đŚ Continuous Problem Definition
From Symptoms to Signals: observability, evaluation, rollouts, forecasting.
One of my favourite leaders to to learn from is Dave McJannet. He is one of those unicorn leaders, who has vastly diverse experience across complementary areas, like economics, product, and marketing1. Today I want to talk about how tackling Problems2 evolves in a world where the company systems are very dynamic. (yes, thanks AI). And who better to serve as inspiration than Dave!
In a talk once he summarised companies as entities that comprise of 4 big systems: the systems of GTM, Product, Finance and HR. In large orgs each of these systems will have a seasoned leader. Each owns a broad scope: GTM = Marketing, Sales, CS. Product = Product, Design, Engineering. You get the gist.
In large organisations, like the one Dave led, each system has a seasoned owner: CRO, CPO, CHRO, CFO. And the COO triangulates across them. Every C-level has two jobs: spot when there might be a Problem in their system, then solve it or escalate to the COO and broader leadership team. With four strong points of view on the table, the COOâs task is to integrate, judge and decide.
Startups are flatter, but also messier. You donât have four owners; you have ten: Marketing, Sales, CS, Product, Eng, HR, BizOps, Finance, Legal + the founders! They also need to spot and resolve problems. But⌠Ask a Marketing VP and youâll hear symptoms: âMQLs are down; CAC is up.â The temptation is to toss the hot potato downstream: âMQLs arenât down, therefore Sales isnât working them.â Thatâs how finger-pointing starts.
Hereâs the trap: most of what start up leaders see are symptoms, not capital-P Problems. Metrics donât live in isolation. A GTM symptom might be rooted in pricing, activation, or data debt. What looks like Marketingâs Problem is often a cross-system cause hiding in plain sight.
This is where Strategy & BizOps (and strategic COOs) earn their keep. They look at a symptom and know it is just one Signal in all the noise. They work on a portfolio of Signals across silos, map them to shared causes, and drive the initiatives that fix the root Problem, rather than applying band-aids to each metric.
In this post, Iâll show how BizOps manages that portfolio of Signals and leads cross-functional fixes that actually move the revenue machine. And how thanks to new expectations, and demands, call for Continuous Problem Definition and how to get there. Supporters of the SBO Hub also get a full list of things to tweak across the Big Four systems to get to Continuous Problem Discovery, at the end of this article.
So letâs dig inâŚ
Defining Problems, not Symptoms
Scenario 1: ICP drift
For the sake of clarity I will group the following symptoms into the 4 big systems, but if you have a separate leader for each area of these, you can imagine the breakdown. So, each leader will be looking at their own dashboards and spreadsheets. And here is what they might see:
GTM: CAC up, win-rate vs âno decisionâ is weak, need to discount heavily to close.
Finance: Payback period stretching; NRR on new cohorts is lower; sales cycle is up and higher than granular industry benchmarks.
HR: SDR/AE burnout, and attrition; hiring spec keeps changing.
Product & Eng: Activation low for new cohorts; roadmap thrash from fragmented asks.
When I took on a situation like this, what mattered to me was that we got to the root cause of the problem. It wouldnât be enough for the GTM org to slash down spending to reduce CAC; or go into a cheaper channel. This solves CAC. But it would open a ton of issues: NRR could plummet, because the people coming through the new channel are not suitable. The Sales cycle could go drastically up for the same reason. In stead of playing whack âem all, all of the above Signals pointed to one culprit: weâre attracting the wrong customers. Thatâs it. The solution isnât more hiring, or cleaning up the roadmap. It is defining the ICP and sticking to it.3
Level of difficulty: low.
Scenario 2: Pricing misaligned with value metric
GTM: Interest is unchanged, but win-rates going drastically down, low expansion, most new deals are custom deals.
Finance: margin leakage via overage waivers or refunds.
HR: Comp plan disputes; AEâs frustration and attrition.
Product & Eng: Metering is non existent; no way to invoice correctly.
This was the situation after a botched pricing change. In stead of taking 4-6 months to do this properly4, the company had done this in a week. âAs long as it is on the marketing site, we are ready to go.â No. They werenât. Using creative accounting to âmassage the sheetsâ can help only thus far. (this is why Pricing belongs to the COO, not marketing). With this one, if you are gentle you can put the root cause of the problem to be âpricing didnât match the value metricâ but you can also point to big management and ego problems. Those are very difficult to fix, and I am going to leave them out of scope for this post, and in scope for the next BizOps meet up đ.
Level of difficulty: high.
Scenario 3: Data/measurement debt
GTM: Attribution fights; MQL/SQL definitions drift.
Finance: Forecast vs actual whiplash; CAC/payback re-stated.
HR: RevOps burnout.
Product & Eng: Event/schema drift; analytics gaps block decisions; analytics requests pile on.
For anyone experienced, this is a tale you have seen 100 times over. It is the lack of a single source of truth. This is first order of business for any BizOps leader or COO. You simply canât have that. And again, the solution to the above Signals wouldnât be for Finance to simply âget on top of reportingâ, or to hire more RevOps folk. It is to solve for all these symptoms with one initiative. It is not an easy one, and takes time. But it is doable.
Level of difficulty: medium.
To stop patching symptoms, first know they are just Signals. Always trace the shared root and fix it once. Ask âWhyâ many times, and form a holistic understanding. Integrate the signals, name the real Problem, and sponsor the one initiative that collapses the noise.
Continuous Problem Definition across company Systems
Insight used to take months, now surfaces in days. MBRs and QBRs are too slow. With todayâs data appetite and tooling, teams expect near-instant feedback from the big four systems. BizOpsâ job is to make those signals continuous and actionable. Letâs look at how to do this reliably.
Hereâs the steps to take:
1. Observability Infrastructure: start with the data pipelines and dashboards. They are critical for every internal, or external âproductâ.
e.g. in Product you have to monitor the Golden signals + Aha moment, on a per cohort basis.
Tie performance to customer and revenue impact.
New tools allow now for BizOps teams to set up a lot of this themselves. You donât need to wait for an Engineer. And no, you donât have to call yourself a GTM Engineer to have the remit to do so. Just bloody do it.
2. Continuous Evaluation: each initiative should undergo its own unit testing, not periodic reviews.
e.g. for GTM treat lead scoring and routing like models with weekly evals.
Tie to ACV and sales cycle, and monitor weekly fluctuations.
There is no other way to put this: if BizOps is not in the weeds of daily operations, you will have a problem. BizOps needs to spot metrics going in the red before the Sales leaders comes knocking on your door. Take it a step further and you should be offering options for hypothesis first.
3. Gradual Rollouts: plan for a staggered roll out, using feature flags (or versions), tiered rollouts, and instant rollback capabilities.
e.g. for Finance you can implement phased spend gates for headcount, infra, and programs.
e.g. for Pricing launch new packages, discounts, or playbooks behind segment flags (ICP tier, region, plan). Shadow-test qualification rules before enforcing.
Every initiative that I have rolled out in the last 2 years is gradual, gathers continuous feedback, and is easily reversible. BizOps doesnât have time to over plan and do big Enablement sessions and all that jazz. Do it gradually, write the Enablement step by step. You will have it at the end and you can do the âofficial kick offâ then. By that point it is 20min, not an hour and everyone goes back to their business feeling productive.
4. Adopt Probabilistic Modelling: plan for multiple potential versions of the future
e.g. for GTM and Product, model scenarios of release, adoption, and expansion across new ICPs through specific product features, campaigns or geos.
Tie each release to a hypothesis, success criteria, rollback plan, and a comms brief (internal + customer-facing).
Bonus points: monitor your forecasting accuracy!
Only once you did the previous steps can you go into probabilistic modelling. If you got observability, and are on top of monitoring and analysing, you will start plotting scenarios in a model. I used the help of a data scientist to get very sophisticated. You donât have to. But you can paint various scenarios, based on probabilities about what would happen and when. Itâs delightful. Over time you can tweak your forecasts to better them for accuracy.
For supporters of the SBO Hub, here is a detailed list of things you can implement across the Big Four systems that will help you get closer to continuous problem definition. It is not exhaustive but it should give you a good thought starter.
Final Word
If your dashboard canât tell you what changed, for whom, and with what impact in the last 7 days, youâre not executing, youâre guessing. Itâs BizOpsâ job is to turn guessing into gradient descent: smaller steps, faster learning, fewer faceplants. Do this and âProblemsâ stop being fire drills and start being unit tests you pass.
Less heroics. More clarity.
Less big updates. More improvement.
And ultimately, more accurate forecasting!
We do have a similar background (I started in International Relations, then Marketing, then BizOps) so I might be a little biased in my appreciation but then hey, according to observers, under Dave McJannetâs leadership (2016â2025), HashiCorp experienced a decade of aggressive growth, financial stabilisation, and eventual acquisition by IBM for $6.4âŻbillion in 2025, marking one of the most notable SaaS exits of the decade. He must have done something right.
We have spoken about this before, but very briefly, and through the lenses of a lived experience. So today the focus is a more rigorous one.
I should probably coin âFocus on the ICP, stupid. = FISâ
Pricing Strategy and Operationalisation does include the plan but also initial testing and customer feedback before big changes are rolled out to the world.