Software Startup Heuristics
What are the common models for company creation in the enterprise?
After surveying the wide landscape of business software from seed to public equities over the past few years, I’d like to lay out common heuristics or patterns I’ve seen entrepreneurs use to start their businesses. It’s important to note that these are really just jumping off points. The best companies usually take a first principles or unique approach to their market. Also, if you haven’t read it yet, it is worth going through Avichal’s post from a couple years back. He has a broader lens, and I focus my attention to business-facing software.
Follow the money (i.e. build for Sales): Sales departments are budget behemoths inside large organizations given their direct ties to revenue generation. Salesforce is the >$200B incumbent in the space, and I’ll leave the attempts to disrupt it aside for now. The size of the opportunity for other functional tools inside the sales department have blown me away, with the likes of Outreach and Gong scaling rapidly. Newer entrants like Cresta and Forethought in the space are augmenting the workflows of customer-facing teams with AI and ML.
Follow the money part deux: Go after sales and marketing teams (and potentially even replace CRMs or build them from scratch), but do it in a rapidly expanding SMB market segment. Attentive and Klaviyo have succeeded with this approach in e-commerce, and Aurora Solar has done so for solar installers across the country.
Rearchitect a well understood category or budget: Enterprises tend to have common business and IT applications they pay for over many cycles – these include CRM (Salesforce), ITSM (ServiceNow), HCM (Workday) on the business side and tooling in security, observability, networking, etc. on the IT end. With subsequent technical breakthroughs or cloud platform shifts, founders can build new, but familiar, solutions for enterprises. Recent examples include Panther in SIEM and Chronosphere in observability (the child of APM).
Build a complement to an emergent platform: I once spoke to a data buyer at a fast-growing, web-scale company who told me that once they had adopted Snowflake, any new vendor who pitched an integration with Snowflake had a 10x easier time making the sale. Certain pieces of software form such a center of gravity around them that they enrich whole new ecosystems. Companies like Fivetran, dbt Labs, and Hightouch have all benefited from the continuous growth of Snowflake.
Fill a spreadsheet-sized hole: Organizations tend to be litter boxes of spreadsheets and email. Once in a while, common workflows emerge in these operational processes. There have been recent examples of software companies that add a nice UI, version control, and audit logs on top of commonly-used spreadsheets; but can accrue a ton of value. Instances include Carta and Pave.
Make something collaborative or interactive: This trend was all the rage in 2019 and 2020, and part of me feels like it’s a bit played out. Turn a piece of software that used to be a single-player utility into a multiplayer experience given the enhancements in the browser. The canonical examples are Figma and Miro. A recent one that looks cool (“Zendesk reimagined”) is Cohere.
Reduce the friction in creating assets: Some of the best software companies turn scarce resources or assets into abundant ones. By enabling creation, these tools themselves capture value. Airtable makes any enterprise user into an application builder. Retool directs the attention of developers away from wrangling the mechanics of CRUD apps.
Provide structure and an opinion to a software development workflow: I once tweeted that that “bash scripts and config files are to dev tools what spreadsheets and emails are to SaaS (see above).” There are tons of unstructured problem areas in the development workflow (especially in building applications and infrastructure at scale), and tools with a strong opinion on process and workflow often end up succeeding. Companies that come to mind for me include Temporal, PlanetScale, Materialize, and Buf.
This list wasn’t meant to be exhaustive or comprehensive, but hopefully it provides folks with some inspiration. Vertical software and all the different wedges there could provide material for a follow up post.