceived as bespoke.

2. Accelerate and increase sales

Qualifying inbound leads, drafting a first version of a proposal, following up at the right moment: all stages where AI shortens the sales cycle. Recruitment giant Indeed reports, in a May 2026 Forbes article, a 30% improvement in its sales functions and customer experience after deploying AI — a clear signal that the "revenue" lever is as tangible as the "cost" lever.

3. Create new offers

AI does more than improve the existing: it lets you launch services that were previously too costly to produce. An automated audit, a personalised report, round-the-clock assistance — additional offers you can charge for, and that set your business apart.

What this changes concretely for your business

The shift is not technical, it is strategic. Before deploying a tool, ask a different question: "Does this save me money, or does it win me a customer?" Both are legitimate, but if 100% of your AI projects target savings, you leave half the value on the table. A good balance is to fund your first growth use cases with the time freed up by your efficiency automations.

Use cases by company size

• Solo: use AI to produce personalised proposals in minutes rather than hours, and respond faster than competitors.
• Micro-business: set up automatic qualification of inbound leads so no quote is ever left without follow-up.
• SME: deploy per-segment personalisation on emails and the website, then measure the effect on conversion.
• Larger company: industrialise a new AI-assisted service offer, with human supervision and quality indicators.

A 5-step action plan

• Step 1 — List your revenue streams and spot where a delay or a lack of personalisation loses you sales.
• Step 2 — Pick a single high-impact, low-complexity "growth" use case.
• Step 3 — Deploy it small, with a measurable goal (e.g. cut sales response time by 24h).
• Step 4 — Measure the real effect on revenue at 30, 90 and 180 days, not just time saved.
• Step 5 — Reinvest the gains into the next use case.

Limits and conditions for success

This optimistic case calls for an honest caveat. MIT researchers estimated in 2025 that the vast majority of enterprise generative-AI pilots did not lead to usable results — a figure relayed by Forbes. Growth through AI is not automatic: it requires clean data, human supervision, rigorous measurement and upskilled teams. Without those conditions, you stack tools with no effect on the bottom line.

In other words: the potential is real, but it must be earned. One well-executed use case beats ten abandoned experiments.

The question, then, is not "how can AI save me money?" but "how can AI help me create more value for my customers?" The first lowers a cost line. The second builds your future.