Account-based scoring for non-traditional accounts
Account-based scoring for non-traditional accounts
Creating scores for companies that are not very online is a challenge. Here's how you can build true/false factors to rank them. Read: "Everything Starts Out Looking Like a Toy" #193
Hi, I’m Greg 👋! I write weekly product essays, including system “handshakes”, the expectations for workflow, and the jobs to be done for data. What is Data Operations? was the first post in the series.
This week’s toy: breaking down the samples from a hip-hop classic to explore the world of analog sampling. If AI music is going to become a thing, it needs to figure out how to document the lineage of sampled music (and pay the creators). Edition 193 of this newsletter is here - it’s April 8, 2024.
Brought to you by Apollo.io, an all-in-one solution providing RevOps teams with access to data, enrichment, outreach, call intelligence, scoring, calendaring, and email automation. They integrate seamlessly with your CRM and existing workflows. The quality of Apollo’s data is unmatched, they are ranked #1 for contact and company data accuracy on G2 with over 6000 reviews and a 4.8 star rating. Learn more …
If you have a comment or are interested in sponsoring, hit reply.
The Big Idea
A short long-form essay about data things
⚙️ Creating reliable account-based scoring
Receiving a new lead as a seller is great. It’s even better when that lead is enriched with valuable information based on the contact. Yet many of the leads we receive don’t have a social presence or sufficient coverage from people enrichment tools to show up with meaningful data that lets you make a decision.
The first place most people fall back to is account-based scoring. In a tool like Apollo, it looks like building a score based on observable technographic and firmographic aspects of the company. Technographic points include the information you can see on their website (tools they use) and firmographic values are sourced from a variety of places to identify the employee count, revenue, location, industry, and other relevant points for an account.
But how do you know which data points are relevant for finding the prospect in your ideal customer profile, particularly if it is outside the typical technology company? Sources like Apollo can help if you know what you’re looking for in the ideal customer. But what if your version of the “ideal customer” is not easily found in traditional data sources?
Matching a “Good Customer” and a “Bad Prospect”
For account-based tools to work in a novel environment, you need to identify a combination of signals that reliably match the kind of accounts you want to attract (and the ones you want to avoid).
Here’s an example: selling to service-based businesses. They are likely to have an online presence, but some other factors might be a bit harder to discern from a careful glance.
Let’s take a case of searching for plumbers in Seattle, WA and trying to find highly ranked smaller shops that are not franchises.
You might start by searching for “highly ranked plumbers near 98101” in a search engine, you’ll find both sponsored rankings (indicating who is spending advertising money) and organic rankings (who is ranking based on content and visits)
At the top of the list, visit five websites and pick one that you like (pleasing, modern design with small details like an updated footer, social links, and easy-to-find contact info)
For a regulated industry, you want to see a valid contractor license listed on the web site (an additional data point you could look up in a state database if available)
For customer contact, you might list what methods they have for taking appointments (chat, email form, phone), and test how long they take to respond to a simple question.
For additional research, look up that business in the state’s contractor license and validate that they are licensed, bonded, and insured and that their qualifications have not expired.
Each of these items can be detected manually - and there might also be a way to validate them automatically if you have access to the right data set.
Mapping signals to accounts
Great! Now we have some positive and negative signals to apply to new prospective accounts, and we need to apply some bounds to our newly found factors.
For our search, here are factors to differentiate between accounts:
Search performance: is it good enough that the business appears in the first two pages of search results for a very specific query, like “highly ranked plumbers 98101”, or does the query need to be wider, like “best plumbers in Seattle area”? This is a good example of having a “better” or “worse” score for “shows up in search results” if you’re thinking about this from the perspective of the buyer.
Industry qualifications: if it’s a regulated industry, can you validate their qualifications? What services do they offer, and are those similar or different to our key customers? Is there other social proof that would indicate industry success, such as membership organizations or badges from a review site such as Angi or G2?
Design sense: this one’s a little bit more subjective. When you open the website, is it easy to read? Does it load quickly? Is the call to action obvious? Once you look at several examples, it’s easier to say “well-designed” or “not well-designed.”
Customer focus: now put yourself in the role of the customer. If you were trying to purchase from this business, could you get that done without searching multiple pages? Is it easy to contact the business via chat?
For each of these factors, you could score them from 1-5 or as a true/false switch, then rank the factors to determine a score that clears your scoring threshold.
Here’s a matrix you can use to build the criteria for your manual scoring before you try to replicate the results:
What’s the takeaway? Account scoring is highly objective by industry and by ideal customer profile. Building true/false factors or items that could (eventually) be ranked by a computer makes it possible to build a scoring model even in industries that don’t have good coverage in traditional firmographic and technographic fields.
Links for Reading and Sharing
These are links that caught my 👀
1/ The design behind useful tables - Tabular data presented well is one of the most effective data visualization techniques I know. That’s why I was excited to find this short article on the design philosophy for excellent data tables. Whether you use the Python package this article describes or just soak in the information architecture gems, there’s some great stuff here.
2/ Using AI makes the average user faster - A study done at Intuit tracked the performance of analysts writing queries and found that using AI sped up the efforts for most users. Note that this result was enablement completed with an in-house LLM and not an off-the-shelf AI tool, but the insight is clear. Getting from 0 to 1, particularly with an unfamiliar data set, is easier with an AI tool that answers “dumb questions.” Remaining to be seen: how well the LLM extends knowledge for the median user in addition to accelerating the path to existing knowledge.
3/ What makes housing expensive - The average price of housing units in the US has skyrocketed over the last 4 years. What are the components of this price increase? The biggest contributors are the expectations for the cost of land and the zoning and other regulations that determine what is built on that land. In other words, enable infill/ADU and multi-family housing in desirable areas to lower the cost of housing.
What to do next
Hit reply if you’ve got links to share, data stories, or want to say hello.
The next big thing always starts out being dismissed as a “toy.” - Chris Dixon