Most research agencies have made a quiet decision about AI. Some have gone all in. Some have stayed away. And a lot of them are using it without telling clients — then billing at the old rate.
I'm not comfortable with that. So we changed how we price.
What We're Doing
Starting now, every Harvest Insights proposal includes two prices:
- A fully human cost — the way we've delivered research for 13 years
- The cost with AI woven into specific parts of the workflow
Clients choose. Fully human, fully AI-assisted, or any mix in between. The difference is spelled out line by line in every proposal.
Why This Had to Happen
AI genuinely saves time on certain parts of the research process. Coding, analysis, transcription, reporting — there are real efficiency gains in the right places. Those savings average around 18–19% depending on methodology. A 15-minute quantitative survey runs nearly 20% less with AI assistance. Six 90-minute focus groups come in around 18% less.
The question every agency has to answer is: where do those savings go?
We're not hiding that AI is part of how we work. And we're not pocketing the difference. Clients deserve to know — and to benefit.
That's the whole point.
This Isn't a Discount for Worse Work
The research question, the design, the analysis, the interpretation — that's all still us. AI handles the parts it's actually good at. A faster transcription is not a worse transcription.
What doesn't change: the strategic thinking, the senior oversight, the quality of what lands in front of a client.
What We Built to Make This Responsible
Anyone can slap "AI-assisted" on a proposal. We spent months building the governance behind ours.
Only vetted tools. We formally evaluate every AI tool we use for security, privacy, hallucination risk, and research suitability. This isn't "we use ChatGPT." It's a documented approval process.
Your data stays clean. PII never goes into an AI system. We anonymize before any AI touches data, and we have a written policy on it.
Humans check everything. Every AI output has a documented human review checkpoint before it goes into a deliverable. No exceptions, no deadline pressure excuses.
Participants are informed. We built ICC/ESOMAR-aligned consent language so participants know when AI is involved in analyzing their responses.
We can prove it. Staff training records, incident response plans, compliance statements — the documentation exists, and clients can request it.
What to Do With This Information
If you're a research buyer, ask your current agency what they're doing with AI on your projects. If they can't give you a straight answer, that's your answer.
If you want to see this in action, reach out. Every proposal we send now shows you both numbers.
The era of pretending AI isn't in the room is over. Let's just be honest about it.
See more on YouTube - https://www.youtube.com/watch?v=2TnzFe6yqpc
