AI and veterinary complaints. Hear me out.

I’ve been a vet for more than 20 years, and I don't think I've ever met anyone in practice who enjoys receiving a complaint. They're horrible to deal with and we never lose that feeling of dread when one comes in.

Someone in the team may be really upset by it, the client is unhappy, and you've now got to somehow work out what's gone on while still getting through everything else in a busy practice day.

How we deal with complaints has changed enormously during my career. I still remember an older vet dealing with an unhappy client in a way that basically amounted to telling them where to go, in rather more ‘direct’ language: “Get out of my practice!”

I don’t think that was unusual at the time. We've thankfully come a very long way from that, although I'm not convinced that, in the interim, we ever really cracked the complaints piece.

For quite a few years, my own approach was essentially to make complaints “go away”. I'd pick up the phone, listen, work out what I thought would make things better, try to be fair to the client while protecting my team, and hopefully get everyone to a point where we could move on.

Quite often that worked. What it didn't necessarily do was make the practice any better.

The CMA changes give us a reason to think about that.

Practices will need a written complaints process, along with a complaints log which is reviewed periodically, meaning that we learn from individual complaints and identify wider problems. The CMA's final decision goes further in saying practices should consider improvements to their services or processes and make changes where it is reasonable to do so.

I have to say, I think that is a good thing. Uncomfortable though they are, some complaints tell us things we need to know.

Start by bringing the whole story together

A complaint rarely comes in one neat package.

Previously, I may have picked up a complaint letter, had a quick glance at the clinical notes and picked up the phone. A quick conversation will still resolve many complaints. There will, however, be situations where a complaint needs more thorough investigation.

There's a huge amount of context that might be useful to have: a Google review, the clinical record, emails, telephone conversations, estimates, consent discussions, messages between colleagues, practice policies and SOPs. Then there are the accounts of the people involved, which may differ considerably.

What did the client think was going to happen? What did the vet think had been explained? What did reception understand? Was there a handover? Did our own process say something should have happened that didn't? Did the team follow a perfectly reasonable process, and the outcome was still not what anyone wanted?

It's a huge amount to pull together, particularly when you're busy, and I think there is a temptation to get enough of the story to deal with the immediate problem rather than really understanding it.

That was certainly true for me at times.

Try not to start with the person who got it wrong

Looking back, some of the conclusions I reached when dealing with complaints myself were far too simple. I can remember thinking things like, “This person just needs to be kinder to clients”, or “We need to make sure they don't say that again.”

To my shame, that was sometimes about as far as the learning went.

Perhaps that person needed more training. Perhaps we'd put them in a difficult situation without giving them the tools to deal with it. Perhaps our process was poor. Perhaps the SOP was fine, but it lived in a dusty folder nobody had opened for three years. Perhaps I simply hadn't spent enough time working out why something had happened in the first place.

With that in mind, I think there's a useful change of mindset here. When something has gone wrong, it's very easy to look for the person responsible. Starting instead with the assumption that there may be something in the way the practice works that contributed to it gives you somewhere more useful to look.

There will obviously be occasions where an individual has made a mistake or behaved badly. There will also be complaints where, having looked at everything, you conclude that the team did nothing wrong. That's important too. Sometimes shit just happens, right?!

The point is to do enough work to know which of those situations you're dealing with.

Give complaints a process

Towards the end of my time running our practices, we became much more structured about complaints. We recorded them and started looking at what we were getting across our different sites. Most importantly, a wonderful vet took on the role of complaints champion. (Thank you, Roseann!)

She did additional VDS training and became extremely good at it. I had enormous admiration for her because dealing with complaints week after week is not an easy job. We checked in regularly and, over time, she became a real expert at it.

Having someone with expertise was a huge improvement on me picking up the phone and trying to sort everything out. I now think the process around that person is just as important.

A decent complaints process should help you gather the information, understand the different parts of the complaint, involve the people who need to be involved, keep track of what's happening and reach a conclusion which you can explain. It should also leave you with something more useful than “complaint closed”.

The question at the end should be:

“What, if anything, should we learn from this?”

That might mean changing an SOP, providing some training, improving a handover, changing the way something is explained to clients or simply reminding people of a process which already exists.

Sometimes the answer will be that nothing needs to change. Having gone through the process and reached that conclusion is very different from assuming it at the beginning.

Don't make complaints somebody's lonely job

There is another reason I think a process helps. Complaints can be miserable for the people involved in them.

The person named in the complaint may be distressed by it, while the person investigating it can end up carrying a huge amount themselves. Becoming the person in a practice who “deals with the complaints” could indeed be a very lonely job.

There needs to be some distance between the investigation and the people directly involved, which can be extremely difficult in a practice team, although as I have already said, that doesn't mean the investigator should be left on their own. It should be easy to ask somebody what happened, bring another person into the discussion, see where the complaint has got to and check in with the people affected by it.

A good complaints process should look after the team dealing with the complaint as well as the client making it. From my experience with Roseann, I'd also make sure that whoever takes responsibility for complaints has some training, support and someone checking in with them. Giving somebody the title of Complaints Champion and then sending every difficult client their way isn't much of a support system.

Use AI as another perspective, not as the complaints department

I've spoken to several vets who feel quite strongly that AI shouldn't be involved in complaint handling, and I understand the instinct. If “using AI” means dropping a client's complaint into ChatGPT and asking it to write a nice response, I'm not on board with that at all.

The bit I find much more interesting is whether AI can help us with the difficult work that comes before the response.

As said, a complaint may involve many pieces of context from many different places. Working through all of it, comparing different accounts, checking what the practice's own policies say and spotting something that has been missed takes time. There are ways in which AI could make that job considerably easier, provided it is being used within an appropriate, secure system and the people involved remain responsible for the decisions. AI can be very good at taking a large amount of information, filtering and sorting it, and presenting it back in a more useful form.

I quite like the idea of AI being almost a third person in the room. It hasn't been embarrassed by what happened, it doesn't work with the person being complained about, it isn't worried about upsetting the client and it hasn't already decided whose version it believes.

I wouldn't want it deciding:

“Who was right?”

I'd be much more interested in it helping us ask:

“What have we missed?”

What supports each part of the complaint? Are there gaps in the information? Do the accounts contradict each other? What did our own policy say should have happened? Have we seen something similar before? Is there something here that we could change?

The judgement remains with the people responsible for the complaint. The useful bit is having some help to look across the information before making that judgement.

Put another way, the most interesting role for AI in complaints isn't writing the response at all. It could be helping us understand how we got there and how we avoid ending up in the same place again.

Close the complaint, then do one more thing

I don't want to trot out the cliché that “a complaint is a gift”. It certainly doesn't feel like one when it lands in your inbox. I'd rather have the card and chocolates!

Looking back at the practices I ran, though, some complaints genuinely told us things we needed to know. We weren't always good enough at turning what we'd learned from one unhappy client into something that made life better for the next client, or for the team looking after them.

There are complaints I dealt with where I was pleased with the outcome. The client was happier, I'd protected my team and we'd put the problem to bed. I'm less certain that I could always tell you what we changed afterwards. That's the bit I think we need to get better at.

Once the complaint is dealt with, I'd ask:

“What are we going to do differently because of it?”

There won't always be an answer. Sometimes you've looked into it, the team did the right thing and you'd do the same again.

Other times there will be something. A conversation that might have been worded better, some training we need, an SOP nobody knew existed, a handover that didn't happen, or perhaps something that starts to look like a bit of a pattern when you look across several complaints rather than one at a time.

I still don't think I'll ever be pleased to see a complaint arrive in my inbox, but I do think that, having gone through all the pain of dealing with one, we should at least make sure we've learned everything we can from it. If AI can help us see something we might otherwise have missed along the way, I think that's worth exploring too.