Measuring CS: Affordability
This is part 2 of my article on how to measure customer support.
To recap, I had mentioned that there are two top level metrics – one each in the area of effectiveness and affordability. Part 1 covered effectiveness.
This article will talk about affordability. Let's begin with the definition:
Affordability: Total support cost as a percentage of the total P&L costs
Why am I recommending this ratio? Why not just the total support cost? Because support cost is essentially a variable cost. (Not directly variable in the sense that, say, payment gateway fees are, but variable nevertheless because they move mostly linearly with business volumes) And variable costs are best understood in relation to revenue. But why I am recommending dividing by costs rather than revenue? Because it is easier to get a sense of affordability when you look at share of costs.
Cost of Support
The first step to get to this ratio is to have a good handle on the numerator- the support costs themselves. Unfortunately, often that’s not easy to get. Because support costs are spread across buckets that finance teams usually look at. Granular data that’s needed to separate out support costs from those buckets is often not available with finance teams.
Let’s talk about the core buckets:
I’m excluding any money spent on tech and product teams from this. Only because it is a bit difficult to separate out opex and capex in tech spends. But if you’re able to do that, then definitely add the opex from there as well for a full picture!
Once you have the affordability metric, based on your P&L structure, you can think about what’s a reasonable number for you. And use that to decide how to prioritise efficiency v/s effectiveness initiatives.
(Note that when you look at support costs, there is this element that’s not reflected in the P&L – customer LTV (Life Time Value). Bad support will lead to lower LTV and higher cost of acquisition. It’s not easy to estimate that impact, but keep in mind that it’s there.)
Another way to look at the cost is at a unit level. If your org does unit costing, this one is easier to get a mental model of, v/s the P&L structure. The unit varies depending on the business- for travel, it will be each room-night booked, or flight-passenger. For ecommerce, it will be every product shipped.
Allright. So now we have a handle of the Level 1 cost metric, and the one the board/CEO should be looking at. But, just as with the experience metric, this metric doesn’t tell you about the levers. For that, we need to go to the next level.
Cost Drivers: Level 1
Here's the formula that breaks down total support cost:
Total Support Cost = No. of transactions?*? Contact Rate * ?(Cost/Contact)
Let’s discuss all three.
If business grows, everything else being equal, cost of support will go up. Folks need to understand that - it's a variable cost!
(More broadly – divide actual support contacts by the number of occasions the customers might have to reach out to support.)
This is also the metric to measure the very first stage in the customer support journey – Need for Support.
The contact ratio is the primary metric that tells you the overall health of your upstream products. A low contact ratio means the products are healthy.
( As with most metrics, this comes with a qualifier – contact ratio is dependent on how easy or difficult it is for customers to reach support. So, as that changes, the CR metric can also change. Keep that in mind. You can get a sense of how bad access is, through your NPS program- we covered that in Part 1.)
With this formula, it’s pretty clear that the support team has zero control over two of the three levers, and only partial control over the third one. Despite that, the irony is that the support team is tasked with reducing the cost of support. No wonder they end up using the levers they control- reducing the cost of agents, reducing overall capacity, floor management to keep AHT low etc.
All of these levers end up worsening customer experience. And end up costing the business more! They just push customers away- take these measures to the logical extreme, and you might as well shut down customer service completely!
Cost Drivers: Cost/Contact
Cost/Contact has a lot going on within itself. Let’s break that down further, one more level.
Cost/Contact = SoSS * cost/self-serve contact + (1-SoSS) * cost/agent contact
Where SoSS = Share of Self-Serve in your total support contacts
领英推荐
Given that self-serve cost is close to zero in comparison to agent driven cost, it makes sense to dive deep into self-serve, define relevant metrics, and find targeted ways of improving its share. Will take that up in a separate article.
For now, let’s dive two more levels for agent channels.
Cost/Contact = (Cost/FTE) / (Contacts/FTE)
Contacts/FTE = (Total Hours/FTE) * Occupancy/AHT
Where FTE= Full Time Equivalent (Basically a support agent)
And Occupancy = Percentage of total login time agents are helping a customer
Now that we have a good handle on the cost metrics, let’s talk about the major levers to reduce costs.
Cost reductions: Principles and Ownership
One way to look at support costs is to think of them as cost of bad quality. Support exists because there is some gap in the core product/service. If there was no gap, customers wouldn’t need support.
So the primary ownership of support costs should rest with business/product teams. They should take the cost/contact as input, and use that to prioritise product efforts. The business leader owns the P&L – they should decide where product bandwidth is best utilized.
Within the overall support cost, the product team should own the contact rate. There should be a two-way contract between the contact center and the business team. The product/business team projects a contact rate and ensure that’s achieved. And the contact center promises a cost/contact and delivers on it.
Contact Rate is one of the two primary levers for cost reduction. It also serves a dual purpose – reducing contact rate means fixing upstream issues, which will improve overall experience and therefore lead to higher retention as well.
The other major lever is share of self-serve. That should also be owned by the product team, as most of the levers to improve self-serve lie with them.
That means that in aggregate, the contact rate into the contact center is the responsibility of the product/business teams.
The contact center team owns the cost/agent, repeat contacts, and the experience metrics (CSAT, escalation rate, backlog etc.).
Generally, the contact center team owns the AHT as well. But I believe AHT should be owned by product teams. That’s because AHT is largely determined by the complexity of resolution process and the tools we provide to the agent. And those are driven by the product team.
It’s very important to define the ownership of the cost levers across teams very clearly. One of the reasons organisations struggle to do a good job on support improvements is that the accountability typically lies with the contact center team, who have limited power or influence on product teams.
I’ll talk in some more detail on how to move the major levers in a subsequent article.
But for now let me talk about one lever that I feel gets abused too much.
What not to do: Cost/FTE, or hours/FTE
At least for India businesses, CS cost reduction efforts end up focusing on cost/FTE, as that’s the easiest lever to pull, and pretty much the only one in the hands of the contact center. (This is probably not as true for UK/US businesses with contact centers based in India/Philipines/Eastern Europe, as they are already getting benefit of significantly lower labour costs compared to revenue).
All you need is to renegotiate commercials with your contact center partner. Or move to a cheaper partner/location. Both of which will end up impacting quality beyond a point.
The irony is that reducing the cost/FTE doesn’t really move the needle much! In a typical CS set-up, you’ll have automation channels, in-house contact center, in-house social teams, and external contact centers. In set-ups I’ve seen, agent cost for outsourced centers ends up being around 40% of total CS costs. Which means that even a 20% reduction is cost/FTE ends up reducing overall CS costs only about 8%. And 20% is super hard to do unless you’re cutting corners!
If you're partnering with a tier 1 vendor, the commercial benefit of moving to a tier 2 town is often only about 10%. If you move to a tier 2/3 vendor AND to a tier 2/3 town, you might be able to save 20%, but it will result in other issues – scaling, quality etc.
Add to it the fact that it takes at least 3-4 months for any new center to stabilize, you’re looking at a 6 month long project to move centers, all for a ~8% savings in overall CS costs! (And because of the resulting quality issues, your repeat calls might become higher, which might negate the theoretical savings!)
I’m not saying there is no scope for saving money by reducing agent cost – but take a considered call keeping in mind the risks and the potential cost savings.
Another trend I’ve seen in the last few years in the Indian (domestic) business is increasing the total hours/FTE. Earlier, we used to have 176 or 184 hours per month – based on a 5 or 5.5 days week. Now folks contract at 196/208 hours – that’s 6 days a week, with no room for time-off. This leads to agent burnout, again impacting quality significantly.
Along with this, occupancy is tightly controlled as well- agents are helping customers 80-85% of the time they are in office! (The standard for international customers is 60-65%)
With high hours and high occupancy, you have no flexibility to generate additional capacity when there’s a supply or demand shock – you have no scope for overtime!
As of 2023, for India businesses, I think if you’re in the 32-36k monthly per FTE range, 184 hours per FTE per month, and 75% occupancy, you’re broadly in a good range. Downside risks of tightening any of these further are probably higher than the upside, and your effort might be better spent elsewhere.
In the next couple of articles on CS measurements, I’ll go deeper into the self-serve and contact reduction levers.