Ten Standards That Keep AI Consulting Clients
I keep a short list for how client work runs. None of it is about models or tooling. It is the professional standard I hold while the market keeps lowering the bar.
1. Over-communicate until it feels excessive
Same-day recaps after every call. Progress updates as often as the work allows. A quick "on it, update tomorrow" within a few hours even when I do not have the answer yet.
Clients get nervous in silence even when the work is fine. They do not know what was done unless you show them or say it. I never assume they magically know four hours went in yesterday.
2. Show work every single day
Daily demo items in a shared Slack channel or email. When a client sees something new every morning, you become part of their team. A week without hearing from you creates uncertainty, and uncertainty leads to churn.
3. Bad news travels from me first
If something breaks or a deadline slips, they hear it from me before they discover it themselves. It is uncomfortable. Clients remember who told them the truth early.
4. Under-promise every number, then over-deliver on it
Project £8k a month in savings and deliver £12k, and you have earned a client for life, a referral, and a case study. (That shape is the deal I aim for, not a result I am claiming yet.) Client sentiment carries more than anything else on this list.
5. Talk about money the moment it comes up
I keep a change-order process and I use it. Scope changes, extra requests, anything that adds cost gets discussed the same day, calmly and in writing. One surprise line item on an invoice can undo six months of good work.
6. Speak their language
Most clients do not care about webhooks, vector databases, or which model I used. They care that intake used to take 45 minutes and now takes 4. I translate everything into hours and pounds.
7. Document everything and keep it organised
Loom walkthroughs, admin guides, troubleshooting docs. That is what gets a build used, and adoption is everything. Documentation is baked into the price. Plenty of agencies burn out because the client stops using what was built.
8. Spend 30 minutes a day immersed in AI
Read the releases, listen to a podcast on the drive, test the new model the week it drops. This space moves weekly and clients pay for currency. Fall behind for a month and the advice is already stale.
9. Run my own business on what I sell
This practice should be its own best case study. If my onboarding, reporting, and follow-up are still manual, I am selling something I do not trust enough to use myself.
10. Book the next step before the current one ends
Every call ends with the next call on the calendar. Every project ends with a conversation about what comes next. Momentum dies, and it rarely comes back.
The technical stack will keep changing. These ten do not.
Want systems like this in your business? See consulting services →